USA: +1(669)289-8683 
Sign in
Type a set of five questions and answers that reflect your understanding of the concepts of facts, win-win transactions, loopholes, and line drawing.
Your test will be graded on your ability to create thought-provoking, practical examples and demonstrate deep engagement with the material. Be sure to follow these guidelines carefully.
What to Do:
- Create Five Original Questions (20 Points Each):
Write one question for each of the following concepts:
Facts: Focus on how legal decision-makers determine “what happened” in a case.
Win-Win Transactions: Explore situations where all parties seemingly benefit and the legal or ethical boundaries of such agreements.
Loopholes: Investigate gaps in laws or rules and how they can be exploited or closed.
Line Drawing: Examine how courts, lawmakers, or individuals draw boundaries to resolve complex legal issues.Write a fifth question that ties together two or more of these concepts.
- Draft and Explain the Correct Answer (and Incorrect Answers):
For each question, provide an answer key.
Explain in 3-5 sentences why your answer is correct.
Explain why the other choices are incorrect. - Cite Readings and Class Discussions:
Use examples, principles, and authors we’ve studied in class to support your answers.
Draw from readings like Kenneth Arrow on incentives, Saul Levmore on legal innovation, and Michael Pardo on proof.
Grading Criteria:
- 20 Points per Question (100 Points Total):
Creativity & Practicality (10 points): Does your question reflect real-world complexity? Does it require thoughtful application rather than simple recall?
Clarity (5 points): Is the question clear and well-written?
Correct Answer Explanation (5 points): Does your explanation demonstrate depth and reference course materials?
Important Notes:
- Avoid simplistic or overly broad questions, such as “What is a loophole?” or “Define win-win transactions.” Questions like these will receive little to no credit.
- Be practical: Think of scenarios that a lawyer, judge, or policymaker might actually encounter.
- You’re welcome to reference hypothetical scenarios or current events to inspire your questions.
Example:
- You may wish to create a question about a company’s creative interpretation of tax laws (loopholes) or a scenario where facts are disputed in a courtroom. Be sure to focus on the legal reasoning process and the consequences of these situations.
Michael Pardo’s “Judicial Proof and the Best Explanation”
Michael Pardo’s “Judicial Proof and the Best Explanation” presents a philosophical and
analytical perspective on how courts should approach the process of evaluating evidence and
arriving at judicial decisions. Pardo emphasizes the importance of understanding evidence in
terms of its explanatory value to ensure that the legal system reaches accurate and just
conclusions.
In the first section of his work, Pardo critiques the prevalent probabilistic models of
evidence and legal proof, suggesting that these models can sometimes lead to counterintuitive
and unjust results. He argues that probability alone should not be the sole criterion for
evaluating evidence, as it often ignores the complexity and nuances of real-life situations.
Instead, he proposes a shift towards a more explanatory model, where the focus is on how well
the available evidence explains the events or facts in question.
Pardo then delves into the concept of Inference to the Best Explanation (IBE), proposing
it as a more suitable and effective method for judicial proof. IBE is a form of reasoning that
involves selecting the hypothesis or explanation that best accounts for the available evidence. In
legal contexts, this would mean that judges and juries should base their decisions on which
narrative or account of events is best supported by the evidence at hand, rather than solely on
probabilistic calculations.
Pardo also addresses potential criticisms and misunderstandings of IBE, emphasizing
that it is not a subjective or arbitrary method, but rather a structured and rational approach to
evidence evaluation. He highlights the importance of transparency, coherence, and rigor in
applying IBE, ensuring that the chosen explanation genuinely aligns with the available evidence
and is not the product of biases or preconceived notions.
Finally, Pardo connects the theory of IBE to broader themes in epistemology and
philosophy of science, demonstrating its relevance and applicability beyond the legal realm. He
underscores the potential of IBE to enhance the accuracy and fairness of judicial proof,
contributing to a more reliable and just
legal system.
Multiple Choice Questions:
What does Michael Pardo critique in his work “Judicial Proof and the Best Explanation”? A. The
use of witness testimonies B. The prevalence of probabilistic models of evidence C. The role of
judges in the courtroom D. The application of Inference to the Best Explanation (IBE) Answer: B.
The prevalence of probabilistic models of evidence
What alternative does Pardo propose for evaluating evidence in legal contexts? A. Relying
solely on eyewitness testimonies B. Implementing a probabilistic model C. Using Inference to
the Best Explanation (IBE) D. Ignoring circumstantial evidence Answer: C. Using Inference to
the Best Explanation (IBE)
How does Pardo view the relationship between IBE and judicial proof? A. IBE is irrelevant to
judicial proof B. IBE can contribute to a more accurate and just legal system C. IBE makes the
legal process overly complicated D. IBE should only be used in criminal cases Answer: B. IBE
can contribute to a more accurate and just legal system.
Here are three real-life examples to help illustrate Michael Pardo’s work on “Judicial Proof and
the Best Explanation”:
Example 1: Wrongful Conviction
Situation:
A man is convicted of a crime based on eyewitness testimony and circumstantial evidence. The
probability models suggest he is likely the culprit due to his presence at the scene and a vague
identification by a witness.
Application of Pardo’s Work:
Using Pardo’s explanatory model, the court would re-examine the evidence to see which
explanation best fits the facts. The man’s lawyer presents an alternative theory, showing that the
timeline doesn’t add up and the eyewitness was not wearing their glasses. The best explanation
may shift away from the man’s guilt towards a possible misidentification.
Outcome:
The man’s conviction is re-evaluated, and he is found not guilty, demonstrating how the
explanatory model can prevent wrongful convictions and contribute to a more accurate and just
legal system.
Example 2: Patent Dispute
Situation:
Two companies are in a legal battle over a patent, with both presenting extensive technical
evidence to support their claims.
Application of Pardo’s Work:
Rather than calculating the probabilities of each company’s evidence being true, the court
applies the Inference to the Best Explanation. They analyze which company’s evidence better
explains the technology in question and how it aligns with the patent.
Outcome:
The court rules in favor of the company whose evidence provides the most coherent,
comprehensive, and plausible explanation of the technology, showcasing how Pardo’s model
aids in complex cases where a nuanced understanding of evidence is crucial.
Example 3: Medical Malpractice
Situation:
A patient sues a doctor for medical malpractice, presenting symptoms and medical records as
evidence. The doctor’s defense brings their own medical experts and records.
Application of Pardo’s Work:
Instead of relying on the probability of malpractice based on symptoms alone, the court utilizes
the explanatory approach. They assess which side provides a better explanation of the patient’s
condition, taking into account all medical opinions and records.
Outcome:
The court reaches a verdict based on the most comprehensive and plausible medical
explanation, ensuring a just decision that is grounded in a thorough understanding of the
evidence. This exemplifies Pardo’s call for a more rational and structured approach to judicial
proof.
Saul Levmore’s “Unifying Remedies: Property Rules, Liability Rules, and Startling Rules
Saul Levmore’s “Unifying Remedies: Property Rules, Liability Rules, and Startling Rules
” critically examines the existing frameworks of legal remedies, particularly focusing on the
categorizations introduced by Calabresi and Melamed, who separated legal remedies into
property rules and liability rules. Levmore acknowledges the transformative impact of their work
while highlighting its limitations and the potential for further development in understanding
legal remedies. He emphasizes the necessity of exploring new and innovative remedies that have
emerged since the original framework was proposed, with the aim of developing a more
comprehensive and unified theory of
legal remedies.
Levmore delves into the intricacies of legal remedies, considering how rights, remedies,
and private bargains can be combined to serve a variety of functions while upholding values of
efficiency, fairness, and redistribution. He brings attention to “startling rules,” a category of
remedies that break conventional norms, such as requiring a party in the right to pay when they
have been wronged. These rules, Levmore argues, can play a crucial role in encouraging honest
claims and enhancing the integrity of the legal system.
The work also revisits the Calabresi-Melamed framework, identifying gaps and areas
left unexplored. Levmore highlights contributions from other scholars, such as Krier and
Schwab, who have added depth to the understanding of legal remedies, introducing new
perspectives and possibilities. He emphasizes the importance of considering the unique
qualities of these innovative remedies and their potential contribution to a unified theory of
legal remedies.
Levmore’s analysis does not stop at exploring new remedies; it also revisits familiar
ones, scrutinizing their content and the implications of their application. He seeks to expand the
application of the Calabresi-Melamed framework beyond traditional property and tort law,
advocating for a comprehensive approach that encompasses a broader range of legal disputes
and remedies. Through this, Levmore aims to contribute to the ongoing development of legal
theory, pushing towards a more holistic understanding of legal remedies.
Multiple Choice Questions:
What is Saul Levmore’s main critique of the Calabresi-Melamed framework in “Unifying
Remedies: Property Rules, Liability Rules, and Startling Rules”?
A) It is completely incorrect.
B) It is outdated and no longer relevant.
C) It has limitations and there is potential for further development.
D) It only applies to property law.
What role do “startling rules” play in Saul Levmore’s analysis of legal
remedies?
A) They are irrelevant and should be ignored.
B) They break conventional norms and encourage honest claims.
C) They only apply to liability rules.
D) They are the only focus of Levmore’s work.
According to Saul Levmore, what is necessary for developing a comprehensive theory of legal
remedies?
A) Ignoring all previous work on legal remedies.
B) Focusing solely on property rules.
C) Combining rights, remedies, and private bargains in innovative ways.
D) Eliminating liability rules from consideration.
Answers: C, B, C
Here are three real-life examples that illustrate the concepts discussed in Saul Levmore’s
“Unifying Remedies: Property Rules, Liability Rules, and Startling Rules”:
1. Eminent Domain and Compensation:
Imagine a situation where the government decides to build a new highway, and part of the
planned route goes through a privately-owned piece of land. According to traditional property
rules, the government has the right to seize private property for public use through eminent
domain, but they must provide just compensation to the owner.
Relation to Levmore’s Work: This scenario mirrors Levmore’s discussion on “startling rules,”
where a party in the right (the property owner) ends up receiving payment as a remedy. Here,
the government’s action is justified for the greater public good, but the property owner is
compensated for their loss, showcasing a blend of property rules and liability rules.
2. Environmental Pollution:
Consider a factory that is releasing pollutants into a river, negatively affecting the water quality
and harming the livelihoods of fishermen downstream. The fishermen could seek an injunction
to stop the factory from polluting the river (a property rule), or they could seek damages for the
harm caused (a liability rule).
Relation to Levmore’s Work: Levmore’s work emphasizes the importance of finding innovative
remedies that balance efficiency, fairness, and redistribution. In this case, a more nuanced
solution might involve implementing stricter environmental regulations and requiring the
factory to install pollution control measures, which would address the root cause of the
problem while allowing the factory to continue operating.
3. Intellectual Property and Licensing:
Imagine an inventor who has patented a new technology. Another company sees potential in
this technology and wants to use it, but they cannot do so without infringing on the inventor’s
patent rights. The company could negotiate a licensing agreement with the inventor, paying for
the right to use the technology (a liability rule).
Relation to Levmore’s Work: This example reflects Levmore’s ideas on combining rights,
remedies, and private bargains in innovative ways. Instead of resorting to a lawsuit (a property
rule approach, where the inventor stops the other company from using the technology), the two
parties come to an agreement that benefits both, illustrating a practical application of liability
rules and private bargaining.
These examples highlight how Levmore’s concepts of property rules, liability rules, and
“startling rules” can be applied in various real-life situations, showcasing the necessity for a
comprehensive and flexible approach to legal remedies.
Kenneth Arrow, Social Choice and Individual Value
Kenneth Arrow’s “Social Choice and Individual Values” is a foundational text in the realm
of economics and social choice theory. Through this work, Arrow grappled with a question that
has profound implications for democratic societies: can we convert individual preferences into a
coherent group decision that represents the collective will? The inquiry’s essence touches upon
the challenges of collective decision-making and the intricacies of reconciling individual desires
with group outcomes.
In the book, Arrow presents his famous “impossibility theorem” or “Arrow’s paradox.”
The theorem essentially states that, when there are three or more options to choose from, no fair
voting system can convert ranked preferences of individuals into a community-wide ranking
without running into certain issues. These issues are manifested in terms of fairness criteria,
which Arrow believed any reasonable voting system should satisfy.
The criteria include factors like “unrestricted domain,” which means that voters should be
free to order their choices in any way; “independence of irrelevant alternatives,” implying that if
a choice is removed, it shouldn’t affect the rankings of other choices; and “non-dictatorship,”
suggesting that no single voter should determine the outcome. Other criteria also emphasize that
a voting system should respect the intensity of individual preferences and not just the order.
However, Arrow’s conclusion, derived rigorously through mathematical proofs, was startling.
He demonstrated that no voting method could satisfy all the fairness criteria at once. This result
showcased the inherent challenges and limitations in aggregating individual preferences into a
collective decision. In other words, it’s impossible to create a perfect voting system that
simultaneously respects all the fairness criteria Arrow outlined.
Arrow’s insights into the complexities and potential contradictions of collective decision-
making have profound implications for democratic theory and practice. They highlight the
challenges democratic societies face in translating individual preferences into collective choices.
While the theorem doesn’t offer solutions, it illuminates the intricacies of societal decision-making
processes and underscores the need for continuous examination and refinement of democratic
mechanisms.
Elaborating on Questions of Law vs. Questions of Fact
The distinction between questions of law and questions of fact is one of the most
fundamental in legal reasoning and adjudication. Questions of law involve the application,
interpretation, or validity of a legal principle, statute, or rule. Courts often determine these
questions, as they require expertise in legal interpretation. By contrast, questions of fact revolve
around what actually happened in a given situation and are typically decided by the trier of fact
(a jury or judge acting as a fact-finder).
In practice, legal decision-makers grapple with separating these two domains, as many issues
present both factual and legal dimensions. Courts must determine the applicable law, interpret
any relevant statutes or precedents, and, through a fact-finding process, establish what
happened to apply the law to those facts.
Real-Life Examples and Exercises
1. Contract Dispute Example: A supplier and a buyer enter into a written contract for the
delivery of 100 widgets. The supplier claims they delivered all 100, while the buyer asserts that
only 80 were delivered. The contract states payment will be due only upon full delivery.
o Question of Fact: How many widgets were delivered? This is a question of fact,
as it relates to determining the actions or events that took place.
o Question of Law: If fewer than 100 widgets were delivered, is the buyer legally
obligated to make any payment under the contract terms? This requires
interpreting the legal obligations under the contract.
o Exercise: Students should identify and argue both the factual dispute (evidence
of delivery, testimony, etc.) and the potential legal consequences (contractual
interpretation).
o Model Answer: Factually, the trier of fact would assess delivery evidence, such
as records, receipts, or testimony. Legally, the court would interpret whether
“full delivery” is a condition precedent to payment or whether partial
performance may allow for prorated payment.
2. Criminal Case Example: A defendant is charged with theft, and the key issue is whether the
defendant intended to permanently deprive the victim of their
property.
o Question of Fact: Did the defendant intend to steal the property? Determining
intent involves assessing the defendant’s actions, statements, and any
circumstantial evidence.
o Question of Law: What does “intent to permanently deprive” mean under the
applicable criminal statute? The court defines this legal standard and instructs
the jury accordingly.
o Exercise: Provide students with hypothetical witness statements and evidence,
asking them to determine intent based on available facts.
o Model Answer: Students would evaluate witness credibility, defendant
behavior, and circumstantial evidence to establish intent. Legally, they would
identify the statute’s definition and discuss how intent fits the statutory
requirement.
3. Tort Case Example: A pedestrian is injured after slipping on a wet floor in a grocery store. The
store owner argues that they placed warning signs and mopped the area recently.
o Question of Fact: Was the warning sign visible, and did the floor remain
slippery? This involves assessing the physical facts at the scene and witness
accounts.
o Question of Law: What constitutes “reasonable care” under premises liability
standards? This requires a legal interpretation to apply the standard of care to
the facts.
o Exercise: Ask students to evaluate mock witness testimonies and pictures of the
accident scene to determine factual conditions. They then analyze whether,
assuming their findings, reasonable care was met.
o Model Answer: Factually, students could conclude whether the area was made
reasonably safe based on evidence of warning signs, mopping frequency, etc.
Legally, they would interpret the “reasonable care” standard and decide if the
store owner met their duty.
Hypotheticals and Exercises
1. Hypothetical #1: A homeowner sues their neighbor for trespass, alleging that the
neighbor cut down trees on their land. The neighbor claims the trees were on their
property.
o Exercise: Distinguish questions of law and fact in this scenario, and identify
evidence or legal arguments that could support each party’s claims.
o Model Answer: The question of fact is where the property boundary lies and
whether the trees were on the plaintiff’s land. The question of law involves what
remedies may be available for trespass, assuming a factual finding supports it.
2. Hypothetical #2: A tenant claims a landlord failed to make necessary repairs, leading to
mold that harmed the tenant’s health. The landlord argues the tenant caused the mold
by neglecting proper ventilation.
o Exercise: Break down the factual and legal issues. How would a court determine
the cause of the mold, and what legal duties does a landlord have?
o Model Answer: Factually, the dispute centers on the cause of the mold
(testimony, expert assessments, maintenance records). Legally, the court must
interpret the scope of the landlord’s duty under housing law and the lease
agreement.
Notes on Readings Re: “FACTS”
Frederick Schauer’s work, “Is There a Psychology of Judging?”, as found in The
Psychology of Judicial Decision Making edited by David E. Klein and Gregory Mitchell, and
Stephen A. Weiner’s “The Civil Jury and the Law-Fact Distinction” from California Law Review,
both offer valuable insights into how the distinction between law and facts functions within
judicial decision-making and the impact of cognitive processes and institutional roles.
Schauer’s “Is There a Psychology of Judging?”
Schauer’s writing explores the cognitive processes and biases that shape how judges make
decisions. In the context of distinguishing between law and facts, Schauer emphasizes that
judicial decision-making is not merely a mechanical application of rules to facts but rather
involves complex reasoning shaped by psychological factors. His work suggests that while
judges may strive to adhere to legal standards, their factual assessments, interpretations, and
conclusions are influenced by cognitive limitations, heuristics, and contextual factors.
Key points to consider in this framework include:
• Cognitive Biases: Judges, like all decision-makers, are susceptible to cognitive biases
that can influence how they perceive evidence, assess witness credibility, or resolve
factual disputes. This has implications for how they might blend factual findings with
legal reasoning, even unconsciously.
• Framing and Contextual Influence: Schauer discusses how framing, context, and prior
experiences shape judicial perception and reasoning. This aspect highlights the potential
difficulties in clearly separating fact-finding from legal interpretation, as judges bring
their subjective perspectives to bear.
• Practical Constraints and Heuristics: Decision-making shortcuts, or heuristics, often
guide judges when faced with complex factual scenarios or ambiguous legal standards.
This dynamic can blur the line between what should be a purely factual determination
and what is infused with normative judgments or legal
reasoning.
Schauer’s analysis thus complicates a purely theoretical distinction between questions of law
and fact by showing how psychological realities influence judicial attempts to
compartmentalize these categories.
Weiner’s “The Civil Jury and the Law-Fact Distinction”
Stephen A. Weiner, in his analysis of the civil jury’s role and the law-fact distinction, focuses on
how the division of labor between judge and jury creates challenges in distinguishing between
questions of law and fact. His examination is particularly relevant when discussing civil cases
where juries, as fact-finders, are tasked with making determinations based on their evaluation
of evidence, while judges provide legal guidance and ensure proper application of the law.
Key aspects of Weiner’s work include:
• Institutional Roles and Responsibilities: Weiner emphasizes the jury’s role in fact-
finding and the judge’s role in legal interpretation. However, the boundary between
these roles is often porous. Judges may influence jurors through instructions or rulings
on the admissibility of evidence, while juries may incorporate legal reasoning into their
fact-finding process.
• Complexity of Modern Cases: Weiner argues that the increasing complexity of civil
litigation blurs the law-fact distinction. Complex legal standards often require jurors to
engage with nuanced questions that straddle legal definitions and factual contexts,
complicating their task.
• Potential for Inconsistent Outcomes: The lack of clear lines can lead to inconsistent
verdicts or legal uncertainty, as different juries may interpret the same legal standards
differently based on their factual findings.
Weiner’s perspective highlights that while the law-fact distinction is theoretically essential for
allocating judicial functions, it is less precise in practice due to the interplay between judge and
jury decision-making, the complexity of factual contexts, and the inherent subjectivity involved
in fact evaluation.
Applying These Theories to the Law-Fact Distinction
• Judicial and Juror Reasoning Processes: Together, Schauer and Weiner illustrate that
decision-makers in the legal system operate under a blend of cognitive, psychological,
and institutional influences. This means that even clear-cut distinctions between law and
facts become blurred in practice, as judges’ factual assessments may be swayed by their
understanding of legal implications, and jurors may unconsciously apply legal
reasoning.
• Complex Legal Standards: In scenarios where the law necessitates fact-based
assessments of intent, negligence, or causation, the distinction is even more challenging.
As Schauer notes, cognitive biases and framing can shape judges’ legal interpretations in
these contexts. Similarly, as Weiner discusses, jurors’ assessments may be impacted by
the complexity of applying abstract legal principles to concrete facts.
• Exercises for Teaching: Students might examine hypothetical jury instructions in
complex civil cases, analyzing how judicial framing may inadvertently influence factual
findings. Alternatively, they could explore how cognitive biases might lead judges to
make factual determinations that appear intertwined with legal conclusions.
By exploring these perspectives, students can gain a deeper appreciation for how psychological
factors, institutional roles, and complex case dynamics contribute to the fluidity of the law-fact
distinction in real-world adjudication.
Rational Choice and Categorical Reason
Author(s): Bruce Chapman
Source: University of Pennsylvania Law Review , Jan., 2003, Vol. 151, No. 3 (Jan., 2003),
pp. 1169-1210
Published by: The University of Pennsylvania Law Review
Stable URL: https://www.jstor.org/stable/3312888
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/3312888?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
The University of Pennsylvania Law Review is collaborating with JSTOR to digitize, preserve
and extend access to University of Pennsylvania Law Review
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
https://www.jstor.org/stable/3312888
https://www.jstor.org/stable/3312888?seq=1&cid=pdf-reference#references_tab_contents
https://www.jstor.org/stable/3312888?seq=1&cid=pdf-reference#references_tab_contents
RATIONAL CHOICE AND CATEGORICAL REASON
BRUCE CHAPMANt
I. RATIONALITY AS A NORMATIVE IDEAL
The theory of rational choice, as understood by most economists
and many other social scientists, has both a normative and a positive
content. Normatively, it points to what should be done maximally to
achieve some given end, and, while it might not prescribe any particu-
lar end, it points to what it is to have a consistent set of ends that are
capable of being so maximized. For example, if an agent had a set of
ends that gave rise to a cyclical ordering of available alternatives, that
is, if she preferred x to y, y to z, and z to x, it would not be possible for
her to choose any one of these alternatives without another of the al-
ternatives being preferred to it according to her own criteria for
choice. In other words, it would not be possible for her to satisfy
completely, or maximize, her own ends.’
Positively, the theory of rational choice is used to describe, ex-
plain, and predict human behavior. Agents are assumed generally to
behave in an internally consistent way that can be rationalized by the
theory of maximization.2 Thus, if an agent has already chosen alterna-
tive y over alternative z, and then chooses alternative x over alternative
y, the assumption, and prediction, will be that the agent will choose
alternative x over alternative z.
tFaculty of Law, University of Toronto. I am grateful to Jack Knetsch, Rahul
Kumar, Arthur Ripstein, and all the participants at the University of Pennsylvania Law
School Symposium on Preferences and Rational Choice for their helpful comments on
an earlier draft. Research funding from the Social Sciences Research Council of Can-
ada and from the Connaught Fund at the University of Toronto is also gratefully ac-
knowledged.
For a good discussion of rational choice construed as maximization, and the
properties that are consequently required for the underlying preference relation, see
AMARTYA K. SEN, COLLECTIVE CHOICE AND SOCIAL WELFARE 7-20 (1970).
See, e.g., Amartya Sen, The Formulation of Rational Choice, 84 AM. ECON. REV. PROC.
385, 385 (1994) (noting that rationality is assumed to mean acting to maximize a utility
payoff); Amartya Sen, Maximization and the Act of Choice, 65 ECONOMETRICA 745, 746
(1997) (distinguishing “maximization,” which only requires choosing an alternative
that is not judged to be worse than any other, from “optimization,” which, more
strongly, requires choosing an alternative that is better than all others).
(1169)
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1170 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
Recently, the positive theory has come under attack from experi-
mental psychologists and economists.3 Their experimental results,
gathered together under the banner of behavioral analysis, show that
the maximizing model of rational choice often does not provide a very
accurate account of how agents actually choose. Moreover, the depar-
tures from the model appear systematic rather than random, suggest-
ing that something other than maximization is going on.
However, the general tenor of these studies is not to question the
normative ideal of maximization. Rather, the departures from the
standard account of rational choice are typically characterized, and
criticized, as failures to be rational. Agents are only human beings,
after all, and human beings are subject to the limitations that must,
inevitably and systematically, arise out of personal biases, limits on the
salience and availability of important information, and the distorting
effects of how a given problem is framed. Thus, real-world agents are
only, it is said, capable of a “bounded rationality,” using “rules of
thumb” and various “heuristics” (sometimes helpful, sometimes not)
rather than the fully fledged maximizing rationality that is still largely
accepted as the ideal for rational choice.4
In this Article, I argue that for many decision-making problems,
the normative account of rationality that animates rational choice
theory, and notjust the positive account that is criticized by the behav-
iorists, is deficient, even as a theory of ideally rational behavior. Ra-
The literature is now huge. Good selections can be found in CHOICES, VALUES,
AND FRAMES (Daniel Kahneman & Amos Tversky eds., 2000); JUDGMENT UNDER
UNCERTAINTY: HEURISTICS AND BIASES (Daniel Kahneman et al. eds., 1982); RATIONAL
CHOICE: THE CONTRAST BETWEEN ECONOMICS AND PSYCHOLOGY (Robin M. Hogarth
& Melvin W. Reder eds., 2d ed. 1987). For a wide-ranging textbook treatment of many
of the relevant issues, seeJONATHAN BARON, THINKING AND DECIDING (3d ed. 2000).
This is quite clearly Jonathan Baron’s view. See BARON, supra note 3, at 66 (not-
ing that, for one reason or another, people often fail to follow prescriptive models of
decision making and rationality). However, not all theorists of “bounded rationality”
think of its “boundedness” as setting constraints on what, ideally, rationality would
otherwise require of us. As one pair of theorists wrote:
Bounded rationality is, however, not simply a discrepancy between human
reasoning and the laws of probability or some form of optimization. Bounded
rationality dispenses with the notion of optimization and, usually, with prob-
abilities and utilities as well. It provides an alternative to current norms, not
an account that accepts current norms and studies when humans deviate from
these norms. Bounded rationality means rethinking the norms as well as
studying the actual behavior of minds and institutions.
Gerd Gigerenzer & Reinhard Selten, Rethinking Rationality, in BOUNDED RATIONALITY:
THE ADAPTIVE TOOLBOX 1, 6 (Gerd Gigerenzer & Reinhard Selten eds., 2001). On
this view, bounded rationality provides an alternative account of ideally rational behav-
ior.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CA TEGORICAL REASON 1171
tionality, I shall suggest, provides for an ordered particularity, including
particular decisions, but the notion of an ordering that informs this
alternative account of ideally rational behavior, and which is more ap-
propriate in some decision-making contexts (including many legal
ones), is very different from the idea of an ordering that informs the
standard account within rational choice theory. The latter, which, as
already suggested, is closely allied to the idea of maximization, remains
largely quantitative and single-minded in its orientation, this despite
the pluralism of motivations that it appears to be able and willing to
accommodate within its seemingly minimalist structure.5 The alterna-
tive account is more qualitative, or categorical (although not abso-
lute), offering a conception of a rational ordering of particularity that
is more allied to the idea of an understanding or interpretation (under
6
rules or principles) than it is to maximization. At the risk of import-
The structure of conventional rational choice is minimalist in the sense that it
only seems to require that an agent (1) be able to order any set of available alternatives
from best to worst, and (2) not choose an alternative x from that set if there is another
alternative that is better than (or more preferred to) x according to this ordering.
(More structure is required for rational choice over uncertain alternatives, including,
most importantly, the so-called “sure thing” principle. For a discussion of this princi-
ple, see infra note 35 and accompanying text.) Requirement (1) appears to be open to
any possible motivation or criterion for choice (including concerns for justice, altru-
ism, respect for the environment, process values, etc.); requirement (2), while captur-
ing the idea of maximization, seems to follow simply from taking these different moti-
vations or criteria seriously. Why settle on choosing some alternative if there is
another alternative available that is better according to one’s own criteria for choice?
However, as I hope to demonstrate in this Article, there is already enough in this ap-
parently minimalist structure to prevent us from accommodating some attractive (non-
maximizing) principles of choice as rational.
Compare the characterization of these two alternative accounts of rational deci-
sion making provided by Drazen Prelec:
Decision analysis, which codifies the rational model, views choice as a fun-
damentally technical problem of choosing the course of action that maximizes
a unidimensional criterion, such as value or utility. The primary mental activ-
ity … is the reduction of multiple attributes or dimensions to a single one,
through a specification of value trade-offs …. For rule-governed action, the
fundamental decision problem is the quasi-legal one of constructing a satisfy-
ing interpretation of the choice situation. The primary mental activity in-
volved in this process is the exploration of analogies and distinctions between
the current situation, and other “canonical” choice situations in which a single
rule or principle unambiguously applies.
Values and Principles: Some Limitations on Traditional Economic Analysis, in SOCIO-
ECONOMICS: TOWARD A NEW SYNTHESIS 131, 131 (Amitai Etzioni & Paul R. Lawrence
eds., 1991). For some suggestions about how the differences between maximization
and reasoning by analogy might be captured in choice theoretic terms, see Bruce
Chapman, The Rational and the Reasonable: Social Choice Theory and Adjudication, 61 U.
CHI. L. REv. 41 (1994).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1172 UNIVFRSITYOFPENNSYLVANIALAWREVIEW [Vol. 151: 1169
ing some unnecessary baggage, but for reasons that I hope will be-
come clearer as the argument unfolds, I refer to this alternative con-
ception of rationality as categorical reason. If that phrase suggests a
longstanding rationalist tradition, exemplified by Kant, but rejected
by the British empiricists like Hobbes and Hume, who are the most
likely intellectual ancestors of contemporary rational choice theorists,
that is not entirely unwelcome.7
The real challenge for this Article, however, is not so much to ar-
ticulate two alternative accounts of rationality that have had some tra-
ditional followers, but to begin to make each accessible to the other
within some common intellectual framework. While I think rational
choice theory provides a useful and precise set of tools for beginning
this process of achieving mutual understanding between the tradi-
tions, I shall argue that some quite fundamental postulates of rational
choice theory (including some of the axioms of choice consistency
and strong independence) will have to be relaxed if the contributions
of categorical reason are properly to be accommodated within
it.
However, I hope to show that there is much advantage in this, even
for what the rational choice theorist hopes to achieve, and to illustrate
For a concise account of the intellectual origins of rational choice theory in the
works of Hobbes and Hume, see Martin Hollis & Robert Sugden, Rationality in Action,
102 MIND (n.s.) 1, 2-7 (1993). In interpretations of Kant, the word “categorical,” as in
“categorical imperative,” is often thought to mean “absolute” or “without qualifica-
tion”; for an example of this interpretation, see CHARLES FRIED, RIGHT AND WRONG 9-
13 (1978). This is not the meaning of “categorical” I mean to suggest in my phrase
“categorical reason.” See infra text following note 77 (discussing the concept of “cate-
gorical reason”). Rather, I mean something more like “within categories” or “accord-
ing to rules,” as in the following:
Everything in nature, both in the lifeless and in the living world, takes place
according to rules, although we are not always acquainted with these rules ….
The whole of nature in general is really nothing but a connection of appear-
ances according to rules; and there is no absence of rules anywhere. If we be-
lieve we have found such a thing, then in this case we can only say that we are
not acquainted with the rules.
The exercise of our powers also takes place according to certain rules that
we follow, unconscious of them at first, until we gradually arrive at cognition of
them through experiments and lengthy use of our powers, indeed, until we
finally become so familiar with them that it costs us much effort to think them
in abstracto ….
Like all our powers, the understanding in particular is bound in its actions to
rules, which we can investigate. Indeed, the understanding is to be regarded
in general as the source and the faculty for thinking rules in general. For …
the understanding is the faculty for thinking, i.e., for bringing the representa-
tions of the senses under rules.
IMMANUEL KANT, LECTURES ON LOGIC 527 (J. Michael Young ed. & trans., Cambridge
Univ. Press 1992) (1800).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1173
the point by reference to some systematic difficulties that the rational
choice theorist faces in the theory of social choice and game theory.
Part II reviews the results of some recent behavioral experiments
that suggest that agents respond to reasons in a way that is not always
consistent with some of the fundamental axioms of (value-based) ra-
tional choice. I look at choice involving certain and uncertain alter-
natives, and focus on the weak axiom of revealed preference in the
former and the sure thing principle in the latter. My claim is that
while some of the choices that some of these experimental subjects
make do seem problematic from a rational point of view, sensible sce-
narios can be constructed that make good sense of these systematic
violations of the rationality axioms.
In Part III, I argue that common law adjudication manifests the
same tension between reason-based choice and conventional (value-
based) rational choice that was shown in the experiments. However, I
argue that the common law idealizes reason-based choice, insisting
not only that a claimant be right, but that a claimant be right and ra-
tional-that is, right for the right reasons. I refer to this reason-based
ideal as categorical reason.
In Part IV, I suggest that the idea of categorical reason can be use-
ful both in the theory of social choice and in the theory of noncoop-
erative games. In social choice, categorical reason brings a kind of
conceptual discipline to the preferences that can be admitted into so-
cial choice, and this helps to avoid certain problems of instability and
collective irrationality. In the theory of games, categorical reason
publicly organizes the particularity of individual agents’ choices so
that coordination and cooperation are more likely to occur.
II. RATIONAL CHOICE BEHAVIORISM AND REASON-BASED CHOICE
A. The Case of Certainty
One might have thought, or even hoped, that a theory of rational
choice would have informed us about how people think or deliberate
about their decisions, or about how their choices are explained orjus-
tified by reasons. That, typically, is how a legal theorist would under-
stand the obligation to offer an account of rational decision making.
At the end of their article on “reasons,” for example, John Gardner
and Timothy Macklem conclude that rationality “is simply the capacity
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1174 UNIVERSITYOFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
and propensity to act (think, feel, etc.) only and always for undefeated
reasons.,8
However, the agenda for developments in the economic theory of
rational choice has, apparently, been one of psychological reduction-
ism. The idea, which began with Vilfredo Pareto’s replacement of
cardinal with ordinal utility as a motivation for choice in the early part
of the twentieth century,9 has been to rely less and less on any claims
about what might be going on in someone’s head.’0 With the advent
of revealed preference theory, as originally developed by Paul Samuel-
son in the 1930s, ” the expunging of anything psychologically substan-
tial that might explain a set of rational choices, like the maximization
of self-interest, utility, or, now, even preference, is more or less complete.
What matters for rationality is the consistency of externally observable be-
havior, not any particular subjective motivation.” This reliance on
what is objectively observable is typically thought to be “scientifically
more respectable”’13 than any attempt to speculate about, and model,
private thoughts, motivations, or reasons.
Of course, the requirements of a rational consistency of observ-
able choice are not unrelated to the requirements of a rational maxi-
mization of unobservable preference or utility. Indeed, the former,
while considered a fully autonomous subject matter for the scientific
and systematic study of choice, is still typically thought capable of be-
ing “rationalized” by the latter. Thus, Samuelson’s Weak Axiom of Re-
vealed Preference (“WARP”), 14 which is still the central postulate of the
new behaviorism in cases of choice over certain outcomes, has been
John Gardner & Timothy Macklem, Reasons, in THE OXFORD HANDBOOK OF
JURISPRUDENCE AND PHILOSOPHY OF LAw 440, 474 Jules Coleman & Scott Shapiro
eds., 2002).
VILFREDO PARETO, MANUAL OF POLITICAL ECONOMY 103-209 (Ann S. Schwier &
Alfred N. Page eds., Ann S. Schwier trans., Augustus M. Kelley Publishers 1971) (1927).
10 For a discussion of the historical developments in the theory of utility as a moti-
vation for rational choice, see MARK BLAUG, ECONOMIC THEORY IN RETROSPECT 343-53
(3d ed. Cambridge Univ. Press 1978) (1962).
P.A. Samuelson, A Note on the Pure Theory of Consumer’s Behaviour, 5 ECONOMICA
(n.s.) 61, 61-71 (1938).
12 See, e.g., J.R. HICKS, A REVISION OF DEMAND THEORY 6 (1956) (“[T]he
econometric theory of demand does study human beings, but only as entities having
certain patterns of market behaviour; it makes no claim, no pretence, to be able to see
inside their heads.”).
13 See I.M.D. Little, A Reformulation of the Theory of Consumer’s Behaviour, 1 OXFORD
ECON. PAPERS (n.s.) 90, 97 (1949) (noting that objective observation has been deemed
“scientifically more respectable” because it allows explanation of behavior “without
reference to anything other than” behavior).
14 Samuelson, supra note 11, at 62-70.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1175
shown to be logically implied by, and consistent with, what would be
chosen by a rational maximizer of preferences.’5 According to WARP,
if an agent ever chooses an alternative x over alternative y from some
set of alternatives, then that agent should never (on pain of inconsis-
tency) choose alternative y over alternative x from any other set of
available alternatives.16
One is tempted to add “unless, of course, her preferences have
changed,” but this would be to seek refuge in a preference-theoretic
explanation of a possible departure from what is supposed to be a
purely choice-theoretic requirement. Nevertheless, the temptation is
revealing in that it shows what really lies behind WARP as a plausible
requirement for rational choice. The idea, surely, is that an ideally ra-
tional agent can arrange all conceivable alternatives in order of prefer-
ence and would choose, from any available subset of those alternatives,
the one that was highest in that ordering. Such an agent would never
violate WARP. Further, an agent satisfying WARP would always
choose as if she had such a preference ordering and were maximizing
it.
At first glance this last claim might seem odd, since there does not
seem to be enough in WARP to generate the thought that there must
be an underlying transitive preference relation motivating choice. For
example, if an agent chose x from the choice set (x, y), y from the
choice set (y, z), and z from the choice set (x, z), there would not yet
be any violation of WARP, although the choices do seem to reveal an
intransitive preference ordering, something that can frustrate maxi-
mization.17 However, the violation of WARP is manifest if we can re-
quire that the agent now show us a consistent choice over the choice
set (x, y, z), that is, if we take seriously the idea that the agent must, in
a way analogous to the complete preference requirement, be consis-
tent in her choices over any conceivable set of available alternatives.
For, given her first three choices over the three different pairs, the
agent cannot now choose any alternative from the triple (x, y, z) with-
See Kenneth J. Arrow, Rational Choice Functions and Orderings, 26 ECONOMICA
(n.s.) 121, 126 (1959) (“The most interesting conclusion is the complete equivalence
of [WARP] with the existence of an ordering from which the choice function can be
derived.”); see also Amartya K Sen, Choice Functions and Revealed Preference, 38 REV.
ECON. STUD. 307, 310-11 (1971) (proving the logical equivalence of WARP with choice
under a rational preference relation).
16 Arrow, supra note 15, at 123.
See supra text accompanying note 1 (introducing the normative theory of ra-
tional choice as maximization).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1176 UNIVERSITY OF PENNSYLVANIA LAWREVIEW [Vol. 151: 1169
out violating WARP.18 Thus, a rational maximizer of preferences will
choose in a way that satisfies the behavioral requirements of WARP,
and an observable chooser satisfying the requirements of WARP will
choose as if she had a fully transitive preference ordering that she was
maximizing. It does not appear, therefore, that the behaviorist revo-
lution has offered up any real surprises at the level of principle.
Of course, where behaviorism has offered up something new is in
the recent experimental research that shows that agents do not actu-
ally choose in the way that the most minimal consistency conditions,
like WARP, seem to require. That is, the choice between two alterna-
tives x and y can vary-indeed, it can be reversed-according to what
else is in the available set of alternatives. To the extent that this ap-
pears to be systematic and predictable, it suggests that these choices
cannot be rationalized as the maximization of preference. It is a dif-
ferent question, perhaps, whether they can be rationalized at all.
The most interesting experimental results for the purposes of this
Article are those offered by Eldar Shafir, Itamar Simonson, and Amos
Tversky around the idea of “reason-based choice.” 19 These authors
begin by contrasting reason-based choice with value-based choice, the
latter being their name for the economic theory of rational choice.
On this latter view, a value is associated with each alternative and
choice is characterized as the maximization of value. Reason-based
choice, on the other hand, is more characteristic of legal scholarship’s
and analyses of historically significant case studies.2′ It “identifies vari-
ous reasons and arguments that are purported to enter into and in-
fluence decision, and explains choice in terms of the balance of rea-
sons for and against the various alternatives.”22 This, they admit, is a
somewhat vague characterization of rational choice, and it might not
be clear why there would necessarily be any incompatibility between
value- and reason-based choice. Surely the “values” of different alter-
18 See AmARTYA SEN, CHOICE, WELFARE AND MEASUREMENT 58 (1982) (“[N]o mat-
ter what he chose out of this set of three alternatives . . . he must violate [WARP].”).
19 Eldar Shafir et al., Reason-Based Choice, in CHOICES, VALUES, AND FRAMES, supra
note 3, at 597, 597-619; see also Amos Tversky & Itamar Simonson, Context-Dependent
Preferences, 39 MGMT. SCI. 1179, 1179 (1993) (discussing empirical findings inconsistent
with value maximization and presenting a context-dependent model of choice).
Shafir et al., supra note 19, at 598; see also infra Part III (suggesting that common
law adjudication is a form of reason-based choice).
21 For example, reason-based choice analysis has been applied to a study of the
Cuban Missile Crisis. GRAHAM T. ALLISON, ESSENCE OF DECISION: EXPLAINING THE
CUBAN MISSILE CRISIS (1971), cited in Shafir et al., supra note 19, at 598.
22 Id.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CA TEGORICAL REASON 1177
natives provide good reasons for choosing them, the argument might
go, and it seems natural to think that the balance of values would pro-
vide a good indication of where the balance of reasons is ultimately to
be found. However, their precisely constructed experiments serve to
indicate more clearly what is meant by reason-based choice and why
the conflict with value-based choice can potentially arise.
The experiments show that agents will often latch onto a reason
for choosing a particular alternative just to resolve the conflict that
they feel in facing choice. The “irrationality,” at least from a rational
choice perspective, is that almost any reason, including a seemingly
“irrelevant” one, will do. For example, some subjects were asked to
choose between alternative x, six dollars in cash, and y, a high quality
pen.23 The pen was chosen by 36% of the subjects and the remaining
24
64% chose the cash. However, when the subjects were presented
with a choice from these same two options together with a third, z,
another pen that was of clearly inferior quality to the first, then the
percentage of subjects that chose y, the higher quality pen, rose dra-
matically.25 This appears to suggest that many subjects who would
choose x over y, when only those alternatives are available, will choose
y over x when some third alternative, z, is added to the set of available
alternatives, a violation of WARP.
The explanation offered is that these subjects now have a reason to
choose y, namely, that it is a pen of clearly superior quality to z, a rea-
son that they did not have when z was unavailable.26 However, this pat-
tern of choices does appear somewhat “irrational.” The fact that y is
an alternative that is obviously better than z provides a good reason
for choosing y over z, but it appears to provide little reason for choos-
ing y over a quite different alternative x. Indeed, as nothing about the
values of x and y is changed by introducing z into the choice set, one
might have thought that nothing would change for a rational chooser
in a value-based choice between x and y. This property of “rational”
consistency is what is captured by WARP and what is violated so sys-
tematically by the experimental results.
Nevertheless, it is not difficult to construct a different scenario
where it seems more sensible to think that the addition of some alter-
native z might change the choice between alternatives x and y. Sup-
23 Id. at 609.
24 Id.
25 Id.
26 Id. at 610.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1178 UNTVERSITY OF PENNSYLVANIA LAWPREVIEW [Vol. 151: 1169
pose, for example, that someone is offered a choice of fruit at the end
of a dinner.27 If only a large apple, A, and a large orange, 0, are of-
fered to her, she would choose the large apple. Both fruits are large
and, all else equal, she prefers apples to oranges. However, if she is
offered A, 0, and a small apple, a, then different considerations arise.
For now there is an issue of etiquette to be addressed. The rule, let us
say, is that one should never choose the larger of two items of the
same kind. Our chooser now reasons that, in the choice from the set
(A, 0, a), she cannot now choose A, because that would be in breach
of the rule of etiquette. She therefore chooses 0, a piece of fruit that
is larger than a, but a fruit of a different kind. Thus, from the set of
alternatives (A, 0), she chooses A; but from the set of alternatives (A,
0, a), she chooses 0, a violation of WARP.
The chooser would also reveal an intransitive preference ordering
if the different fruits were offered to her in pairs. She would choose A
from the pair (A, 0), 0 from the pair (0, a), and a from the pair (a,
A), in violation of transitivity. The reason, of course, is that the rule of
etiquette does not come into play until the third choice, when the big
and small apples are presented together. Until that point the chooser
can select between the fruits purely according to taste, or according to
the different values of the different alternatives, choosing the highest
valued one; in other words, she can choose in the way that the theory
of value-based choice and maximization suggests. But when the two
apples are presented together, etiquette becomes an issue between
them, that is, as an issue bearing on the relationship between those two
alternatives, not as a property or value of either of the two alternatives
considered on its own. In this way we can say that the concern for eti-
quette, unlike the concern for taste, is a partition-dependent or categorical
idea; it arises only when the two alternatives, a and A, appear together
within some partition of the alternatives.”
27 This example is now much discussed. The earliest published analysis of it of
which I am aware is in Philip Pettit, Decision Theory and Folk Psychology, in FOUNDATIONS
OF DECISION THEORY 147, 163 (Michael Bacharach & Susan Hurley eds., 1991). The
example, and close variations of it, is also analyzed in Paul Anand, The Philosophy of In-
transitive Preference, 103 ECON. J. 337, 344 (1993); Bruce Chapman, Law, Incommensura-
bility, and Conceptually Sequenced Argument, 146 U. PA. L. REV. 1487, 1498-99, 1503-05
(1998); Amartya Sen, Internal Consistency of Choice, 61 ECONOMETRICA 495, 501 (1993).
See also Amartya Sen. Is the Idea of Purely Internal Consistency of Choice Bizarre?, in WORLD,
MIND, AND ETHICS: ESSAYS ON THE ETHICAL PHILOSOPHY OF BERNARD WILLIAMS 19, 24
(J.E.J. Altham & Ross Harrison eds., 1995) [hereinafter Sen, Bizarre] (showing the epis-
temic relevance of a different menu of alternatives).
For other examples used to make the same point, see JOHN BROOME, WEIGHING
GOODS: EQUALITY, UNCERTAINTY, AND TIME 100-01 (1991); ISAAC LEVI, HARD
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CA TEGORICAL REASON 1179
The etiquette example provides, therefore, another instance of
reason-based choice pulling the chooser in a direction different from
that prescribed by the logic of value-based choice. The presence of a
in the set of alternatives gives our chooser a reason in etiquette for not
choosing A. But, just as for the experimental results referred to ear-
lier, this is a reason that does not seem to be relevant to any compari-
son between A and 0; it is a very partition-dependent consideration.
The values of A and 0 as alternatives would appear to be unchanged,
and the choice between them, one might have thought, would be un-
affected by such an “irrelevant” reason.
Yet, despite this apparent “irrationality,” what is happening in the
etiquette example is hardly incomprehensible to us, at least if we have
any sort of feel for the rule of etiquette that is involved. We simply
understand the choice situation (A, 0, a), where both A and a are pres-
ent and etiquette is at issue, differently from the choice situation (A, 0),
where a is absent and etiquette is not at issue. And this different un-
derstanding, which turns on the availability of an alternative that is it-
self never chosen, requires that a different choice be made over those
two alternatives A and 0, which were always available for choice.
Thus, it is not as if the different understanding arises simply because a
different set of available alternatives means we can now choose some-
thing different, and more particularly something better, which was not
available earlier.29 That sort of different understanding, which does
CHOICES: DECISION MAKING UNDER UNRESOLVED CONFLICT 33, 105 (1986); James F.
Reynolds & David C. Paris, The Concept of ‘Choice’ and Arrow s Theorem, 89 ETHICS 354,
363 (1979).
29 Not surprisingly, this is how the committed rational choice theorist typically
solves the etiquette problem and others like it. See, e.g., BARON, supra note 3, at 235
(arguing that, because of the relevance of etiquette, a large apple does not mean the
same thing when compared to a large orange as it does when compared to a small ap-
ple). What appears as an inconsistent choice over the same pair of alternatives is actu-
ally a choice over a different pair of alternatives and, therefore, the issue of inconsis-
tency cannot arise. Choosing “the big apple A from a set where the little apple a is
available” is simply not the same as choosing “the big apple A from a set where the lit-
tle apple a is not available,” or so the argument goes. We might even capture this idea
by more accurately relabeling the two alternatives for choice as A/a and Al-a, respec-
tively (where A/a is read “A when a is also available” and Al-a is read “A when a is not
also available”). Thus, the apparent inconsistency of choosing A from the set (A, 0)
and 0 from the set (A, 0, a), for example, is resolved under this redescription of the
problem as choosing A/-a from the set (A/-a, 0) and 0 from the set (A/a, 0, a), in
perfect conformity with choice consistency conditions like WARP. This “solves” the
difficulty, but only at the cost of rendering the choice consistency requirements vacu-
ous. As Sen observed, “[iff every time the set from which the choice is being made
changes, the choice of any given alternative … is taken to be a different choicej . ..
then no condition of internal consistency of choices from different subsets can make
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1180 UNIVERSITY OFPENNSYLVANIA LAWRPEVIEW [Vol. 151: 1169
not affect our understanding of the previously available alternatives,
would always be relevant to a maximization of value. Rather, the dif-
ferent understanding arises because the addition of the new alterna-
tive changes how the previously available alternatives, themselves un-
changed, are now conceived. And this new understanding changes how
we choose between those previously available alternatives.30 Thus, this
any demand whatsoever.” Sen, Bizarre, supra note 27, at 26. This is a heavy price to pay
to “secure” the conventional requirements of rationality in rational choice theory.
Moreover, the very act of redescribing the alternatives according to what else is avail-
able in the choice set concedes the point at issue, viz., that what we are doing in choice,
and what we want to do under that description, varies with the set of available alterna-
tives. This variation is only obscured, and not preserved as a subject requiring more
thorough analysis, if we simply provide a new set of partition-dependent descriptions of
the alternatives to preserve a partition-independent choice consistency condition. To his
credit, John Broome has at least recognized that the redescription strategy must be re-
fined so that something of the original force of the conventional rationality conditions
can be preserved without reducing them to the worst forms of “ad hocery.” See
BROOME, supra note 28, at 102-07 (discussing the recognition of rational requirements
of indifference as a way of dealing with the problem of emptiness). However,
Broome’s refinement strategy, which allows alternatives to be “individuated” in the way
described earlier, either begs the question (in that rationally justified differences in
choice still have to be justified as differences between the alternatives separately con-
sidered) or generates a quite different problem for rational choice in that some alter-
natives cannot now logically be compared, something which violates the completeness
requirement. For a discussion of the relationship between completeness and rational-
ity, see infra text accompanying notes 80-81.
For an interesting paper relating the conventional choice consistency condi-
tions (like WARP) to the equally conventional monotonicity requirements of classical
logic, see RuVIN GEKKER, NONMONOTONIC REASONING AND THE FOUNDATIONS OF
RATIONAL CHOICE (European Pub. Choice Soc’y, Working Paper No. 61, 2002), avail-
able at http://www.economics.nuigalway.ie/downloads/papers/0061paper . Under
monotonicity, if some proposition p is sufficient to imply another proposition q, then
the compound proposition p and r should also imply q; in other words, the sufficiency
of p for q should not be undermined by adding r. However, under non-monotonic or
defeasible reasoning, this is precisely what is relaxed. While p on its own might be suf-
ficient for q, the addition of proposition rcan imply not-q.
In the etiquette example, the non-monotonicity is found in the following: Let
proposition p be “options A and 0 are available for choice,” let proposition q be “I
should choose A and not 0,” and let proposition rbe “option a is available for choice.”
Then, under the obligation to choose something according to the desire to eat (larger
pieces of) fruit and the rules of etiquette, while p implies q, p and r implies not-q. (In
fact p and r implies, “I should choose 0 and not A.”)
It is arguable that much of legal reasoning and legal argument proceeds non-
monotonically. Certainly the desire to provide an adequate model of legal reasoning
has been one of the great motivators for the development of non-monotonic or defea-
sible logics in recent years. See, e.g., JAAP C. HAGE, REASONING WITH RULES: AN ESSAY
ON LEGAL REASONING AND ITS UNDERLYING LOGIC, at xiii (1997) (discussing defeasibil-
ity of reasoning with rules, “in particular legal rules”). For example, the addition of a
certain legal defense r, while not relevant or even admissible as a consideration until
the prima facie case p is in place (it is a defense after all), can reverse or defease the le-
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1181
is different choice under a truly different understanding of the (same)
alternatives (themselves unchanged in value); it is not different choice
simply because (trivially, we now understand that) differently valued
alternatives are now available. Where WARP allows the latter role for
an altered “understanding,” it does not allow the former.
Yet the idea that we might choose differently over empirically in-
distinguishable alternatives because we have a different understanding
of the issues that are at stake in our choice is hardly novel. Consider
the example of how one chooses as a friend, something Aristotle dis-
cussed at some length.) It is widely appreciated that there is some-
thing problematic about choosing to be someone’s friend for instru-
mental reasons, especially, say, if one is being a friend (or seeking the
good for one’s friend) simply because it makes one better off. There
may be reciprocity or mutual “back scratching” in that, but it fails
fundamentally as friendship. The test (although not, of course, the
end) of true friendship is in the willingness to continue acting as a
friend even if doing so makes one worse off.
This much is elementary, but we can take the basic insight further.
A true friend cannot even have the value of having friends as the rea-
son she does what friends do. That is also too instrumental; it puts the
value of having friends before the friendship itself. Even if the actor
sees her conduct as perfecting friendship, or seeking (too much) to
do what it is that friends do, it would still be too calculating and too
(self-consciously) goal-oriented for genuine friendship. But suppose
she says, in response to some proposal, “That’s not what friends do; I
cannot do that.” Then the concept of friendship informs what she
does, although it does not guide what she does in the way that a goal
(e.g., the goal or value of achieving or maintaining friendship) might.
She acts under an understanding of what friendship is, even of what
friendship requires, but she does not, strictly, act that way because
friendship requires it. The latter suggests too much that there is a
gal outcome q that would otherwise be implied by p. I have argued elsewhere that de-
feasible rules provide an innovative, rational, and peculiarly legal structure for the ac-
commodation of plural values in collective decision making, and one that cannot be
captured within the conventions of rational choice theory. Bruce Chapman, Law
Games: Defeasible Rules and Revisable Rationality, 17 LAW & PHIL. 443, 446 (1998);
Chapman, supra note 27, at 1494-95; see alsoJohn L. Pollock, A Theory of Moral Reason-
ing, 96 ETHICs 506, 512-20 (1986) (arguing that defeasibility should provide the logical
structure for moral reasoning more generally).
31 Aristotle devotes two books of The Nicomachean Ethics to friendship. SeeJohn M.
Cooper, Aristotle on Friendship, in EssAYs ON ARISTOTLE’S ETHICS 301, 301-40 (Amelie
Oksenberg Rorty ed., 1980) (discussing Aristotle’s writings on friendship).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1182 UNIVERSI7’Y OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
moment when she can understand the alternatives for choice inde-
pendent of the concept of friendship, and then go on to choose
amongst them as friendship, now brought to bear upon the choice,
would have her do. But the concept or category of friendship in-
forms, and orders, the particularity of her choices in a more gapless or
immediate way. She sees the choice as directly implicating friendship,
as a particular immediately implicates the category of which it is an in-
stance. And she chooses as a friend, with the result that, just as imme-
diately, she instantiates the category in the particularity of what she
does. In this way, the concept of friendship can be the reason for, or
the ordering of, her acting the way she does, even if the value of
friendship cannot.
This is not to say that there is no value to be achieved in friend-
ship or that friendship does not connect to something valuable. If
everyone acts under the aspect of friendship, where friendship ration-
ally orders or gives reason to what they do, then the good of friend-
ship is likely to be achieved and enjoyed. And there is value in that.
But still, it seems quite plausible to say that the value of friendship is
no part of our rationale for acting as friends. We act one way rather
than another simply because we know what it is to be friends. The
value we achieve is merely an incidental by-product, maybe even an
essentially incidental by-product,32 of our acting on this knowledge and
for this reason.
This digression into the notion of friendship has served to illus-
trate that there is a long and durable tradition in the idea that an un-
derstanding, or a concept, can act as a reason for choice. This tradi-
tion also suggests that there is a difference between reason based on
such a concept and reason based on value. The importance of the
etiquette example is that it shows precisely where this difference is to
be found within the axioms of rational choice theory. Where a differ-
ent concept gives rise to a different understanding of the alternatives
available for choice, even a different understanding of all those alter-
natives that continue to be available as other alternatives change, then
choice may vary according to that changing understanding, even for
2 For a discussion of social states that are “essentially by-products,” that is, states
that “can only come about as the by-product[s] of actions undertaken for other ends,”
see JON ELSTER, SOUR GRAPES: STUDIES IN THE SUBVERSION OF RATIONALITY 43
(1983). See also Robert Sugden, Rational Choice: A Survey of Contributions from Economics
and Philosophy, 101 ECON. J. 751, 781 (1991) (linking Elster’s idea to problems of self-
defeating rationality in the theory of rational choice and, in particular, the problem of
rational commitment).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CA TEGORICAL REASON 1183
those alternatives that continue to be available. While this might trouble
the rational choice theorist committed to certain choice consistency
conditions like WARP, conditions that make sense on a view of ra-
tional choice that is value-maximizing and goal-oriented, it is less clear
that such variations should surprise those whose conception of ra-
tional choice is to be found more in the idea that particular decisions
are rational to the extent that they can be ordered, or organized, un-
der the aspect of different concepts, understandings, or categories of
thought.
B. The Case of Uncertainty
The discussion so far has related to choice over certain alterna-
tives. We have been questioning whether the idea of maximizing
one’s preferences (or values) over those alternatives, even if that only
takes its behaviorist form as a choice consistency condition like WARP,
is the only sensible conception of rational choice, or whether an al-
ternative conception, sensitive to the different understandings that a
chooser might bring to a choice problem, might also provide an ac-
count of rational ordering of particular decisions.
However, the most general theory of rational choice, which pur-
ports to rationalize behavior as expected utility maximization, deals with
choice over risky alternatives.33 Of course, to handle the more general
case, some further choice axioms are required. In this Section, we will
focus on one in particular, namely, the strong independence assump-
tion or “sure thing principle.”34
The sure thing principle has been characterized as “[t]he key
qualitative property that gives rise to expected utility theory,”35 and so
Within this more general theory, certain choice is interpreted as the trivial case
where the probabilities for the different possible outcomes are reduced to either one
or zero.
34 For a discussion of the axioms underlying expected utility maximization (or the
idea that an agent’s observable choices over uncertain alternatives can be rationalized,
or represented, as the maximization of that agent’s expectation of utility), see R.
DUNCAN LUCE & HowARD RAIFFA, GAMES AND DECISIONS 23-31 (1957). Luce and
Raiffa refer to the strong independence assumption as “substitutibility.” Id. at 27. For
a discussion of the possible origins of the name “sure thing principle,” see LEONARDJ.
SAVAGE, THE FOUNDATIONS OF STATISTICS 21-24 (1972). “Strong independence” is the
name that Paul Samuelson uses for the axiom. See Paul A. Samuelson, Probability, Util-
ity, and the Independence Axiom, 20 ECONOMETRICA 670, 670 (1952) (asserting that
strong independence conditions “create the existence of certain special or canonical
indexes of utility and probability that are additive”).
Amos Tversky & Daniel Kahneman, Rational Choice and the Framing of Decisions, in
CHOICES, VALUES, AND FRAMES, supra note 3, at 209, 210.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1184 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
it is important to have a sense of what it means. Roughly, the princi-
ple allows the chooser to cancel or eliminate from her consideration
any possible state of the world that yields the same outcome (the “sure
thing”) regardless of one’s choice. Suppose, for example, that one
can choose between two lotteries, A and B, where A provides for a sea-
side vacation if one draws (from 100 possible numbered tickets) a
number between 1 and 10 and $100 for any number between 11 and
100, and B offers a mountain vacation on the chance that you draw a
number between 1 and 10 and $100 for any number between 11 and
100. Since all the possible draws of numbers between 11 and 100 yield
the same outcome of $100, the sure thing principle tells us that the
choice between the lotteries should turn only on the chooser’s prefer-
ence between the possibilities of vacationing at the seaside or in the
mountains.
Suppose that the chooser indicates a preference for lottery A over
lottery B, indicating a preference for the seaside vacation. Now con-
sider a third lottery, C, which offers a seaside vacation if one draws a
number between 1 and 10 and $200 if one draws a number between
11 and 100. By the same sure thing principle, the choice between lot-
tery C and lottery A should turn on the chooser’s preference between
$200 and $100. Suppose, as seems reasonable, that the chooser pre-
fers lottery C because, all other (vacation) possibilities equal, it offers
(the same chance of) a larger cash award. Thus, C is preferred to A, A
is preferred to B, and, by transitivity, C is preferred to B. And surely
this last implication (based on transitivity) makes sense even on a di-
rect application of the sure thing principle. If lottery C is preferred to
lottery B in every possible world that might occur (i.e., because of the
preference for a seaside vacation over a mountain vacation if a num-
ber between 1 and 10 is chosen, and because of the preference for
more cash rather than less in the event that a number between 11 and
100 is chosen), then lottery C should be preferred to lottery B even
when the exact state of the world (or which number might be chosen)
is still unknown. If this were not the case, that is, if lottery B were pre-
ferred to lottery C, then there would have to be something attractive
in the lottery as such, that is, in the particular combination of (mutually
exclusive) possibilities that B offers, namely, a less preferred vacation
and less cash, which allows lottery B to be more attractive for the
chooser even though each possibility considered on its own is dis-
preferred. This would, of course, violate the strong independence prop-
erties of the
sure thing principle.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1185
However, despite the apparent rationality of the sure thing prin-
ciple, behavioral research indicates that it is systematically violated by
experimental subjects. The subjects appear to reveal what the behav-
iorists have called a “disjunction effect.”36 That is, the subjects will in-
dicate a preference for some choice x over another choice y when they
know that event A obtains, and they will also indicate this same pref-
erence when they know that event A does not obtain, but they will in-
dicate a preference for y over x when it is unknown whether or not A
obtains. This is a violation of the sure thing principle. What is most
interesting about this research is the explanation for why this disjunc-
tion effect occurs. As we shall now see, the explanation again points
to the importance of reason-based choice.
In one of these experiments, all the student subjects were asked to
imagine that they had just taken a very difficult qualifying examina-
tion near the end of the fall term.37 Some of these students were then
asked whether they would take advantage of a very attractive holiday
38
package in Hawaii if they knew they had passed the exam. Others
were asked whether they would take advantage of the same package if
they knew they had failed the exam.39 A majority of the students indi-
cated that they would choose the vacation package in each of these
two possible states of the world.40 However, when the students were
asked if they would choose the package if they did not know whether
they had passed or failed, less than a third of the students chose the
package and more than 60% were willing to pay five dollars to post-
pone the decision until the following day when they would know the
results of the examination.41 The fact that students (i) are unwilling to
commit to the vacation when there are still two uncertain possibilities
before them, namely, pass or fail the examination, but (ii) will choose
the vacation in the event that either one of these two possibilities be-
comes a certainty, illustrates the disjunction effect and violates the
sure thing principle.
36 See, e.g., Shafir et al., supra note 19, at 612-13 (“Evidently, a disjunction of differ-
ent reasons (reward in case of success or consolation in case of failure) is often less
compelling than either definite reason alone.”); Eldar Shafir & Amos Tversky, Thinking
Through Uncertainty: Nonconsequential Reasoning and Choice, 24 COGNITIVE PSYCHOL. 449,
449 (1992) (explaining that “[u]ncertain situations may be thought of as disjunctions
of possible states”).
3 Shafir et al., supra note 19, at 611.
38 Id. at612.
39 Id.
40 Id.
4 Id. at611.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1186 UN!VERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
The explanation that is offered for this behavior concerns the rea-
sons that the students give for wanting to go on the vacation in each of
the possible states of the world. When the student knows that she has
passed the exam, she reasons that the vacation is ajust reward for her
success; when the student fails, she construes the vacation as providing
some kind of consolation. Thus, whatever the outcome of the exami-
nation, the student has a good reason-albeit a different reason for
each possible outcome-to take the vacation. However, when the
outcome of the examination is unknown, the student lacks either one
of these as “a definite reason”42 for taking the vacation. Thus, under
this last scenario, it is as if she knows it is right for her to go on vaca-
tion, but not that it is right for any particular reason. This, somehow,
makes it hard for her to go.
We could also interpret her difficulty in the following way: when
she knows the outcome of the examination, she knows the reason that
is relevant to her choice and, therefore, she knows exactly what it is
that she is doing when she goes on vacation. When she knows she has
passed, for example, she knows that this is a “reward trip.” Likewise,
when she knows that she has failed, she knows that this is a “consola-
tion trip.” In other words, she can organize what she is doing under
some kind of category or understanding. But when she does not
know the results of the examination, it is as if her vacation lacks any
such identity, or meaning, for her. Again, it is as if she knows what
she should do (after all, she knows that she has no reason to do oth-
erwise), but she lacks any particular understanding of what it is that
she is doing when she does it. To the extent that such an understand-
ing might bring order, or rationality, to the particularity of what she
does, her reluctance to go, far from manifesting irrationality (as sug-
gested by the violation of the sure thing principle), evidences the im-
portance of this alternative conception of rationality for her behavior.
Indeed, without this alternative rationality to bring order to what she
does, she will pay five dollars to wait the one day so that she will know
what it is that she is doing. And she will do this despite the fact that
waiting the extra day will not change what she will do.
The vacation example suggests the importance for choice of hav-
ing an understanding of what it is that one is doing, as opposed to
having no such understanding at all (or, at least, an ambiguous or
equivocal understanding because of uncertainty). However, other ex-
periments reveal that the presence of uncertainty, giving rise to a dis-
42 Id. at 612.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1187
junction effect, can generate, more positively, an alternative under-
standing to that arising from choice under certainty and, therefore,
give rise to choices that violate the sure thing principle because there
is this different understanding informing choice. Consider, for ex-
ample, what Eldar Shafir and Amos Tversky unearthed about how sub-
43
jects play the familiar two person prisoner’s dilemma game. In their
experiment the rate of cooperation in the game was 3% when the sub-
jects knew that the other player had defected, and 16% when they
knew that the other player had cooperated.44 Now one might well
have expected some rate of cooperation between 3% and 16% when
the subjects were uncertain whether the other player had cooperated
or not. However, when the subjects were confronted with this uncer-
tain situation, the rate of cooperation rose significantly to 37%,45 a
number that cannot be explained as some weighted average between
the strategy “cooperate when the other cooperates” and the strategy
“do not cooperate when the other does not.”
Shafir and Tversky attribute this pattern of responses, a pattern
that shows the disjunction effect and violates the sure thing principle,
to the different understandings that a subject will have of her choice
situation depending on whether she knows if the other player has al-
46
ready made his choice of strategy. When she knows that the other
player has already chosen his strategy, whether it be to cooperate or
not to cooperate, the subject thinks of herself as acting “on her own.”47
Given the choice of the other player, she alone will determine the final
outcome of the game. This encourages her to bring a highly indi-
vidualistic perspective to bear on her choice of strategy, a perspective
that leads her more naturally to choose against cooperation. How-
ever, in the disjunctive or uncertain situation, all four possible cells of
the prisoner’s dilemma game are still very much in play, with the out-
come of the game still to be determined by a combination of the
strategy choices of both players taken together. Shafir and Tversky
argue that this provides for a more collective understanding of the
situation, and from this more collective point of view the optimal
3 Shafir & Tversky, supra note 36, at 452-59.
Id. at 454-55. The higher rate of cooperation in the latter situation provides
some support for the idea that players will sometimes reciprocate cooperation from
the other player rather than always choose the dominant noncooperation strategy that
rational choice theory typically prescribes.
4 Id. at 455.
Id. at 457.
47 Id.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1188 UN!VERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151:1169
strategy for both parties is to cooperate.48 Thus, it is less surprising
that cooperation is chosen more frequently in this situation, say Shafir
and Tversky, the sure thing principle notwithstanding.49
It is worth emphasizing, again, that these differences in under-
standing that the subject brings to her choice situation cannot be ex-
plained on the basis of some variation in the properties or values of
any one (or more) of the alternative outcomes considered each on
their own. Nor are they to be explained by variations in properties or
values of the outcomes as weighted by the probability of their occur-
rence. Rather, the differences in understanding arise because the
same alternatives are somehow “differently conceived” depending on
what else is available, something that goes to a relationship that exists
between alternatives in the set of possibilities. For example, in the dis-
junctive case of the prisoner’s dilemma, the presence of all four cells
together as possible outcomes of the game helps to frame any one of
them as instances of the “still-to-be-collectively-determined” category.
On the other hand, when only two cells within a given column of the
game are available (indicating that the other player has already cho-
sen his strategy), the two possible outcomes become instances of the
“it’s-all-up-to-me” category. The idea that there could be a relation-
ship between alternative outcomes, related as instances of a particu-
larly conceived category of possibilities, and that this could influence an
individual’s choice of a strategy producing those outcomes, is an idea
that the independence properties in the sure thing principle are set to
50
deny.
48 Id.
Id. at 457-58. Someone might object that there is a better explanation for the
subjects’ tendency to choose “less rationally” (in the sense that rational choice would
require that the subject choose the dominant noncooperative strategy) in the case
where the four possible cells of the prisoner’s dilemma are still in play. The argument
might be that choosing over four cells is a more complex choice than choosing over
two cells, and that it should not be surprising that subjects do less well, or less “ration-
ally,” in the more complex case. However, Rachel Croson has experimental results
that show this cannot be the explanation; it appears that the disjunction effect disap-
pears in games that are equally complex but which do not have the same scope for dif-
ferent reasons. See Rachel T.A. Croson, The Disjunction Effect and Reason-Based Choice in
Games, 80 ORG’L BEHAV. & HUM. DECISION PROCESSES 118, 129-31 (1999) (testing an
alternative theory of decision making, complexity, as an explanation of the disjunction
effect). Thus, Croson concludes that the explanation for the effect is reason-based,
not complexity-based. Id. at 131.
The famous Allais paradox can be thought of in the same way. Maurice Allais,
The Foundations of a Positive Theory of Choice Involving Risk and a Criticism of the Postulates
and Axioms of the American School, in EXPECTED UTILITY HYPOTHESES AND THE ALLAIS
PARADOX 27 (Maurice Allais & Ole Hagen eds., 1979). Allais argued that agents often
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1189
III. REASON-BASED CHOICE AND THE LAW
The common law is more than a mere collection of authoritative
resolutions for disputes. In addition to providing a decision, common
law judges are typically expected to articulate a set of publicly com-
prehensible reasons in support of the decision. Indeed, the obliga-
tion to provide reasons for one’s choices may well be the one thing
that effectively distinguishes the common law as a method of collective
decision making. In market or political forms of decision making, for
example, individual rights holders and legislators can make perfectly
authoritative decisions without good reason.” But, in the common
law, the reasons that are publicly articulated in support of some deci-
sion form a large part of the authoritative basis for it. Weak reasoning
will undermine the authority of a case and leave it exposed to the in-
dignity of being distinguished into oblivion, if not completely over-
turned or reversed.
This suggests that common law adjudication is very self-
consciously a form of reason-based choice. It remains to be seen,
however, whether we can find the same tension in law between reason-
based choice and value-based choice that we saw above in the behav-
iorists’ experiments. In this Part, I will argue that the same tension
does exist, and that it is often properly resolved in favor of legal rea-
sons that, for groups of judges, can be sensibly organized under par-
ticular categories of understanding, the sort of thing that explained
the behavior of the experimental subjects discussed earlier. This ra-
tional ordering of particular decision making is something I call cate-
gorical reason.
will have a “certainty preference” that cannot be allowed for if the strong independ-
ence condition, or sure thing principle, of expected utility theory is assumed. The cer-
tainty of getting some given payoff in every possible state of nature is not a property of
any one state of nature. Rather, it is a property of, or a property of how we think of, all
the (mutually exclusive) alternative states taken as a whole, that is, a property of their
relationship. Moreover, because the alternative states are mutually exclusive possibili-
ties, their relationship cannot be causal, only conceptual.
I do not mean to suggest that the passage of legislation is never informed by
reasoned debate. Indeed, for the “republican” theory of politics, this sort of delibera-
tive exchange is often thought to be central and important for grounding the authority
of political decision making in general. I only mean to argue that the authority of any
given legislative act is, ultimately, a matter of its proper positing by some legitimate
lawmaker (e.g., a majority of the legislators). It is not affected by the possibility that,
after reasoned debate, the particular reasons that prevailed were not especially good.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1190 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
To see the connection between law and reason-based choice as
understood by the behaviorists, consider the following example:”
Suppose that a plaintiff was injured while using some product and that
she has advanced two separate and independent claims for the recov-
ery of damages from the defendant manufacturer. The plaintiff has
argued that the product was either defectively manufactured (D) or
sold with an inadequate warning (W), where either of these two argu-
ments would be sufficient, if successful, to win a judgment (J) for the
damages in question. In symbols, (D or W)–J. Now suppose that a
panel of three judges has considered the various arguments and that
each judge has come out on the issues in the following way. Judge A
The example is an instance, in disjunctive form, of what Lewis Kornhauser and
Lawrence Sager originally called “the doctrinal paradox.” Lewis A. Kornhauser & Law-
rence G. Sager, The One and the Many: Adjudication in Collegial Courts, 81 CAL. L. REV. 1,
3, 57 (1993); see also Lewis A. Kornhauser & Lawrence G. Sager, Unpacking the Court, 96
YALE L.J. 82, 114-15 (1986) (illustrating the problem of the doctrinal paradox); Lewis
A. Kornhauser & Lawrence G. Sager, Group Choice in Paradoxical Cases 2 (Nov. 26,
2001) [hereinafter Kornhauser & Sager, Group Choice] (unpublished manuscript, on
file with author) (defining paradoxical cases as those “where a group’s consensus on
underlying premises diverges from that group’s consensus on the ultimate outcome”).
For other discussions of the doctrinal paradox, including analyses that link it to more
general problems in the theory of social choice and the theory of games, see Geoffrey
Brennan, Collective Coherence?, 21 INT’L REV. L. & ECON. 197, 200-01 (2001) (examining
the doctrinal paradox and illustrating how justice may conflict with doctrine); Bruce
Chapman, More Easily Done Than Said: Rules, Reasons and Rational Social Choice, 18
OXFORD J. LEGAL STUD. 293, 312-18 (1998) [hereinafter Chapman, More Easily Done]
(analyzing reasons, legal issues, and structure-based equilibrium); Bruce Chapman,
Rational Aggregation, 1 POL., PHIL. & ECON. 337, 341-50 (2002) [hereinafter Chapman,
RationalAggregation] (arguing that the aggregation of reasoned individual judgments is less
subject to paradox than the aggregation of individual preferences); Bruce Chapman, Ra-
tionally Transparent Social Interactions, in COGNITION, RATIONALITY, AND INSTITUTIONS 189,
197-98 (Manfred E. Streit et al. eds., 2000) (using a legal example to illustrate the point
that some strategic interactions are transparent simply because they are more rationally
comprehensible); Christian List & Philip Pettit, Aggregating Sets ofJudgments: An Impossi-
bility Result, 18 ECON. & PHIL. 89, 96-100 (2002) (proving a general impossibility theo-
rem for the aggregation of individual judgments); Christian List & Philip Pettit, Aggre-
gating Sets of Judgments: Two Impossibility Results Compared, SYNTHESE (forthcoming
2003) (manuscript at 10-12, on file with author) (comparing Arrow’s impossibility
theorem with their own impossibility result), available at http://socpol.anu.edu.au/
pdf-files/W20 ; Philip Pettit, Deliberative Democracy and the Discursive Dilemma, in
SOCIAL, POLITICAL, AND LEGAL PHILOSOPHY 268, 274-76 (Philosophical Issues vol. 11,
Ernest Sosa & Enrique Villanueva eds., 2001); Philip Pettit, Groups with Minds of
Their Own 1-4 (Apr. 2001) (unpublished manuscript, on file with author) (discussing
the doctrinal paradox in the context of purposive groups more generally), available at
http://socpol.anu.edu.au/pdf-files/W12 . Saul Levmore used the specific example
in the text, involving two different causes of action, to make a quite different point
about the conjunction of probabilities. See Saul Levmore, Conjunction and Aggregation,
99 MICH. L. REv. 723, 729 n. 1 (2001) (identifying a “reverse conjunction” problem in
the context of the product rule combining independent probabilities).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1191
believes that there is no reason for thinking that the product was
defectively manufactured. Thus, she believes argument D to be false.
But she thinks argument W is true; in other words, she thinks that, al-
though the product has not been defectively manufactured, there is
some risk and the consumer has been inadequately warned. She
would find reason forJ on the basis of W Judge B, on the other hand,
believes that, while adequate information has been provided and,
therefore, that argument W is false, the product has, nevertheless,
been defectively manufactured and, therefore, that argument D is
true. Thus, she too would find in favor of J. albeit for a reason differ-
ent fromJudge A. Finally,Judge C rejects both arguments D and Was
false and, therefore, also rejects J as false. The views of the three dif-
ferentjudges are presented in summary form
in Table 1.
Table 1
1. There should 2. There should b3 There rson
beJ for reason D. beJ for reason W. D or W
Judge A False True True
Judge B True False True
Judge C False False False
Majority False False True
Where:
J is the proposition, “The plaintiff wins ajudgment for damages.”
D is the proposition, “The product was defectively manufactured.”
W is the proposition, “There was an inadequate warning of product
risks.”
And: (DorW)-JJ
It should be apparent that, at the group level, there is something
odd going on here. A majority of the court rejects proposition 1,
“There should be J for reason D,” as false. Further, a majority of the
court also rejects proposition 2, “There should be J for reason W,” as
false. Yet, a majority of the court accepts the disjunctive proposition
3, “There should be J for reason D or W,” as true. There is something
collectively irrational in this combination of votes. At the group level,
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1192 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
it looks like we have generated a disjunction effect similar to what we
observed earlier in the behaviorist experiments.
The collective irrationality would become particularly apparent if
the majority of judges who support an outcome favoring the plaintiff,
A and B, had to articulate a set of publicly comprehensible reasons in
support of their decision. There are, after all, only two possible ar-
guments that the plaintiff can make in this case to win a reward for
damages J. Yet, on each of these essential arguments, the two judges
who form a majority in favor of the plaintiff have completely opposing
views (as indicated by columns 1 and 2). They would, therefore, have
some difficulty saying together what it is that they want to do together.
The two judges may share a common preference, or predisposition,
for the outcome favoring the plaintiff, but it is not at all clear that they
have a shared understanding of what it is that they are doing to reach
that outcome.53 It is as if what they want to do together lacks any real
categorical identity. Nor is the problem merely that there is a plurality
of reasons here, with no particular reason commanding majority sup-
port. There are majority views on each of the relevant reasons and
they are that neither D nor W is a good reason for the plaintiff to pre-
vail in this action.
Now one might ask whether this lack of any common understand-
ing of, or reason for, a shared preference for a particular legal out-
come must translate into any practical difficulty for the majority actu-
ally to act on its preference, that is, whether the failure to be able to
articulate a common sense of what it is that one is doing should im-
pact at all on the possibility of rational choice.54 In some strictly causal
Of course, each judge has a reason for what she wants to do. To that extent, the
position of each judge is reasoned and not merely a matter of preference or predispo-
sition. But as these reasons are not shared between the judges, what is shared, not be-
ing supported by reason, looks more like a brute preference or predisposition.
5 If the different members of a majority are voting in favor of a given outcome,
but each for different reasons, then in an important sense they are not actually voting
on the same proposition. For example, in Table 1, column 3, while it appears that
Judges A and B agree in their voting, Judge A is actually voting for the proposition ‘j
for reason W,” whereas Judge B is voting for the proposition ‘j for reason D.” This se-
verely complicates the epistemic support that majority voting gives to a given proposi-
tion by way of Condorcet’s famous jury theorem. Condorcet’s theorem shows that if
each member of a group of voters is more likely to be right than wrong about some
given proposition (and this probability of being right is more or less equal for each
voter), then a majority of such voters is even more likely to be right than wrong about
that proposition than is any of the voters alone. MARQUIS DE CONDORCET, ESSAY ON
THE APPLICATION OF MATHEMATICS TO THE THEORY OF DECISION-MAKING (Paris,
L’imprimerie Royale 1785), reprinted in CONDORCET: SELECTED WRITINGS 33, 48-49
(Keith Michael Baker ed., 1976); see Bernard Grofman et al., Thirteen Theorems in Search
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1193
sense, of course, there cannot be any such impact. It is always possible
to pursue one’s preferences without good reasons, and possible for a
majority to pursue its preferences without any coherent reason across
its members. The preferred alternative is not less available as an op-
portunity for choice simply because the majority cannot organize its
thinking in favor of that preferred result under a single (coherent) set
of concepts or categories of thought. Thus, at first glance there is lit-
tle in this legal example, perhaps, that provides any reason for think-
ing that there is some necessary connection between what we can say
or think (together) and what we can do (together). The conventions
of rational choice theory, therefore, which emphasize (as WARP does)
consistency of choice over the consistent availability of unchanging al-
ternatives, seem not to be much affected, although one might have
hoped that the idea that one can pursue one’s preferences without
good reasons would give a rational decision theorist some reason to
pause.
However, the experimental results provided by the behaviorists55
do suggest that subjects will have some difficulty making a choice (or
making the same choice) if they cannot organize what they want to do
under a set of particular (or the same set of particular) reasons. Thus,
this literature supports the idea that there can be a genuine tension
between what we want to do and what we have reason to do, and that
this tension is often resolved, as a matter of fact, in favor of reason.
Further, it seems that our legal practices might provide some
normative support for resolving the tension in this way. For suppose
the plaintiff in the Table 1 example were to argue that she should win
a judgmentJ because there appears to be a majority agreement in fa-
vor of this outcome in column 3. In other words, she points simply
and bluntly to the fact that a majority of the judges thinks that she
should win, albeit for different reasons. A lawyer is tempted to reply, I
think, that the plaintiff’s appeal to the column 3 agreement is inade-
of the Truth, 15 THEORY & DECISION 261, 264-65 (1983) (formalizing and extending
Condorcet’s original jury theorem). The theorem provides some reason for using ma-
jority voting in legal decision making to get at matters that have a correct legal answer,
or are questions of judgment rather than preference, but it is crucial for the applica-
tion of the theorem that the relevant majorities have voted for the same, or at least not
inconsistent, propositions. Thus, in the context of the doctrinal paradox, this means
that the Condorcet theorem may provide more epistemic support for the majority
votes down columns 1 and 2 than it does for the majority vote down column 3. For
further discussion of the relevance of the doctrinal paradox to the application of the
Condorcet theorem, see Chapman, Rational Aggregation, supra note 52, at 341-44.
See supra text accompanying notes 37-49 (discussing the results of Shafir, Simon-
son, and Tversky concerning the idea of “reason-based choice”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1194 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
quate because in law she has an obligation to frame her claim against
the defendant as an argument, that is, under some sort of conceptual
structure. It will not do for the plaintiff to show only that a majority of
the judges believes (in some unstructured way) that the defendant
owes (or, even, probably owes) her damages for some reason. Instead,
she must show that (more probably than not) a majority of the judges
believes that there is a reason-at least one particular reason, here
woven out of some particular account of transactional wrong-for the
claim. It is in this respect that some claims, even if they are right (or
right more probably than not), may not be right for the right reasons.
We might reasonably wonder, therefore, whether they are both right
and rational.
IV. CATEGORICAL REASON AND RATIONAL CHOICE
The argument so far has shown that there is an alternative con-
ception of ideally rational choice, which I have called categorical rea-
son, that requires us to relax some quite fundamental axioms of the
economic theory of rational choice. If agents choose under an under-
standing of what it is that they are doing, and not merely in a goal- or
value-oriented way, then they will choose differently under different
understandings, the requirements of WARP and the sure thing prin-
ciple notwithstanding. I have called this alternative conception of ra-
tional choice categorical reason because it emphasizes the idea that a
relationship between certain alternatives, or how they are understood
together or as a category, can inform choice. In this respect, categori-
cal reason is to be contrasted with axioms like WARP and the sure
thing principle, which emphasize the independent properties and
values of particular alternatives, and insist on choice consistency so
long as these properties, independently considered (either within a
set of certain alternatives or within a lottery of uncertain outcomes),
remain the same.
Further, the research done by the behaviorists has suggested that,
as a matter of fact, this alternative ideal of rational choice does inform
how agents actually choose. Of course, the particular experiments
might still have us wondering how rational it is merely to react (un-
thinkingly) to some of these different understandings of one’s choice
situation. For example, to choose a higher quality pen over some cash
merely because a lesser quality pen has been added to the set of alter-
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1195
56
natives does not seem to be a particularly rational thing to do on any
plausible account of rationality. However, I have argued that different
scenarios can easily be constructed that make perfect sense of this sort
of behavior and, further, that the practice of common law adjudica-
tion, in its traditional emphasis on the obligation to offer a particular
argument in support of one’s claim, idealizes a form of categorical
reason.
In this Part, I argue that there is some advantage in this alternative
conception of rationality, even for what the economic theorist of ra-
tional choice seeks to accomplish. I focus on two areas of rational
choice theory that have confronted systematic difficulties, namely, the
theory of social choice and the theory of games, to suggest how the
arguments might go. In the theory of social choice I suggest that the
idea of categorical reason can bring some conceptual discipline to
bear on the individual preferences that are deemed admissible into
the social choice rule and that certain problems of instability in social
choice can thereby be avoided. I relate this argument to certain for-
mal results in social choice theory dealing with required restrictions
on the domain of admissible preferences. In the theory of games, I
argue that categorical reason allows a player to conceive of the game
she is playing in a way that makes certain strategy choices less accessi-
ble to her. The result, I suggest, is a greater propensity for players to
coordinate their choices in a coordination game and, more controver-
sially, a greater propensity for them to choose cooperatively in the
prisoner’s dilemma game.
A. Categorical Reason and Social Choice
In many situations calling for collective action, it seems likely that
the individual members of a decisive majority will not have reasons in
common for what they most want to do. Yet we will not feel that this is
57
in any way problematic for them. Consider, for example, that most
mundane of economic transactions, the purchase of a car. To give
this problem a collective dimension, imagine that there is a three-
person purchasing consortium and that a majority of the consortium
has voted to buy a white sports car. One member of the majority has
voted this way only because the car is white and the other only because
it is a sports car. Table 2 summarizes this scenario in a way that ap-
56 This choice is discussed supra text accompanying notes 25-26.
5 See, eg., Kornhauser & Sager, Group Choice, supra note 52, at 18-20 (discussing
the different normative premises that might motivate individual members of a group).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1196 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
pears to make it fully analogous to the earlier legal example laid out
in Table 1.
Table 2
1. We should buy 2. We should buy 3. We should buy
this car because it this car because it this car because it
is white. is a sports car. ar.
car.
Individual A True False True
Individual B False True True
Individual C False False False
Majority False False True
The fact that the members A and B of the majority in column 3 do
not have “reasons in common” to support their shared preference for
buying this particular car is not thought to present them with any real
difficulty. Nor is it thought to be rationally compelling that this pur-
chasing consortium chooses not to buy this white sports car simply be-
cause a majority rejects both the idea of buying it because it is white
and the idea of buying it because it is a sports car. The decision to
purchase a car, even (more particularly) a white sports car, is not es-
sentially decomposable into two prior atomic propositions: Is it a
sports car? Is it white? That underlying structure, while possibly a
helpful guide to the purchasing decision, is not an essential part of
the problem in the same way that the plaintiffs claim to damages in
Table 1 needs to be grounded in a particular account of transactional
wrong. Rather, the purchasing consortium is out to purchase a car,
perhaps even the best car that is available to it, all things considered. But
that judgment is ultimately made of the car and on the whole, not on a
criterion-by-criterion (or column-by-column) basis.
Despite this structural difference in the examples, there might be
something useful, even for what the economist seeks to accomplish by
way of social choice or what a purchasing consortium seeks to achieve
in the market for cars, in insisting on the greater rationality require-
ment that is inherent in the legal idea that members of a group can
act sensibly together only if they can organize what they would prefer
to do under a common understanding, that is, only if they can act to-
gether under a common set of categories or concepts. That this is
sometimes difficult to do, and that it sometimes frustrates the
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1197
achievement of shared preferences, might be precisely what is so use-
ful about it.
To illustrate this point, suppose that the three individuals in our
Table 2 purchasing consortium, considering the joint purchase of a
car, originally had preferences over three alternative cars as follows
(where for each individual the alternatives are preferred in order
from top to bottom within each column):
Table 3
Individual A Individual B Individual C
White sports car (WS) Black sports car (BS) Black family car (BF)
Black family car (BF) White sports car (WS) Black sports car (BS)
Black sports car (BS) Black family car (BF) White sports car (WS)
This is, of course, the preference profile that makes for the familiar
majority voting paradox. A majority prefers WS to BF, BF to BS, and
BS to WS. Thus, within the social choice framework, there is the dan-
ger here of a kind of excess of rational doing: for every alternative that
one is tempted to choose, there is another that a majority would pre-
fer to have instead. It is this excess of rational doing that gives rise to
cycling and instability.
Now it is common for economists to point out that the problem
here is that individual preferences are not “single peaked”; there is no
general agreement (1) that all the alternatives are to be assessed ac-
cording to some single decisive dimension, and (2) that one of the al-
ternatives is of intermediate value on that decisive dimension. If only
that were so, the argument goes, then that intermediately placed al-
ternative would never be the worst alternative for any voter and the
majority voting paradox would be avoided 9
For a good introductory discussion of the majority voting paradox, see DENNIS
L. MUELLER, PUBLIC CHOICE II, at 63-66 (1989).
59 Id. at 64-66. If there is this sort of agreement across the voters, and preferences
are single peaked, then the alternative chosen under majority rule will be the one
which possesses that amount of the decisive dimension which is most preferred by the
median voter-that is, the voter who is in the middle of the distribution of voters or-
dered along the decisive dimension according to where their most preferred alterna-
tive is on that decisive dimension. Why this alternative would be chosen, and why no
other alternative would defeat it under majority rule, can easily be appreciated in the
following way: Imagine beginning at the extreme left (or right) of the decisive contin-
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1198 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
This is, in effect, to insist that individuals organize their prefer-
ences in a single-minded way along one decisive dimension and to al-
low them only the limited scope of ordering the social alternatives ac-
cording to how these alternatives vary quantitatively (more or less)
along that decisive dimension.60 But, as the example suggests, and as
1 61
multidimensional models show more generally, individuals react,
reasonably, to a broad range of categorically different dimensions or
aspects of the social alternatives on offer. And so the question arises
whether these different and plural dimensions of a social choice prob-
lem can be rationally organized in some way so that instability can be
avoided.
The car example is suggestive. The majority coalition of A and C
can say together (in support of what they might do together), “Given that
the car is black, we would prefer it to be a family car.” Likewise, the
majority coalition of B and C might be able to say, “Given that it’s a
sports car, we would prefer it to be black.” In this respect, these coali-
tions can make use of what are sometimes referred to as generic prefer-
ences.62 But what would the majority coalition A and B say together?
In some sense, of course, they have a shared preference over a pair of
very particularly described alternatives just like the other majority coa-
litions do. Indeed, as already intimated, that is what gives rise to the
instability. But their shared preference for WS over BF lacks any of
the generic structure that characterizes the shared preferences of the
other two majority coalitions. Thus, it is harder for them to articulate
their shared preference in any sort of categorical way, that is, in a way
that makes use of the generic preferences that are in play in the
choice problem. In this respect they, as a coalition, are rendered
uum. Then any move to the right (or left) will receive the support of a majority of the
voters until we arrive at the median voter’s ideal position on the continuum, at which
point any further move to the right (or left) will be defeated by a majority of the voters.
60 The standard example is the ordering of candidates for political office from
“left” to “right” on the ideological spectrum. Another example might be the different
quantities of some uni-dimensional public good that different voters want to buy at a
given tax price. For a good discussion of both of these examples, see ALLAN FELDMAN,
WELFARE ECONOMICS AND SOCIAL CHOICE THEORY 169-70 (1980).
See Richard D. McKelvey, Intransitivities in Multidimensional Voting Models and
Some Implications for Agenda Control, 12J. ECON. THEORY 472, 472-82 (1976) (detailing
how multidimensional voting can “end up at any other point in the space of alterna-
tives”). For a discussion of the multidimensional case in a legal context, see Chapman,
More Easily Done, supra note 52, at 300-16.
SeeJon Doyle & Richmond H. Thomason, Background to Qualitative Decision The-
ory, Al MAG., Summer 1999, at 55, 58 (defining generic preferences as preferences
among classes).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1199
“speechless,” just like judges A and B were in the Table 1 legal exam-
ple. But now we can see that there may be some stabilizing effect in
using the discipline of a shared (or public) categorical reason to re-
strict the formation of this majority group. After all, without this addi-
tional discipline and structure, there is only a senseless (i.e., noncate-
gorical, nonconceptual) aggregation of (merely particular) pre-
ferences and the cycling problem that this permits.
The discipline that is provided by a (public) categorical reason
can be related more generally to a particular form of “value restric-
tion” (specifically, “not-between value restriction”) that Amartya Sen
63
has shown is sufficient for avoiding the majority voting paradox.
Specifically, if all individuals agree that in any triple a given alternative
is “not between” the other two, that is, is either best or worst of the
three, then the majority voting paradox cannot occur. For instance,
in the car example, it is easy to see that an alternative way to express
what A and C have in common is their view that WS is a “not-between”
alternative for them; the real issue between them is whether the pur-
chased car should be a black car (BE, BS) or not (WS). Individual A
puts the white car alternative (WS) first and the pair of black car al-
ternatives (BE, BS) last, whereas individual C has the reverse ordering
of these two partitions. This is something that they could have de-
cided first, before they went on to decide, if necessary, what was a sec-
ondary issue to them, viz., what type of car a black car should be.6′
Likewise, what the coalition of B and C has in common might have
been expressed as an agreement over BF as a “not-between” alterna-
tive, the sort of agreement that asks each to decide first whether the
car chosen should be a sports car or not and, second, if it should be a
sports car, whether it should be black or white. But, again, the pair of
individuals A and B would have some difficulty articulating its own ver-
sion of a common understanding of the relevant issues in this way.
They agree between them that BS is a “not-between” alternative, but
Amartya K. Sen, A Possibility Theorem on Majority Decisions, 34 ECONOMETRICA
491, 492-95 (1966).
64 It could be, of course, that A feels there is a great deal more at stake in the
choice of car type than the choice of color, viz., that the preferential distance between
BFand BS is large compared to the preferential distance between WS and BE But this
cannot be a view that she has in common with C. For C the preferential distance be-
tween BF and BS is contained within the distance between BF and WS. So the search for
a shared (or public) categorical reason for choice, at least one linking A and C, cannot be
found here. As the text following this note suggests, this suggested interpretation (that
car type is a more important issue than car color) is better for the pair of voters B and
C.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1200 UNIVERSITY’OFPFNNSYLVANIA LAWREVIEW [Vol. 151: 1169
what, exactly, is the category or concept that embraces the partition
(WS, BE) of alternatives that is the complement to that not-between
alternative? The problem, again, is that it is hard to “make sense” of
such a partition of the alternatives in terms of the categories or con-
cepts (color and type of car) that are in play in the example. We
might say, as we can for all the other possible pairs of individuals, that
individuals A and B agree at the level of preferences, but that they do
not share any sort of categorical agreement about the sorts of issues
that inform their choice and the order in which these issues might be
considered.
Now one might object that the imposition of a categorical disci-
pline on preferences still leaves too much unresolved to be helpful.
After all, even those two groups of voters, AC and BC, which (unlike
group AB) agree that the salient issues are the type and color of the
car to be purchased, disagree fundamentally on the order in which
these two issues should be addressed. For AC the most salient issue is
color; only after considering that issue would this group turn its atten-
tion, if necessary, to what type of car it should be. But for the coali-
tion BC the most important issue is type of car, and only if a sports car
is chosen would the coalition turn its attention to the issue of color.
Moreover, the order in which the issues are considered is likely to af-
fect the outcome; in this respect the matter is analogous to the prob-
lem of path-dependent choice . For example, if color is considered
first, then it seems less likely that BS will end up being chosen. Indi-
65 The choice of an alternative is path-dependent if the probability of that alterna-
tive being chosen varies with the order in which it is presented for consideration as
compared to other alternatives. In social choice theory, the conventional view is to
think of path dependence as a kind of arbitrariness to be avoided; alternatives should
be chosen, the argument goes, according to the value of their intrinsic properties.
Kenneth Arrow, for example, defended his use of a collective rationality condition in
social choice on the ground that collective rationality, in the form of a fully transitive
social preference relation, would guarantee path independence. KENNETH J. ARROW,
SOCIAL CHOICE AND INDIVIDUAL VALUES 120 (2d ed. 1963); see also Charles R. Plott,
Path Independence, Rationality, and Social Choice, 41 ECONOMETRICA 1075, 1075-91 (1973)
(discussing the relationship between path dependence and collective rationality).
However, not all path dependence should be construed as arbitrary path dependence;
some choice sequences, or paths, might make “more sense” than others and will often
matter a great deal to a process theorist. See Bruce Chapman, Individual Rights and Col-
lective Rationality: Some Implications for Economic Analysis of Law, 10 HOFSTRA L. REv. 455,
466-70 (1982) (reconciling path dependence with deontological thought); Chapman,
More Easily Done, supra note 52, at 303 (arguing that “some legal choice paths or proc-
esses are so permeated by thought,” and so conceptually privileged, that they make
choosing on alternative choice paths almost “unthinkable”); Bruce Chapman, Rights as
Constraints: Nozick Versus Sen, 15 THEORY & DECISION 1, 2-8 (1983) (discussing the im-
plications of rightful choice partitions for collective rationality).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1201
vidual C will vote in favor of black cars, and individual A against black
cars, in the first round. Whether black cars are chosen categorically in
that round depends a good deal on how individual B, whose prefer-
ences are not categorical in this way because they do not satisfy not-
between value restriction on alternative WS, actually votes. But, in the
event of a first round vote for a black car, it does seem likely that BF
will defeat BS in the vote on the issue of type of car. An analogous ar-
gument would suggest that WS is the less likely choice if the issue of
type of car is decided first.
However, in some contexts, there is good reason to think that this
sort of path dependence will be less of a problem for categorically
sensitive choice.66 This is because the categories or concepts that
make sense of certain partitions of the alternatives for choice will often
make sense of certain paths (or sequences of those partitions) as well,
at least if we want to continue to make use of the stabilizing effects of
not-between value restriction. To see this, consider the example of a
criminal trial, where the two issues to be decided are the verdict and
the sentence for the accused. Again, one could imagine a panel of
judges considering three possible final outcomes-innocent (I), guilty
with a severe sentence (GS), and guilty with a lenient sentence (GL).
And again, a natural partition of the alternative outcomes might be
into the two issues, verdict (generating the choice “(I) or (GS, GL)?”)
and sentence (generating the choice “GS or GL?”), a partitioning that
would “make sense” in a way that the alternative partitions, “(GS) or
(I, GL)?” or “(GL) or (I, GS)?,” would not. (What single concept,
category, or issue sensibly comprehends the partition (I, GS), for ex-
ample?) But, still, it seems that one could take these two issues, and
the partitions to which they lend sense, in order of either “sentence
first, verdict afterwards” or “verdict first, sentence afterwards.” The
law adopts the second of the two possibilities (and the Queen at Al-
ice’s trial in Wonderland adopts the first),6′ but is there any reason to
do so? One answer, of course, is simple economy: why bother attend-
ing to the issue of sentencing until we know that the verdict decision
makes it necessary? But the analysis provided here suggests a different
answer. While both sequences respect the partition of the alternatives
that makes the most sense, only the path that has us consider the ver-
See Chapman, supra note 27, at 1507 (arguing that law exemplifies a “categorical
application of plural considerations to decisionmaking” by using a “process of adjudi-
cation as sequenced argument”).
LEWIS CARROLL, ALICE’S ADVENTURES IN WONDERLAND 187 (Univ. Microfilms
1966) (1865).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1202 UNIVERSITY OFPENNSYLVANIA LAWREIVIEW [Vol. 151: 1169
dict first, or the one forcing the initial choice to be over the partitions
“(I) or (GS, GL) ,” imposes any sort of not-between value restriction on
the panel of judges. Under the verdict first sequence, each judge
must order her preferences around the salient legal categories, decid-
ing whether to put the alternative I either better or worse than (but
not between) the alternatives GS or GL. The sentence first sequence,
on the other hand, while paying a kind of lip service to the same set of
issues, does not require any of the judges to order her preferences
around those issues. For example, a judge who preferred the three
alternatives in the order GL first, then I, and then GS, that is, someone
who might be saying, “Whether or not I would find him guilty of the
offense depends on the sentence he would receive,” would have no
difficulty voting these preferences under the sentence first procedure
even though these preferences do not seem to show a categorical
commitment to the issues that are salient in the case. The verdict first
sequence, on the other hand, does force this judge to ask a more cate-
gorical sort of question about the verdict, that is, to show the same
sort of commitment to the issues in the case as does the law she per-
68
sonifies. Furthermore, under a verdict first procedure, we not only
make sense of the issues in the case, but we also impose a domain re-
striction on the preferences that legal decision makers can bring to
bear on legal decisions so that certain problems of instability are
avoided.
The burden of this Section has been to show that categorical rea-
son can provide a useful sort of conceptual discipline on the kinds of
shared individual preferences that should be decisive in social choice.
Certain individuals, who want to do something together, might find
that it is more difficult to act as a decisive coalition in favor of their
shared preferences if they are obliged to think, and talk, about exactly
Requiring this sort of structure can, of course, tempt the judge to “nullify” a
possible guilty verdict for fear of risking the worst (for her) possible sentencing out-
come GS. Verdict nullification has attracted a good deal of critical comment, particu-
larly in the United States where, in jury trials, there is the possibility of the death pen-
alty. Juries are said to be charged with the responsibility of reaching a verdict within
the law as explained by the trial judge; it is the task of the judge to determine the sen-
tence. For members of the jury to worry about the sentence rather than the verdict,
particularly if they think the accused has committed the offense in question, is thought
by some to violate the rule of law. Whatever the merits of verdict nullification by ju-
ries, my analysis here, based on the stabilizing impact of imposing the categorical con-
straints of not-between value restriction, offers an independent reason for supporting
the verdict first procedure. See generally Darryl K Brown, Jury Nullification Within the
Rule of Law, 81 MINN. L. REv. 1149, 1155 (1997) (examiningjury nullification and sug-
gesting thatjury nullification can “occur within the rule of law, rather than subvert it”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CA TEGORJCAL REASON 1203
what it is that they are doing. To the extent that the problems of in-
stability that arise in social choice are explained in large part because
too many decisive coalitions can form too easily, the conceptual disci-
pline that categorical reason provides in this respect could be very
helpful. I have also tried to relate the idea of categorical reason to
some well-known results in the theory of social choice that impose re-
strictions on the domain of individual preferences that can be admit-
ted into social choice if instability is to be avoided. While the latter
results are not in any way new for the economic theory of social
choice, it is novel to motivate these results in, and connect them to,
the more philosophical idea of categorical reason. I now will suggest
that categorical reason can also have a beneficial impact on the possi-
bility of coordination and cooperation in the theory of games.
B. Categorical Reason in Noncooperative Games
Consider the simple two person pure coordination game called
“Heads and Tails.”69 Each person, without consulting the other, must
turn up either “Heads” or “Tails” on her own coin. If each person
turns up “Tails”-a match-then each will win five dollars from the
pot. However, if each turns up “Heads”-another match-then each
will win ten dollars from the pot. In the absence of a match, each wins
nothing. What should each person do? What is the rational thing to
do?
Note that there is no conflict of interest in this game. The two
players will receive identical payoffs in all four possible outcomes and,
therefore, order these four outcomes in an identical way.” Specifi-
cally, they both agree that the outcome generated by each of them
playing “Heads” is best, that the outcome wherein each plays “Tails” is
This game is discussed in Sugden, supra note 32, at 775.
70 Thus, this could be a game in which all the players are act utilitarians: each
seeks to act in such a way that total welfare for her society is maximized, but must do so
without the benefit of prior consultation about what she should do to achieve that
shared goal. Such a group has a coordination problem (the inability to communicate
is what makes the game “noncooperative”), even though there is an identity of inter-
ests across the players. See D.H. HODGSON, CONSEQUENCES OF UTILITARIANISM: A
STUDY IN NORMATIVE ETHICS AND LEGAL THEORY 58-62 (1967) (illustrating that even
correct application of act utilitarianism would not necessarily have better conse-
quences, and would possibly have worse consequences, than would acceptance of spe-
cific conventional moral rules and personal rules); DONALD REGAN, UTILITARIANISM
AND CO-OPERATION 66 (1980) (analyzing the problem of coordination between act
utilitarians). For a good introductory discussion showing how rational choice theory
implies counterintuitive results in the analysis of coordination games, see Sugden, su-
pra note 32, at 774-78.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1204 UNITVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
second best, and that the two non-matching outcomes, “Heads-Tails”
and “Tails-Heads,” are tied for worst. One might have thought that
this would make the choice of actions relatively easy: each player
would choose that action, “Heads,” which so clearly, and without con-
flict, makes both players better off.
Surprisingly, however, the choice of this action is less obvious for a
player deemed to be rational in the way that rationality is understood
within game theory. This is because rational play for any one player
depends crucially on what that player thinks the other player will do
in the game. It is simply false, the argument goes, to think that one
should always turn up “Heads.” If the other player turns up “Heads,”
then, but only then, should the first player match with “Heads” her-
self. Otherwise, the first player should turn up “Tails” and secure the
second best of the matching outcomes. The problem, of course, is
that both players are thinking through this same problem of strategic
choice at the same time (or, at least, without prior consultation or
revelation of their choices), and so neither can really condition her
choice on the given choice of the other. Moreover, that each player is
rational in this way is typically assumed to be common knowledge in
the game. Thus, each player knows that the other is likewise at-
tempting to work out this conditional strategy which conditions on a
strategy that is itself conditional on the strategy of the first. The result
is an infinite (self-referential) regress that has the effect of leaving
each player in a kind of strategic limbo, unsure about what to do.
Nor do the difficulties disappear if we allow the individual player
to develop a strategy that appears to recognize and confront this proW
lem as one of uncertainty. The mixed (or probabilistic) strategy that
survives the common knowledge assumption requires that each player
play “Heads” with a 1/3 probability and “Tails” with a 2/3 probabil-
ity.72 However, while this allows both players simultaneously to step
Common knowledge is information which is known to all the players in a game,
which each player knows the others know, which each knows the others know that she
knows, and so on. Common knowledge of rationality (and of the rules and payoffs of
the game) is typically crucial for solving games because it allows players to put them-
selves in the place of other players, to replicate their reasoning (that is, think through
what they will rationally do in their situation), and act accordingly. For a discussion of
the importance of the common knowledge assumption to game theory, see CRISTINA
BICCHIERI, RATIONALITYAND COORDINATION 39-43 (1993); Chapman, supra note 30, at
443-45.
7 This is the Nash equilibrium mixed strategy. “A Nash equilibrium is an array of
strategies, one for each plaver, such that no player has an incentive (in terms of im-
proving his own payoff) to deviate from his part of the strategy array.” DAVID M.
KREPS, GAME THEORY AND ECONOMIC MODELLING 28 (1990). Any other assignment of
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CA TEGORJCAL REASON 1205
out of the strategic limbo in which they originally found themselves,
the consequence is hardly comforting. Now, under this choice of a
mixed strategy by each player, the most preferred outcome where
each player matches on “Heads” arises with only a 1/9 probability (the
product of each player independently playing heads with a 1/3 prob-
ability). One might have hoped that rational choice would do better
than that.
The economist Michael Bacharach has characterized the sort of
thinking that generates this difficulty as thinking in an “I/he” frame.73
The “I/he” frame accommodates the idea, central to game theory and
Nash-like thinking, that a player should ask what strategy is best for
herself given what the other player might do, and allows that player,
again under common knowledge of such reasoning, to replicate that
probabilities between the two choices, “Heads” and “Tails,” would not be stable as each
of the two players tested out its rationality under common knowledge, an assumption
that allows each to replicate the reasoning of the other and then make corresponding
adjustments in a proposed strategy choice.
For an explanation of how to derive (and interpret) a mixed strategy, see ERIC
RASMUSEN, GAMES AND INFORMATION: AN INTRODUCTION TO GAME THEORY 69-73
(1989). Mixed strategies require an odd interpretation. The idea is to calculate the
probability distribution over one’s own possible actions such that the other player will
be indifferent between which strategy she chooses. Thus, in the game “Heads and
Tails,” if player 1 chooses to play “Heads” with a 1/3 probability, the expected payoff
for player 2 in playing “Heads” is equal to the expected payoff in playing “Tails,”
namely, 10/3. But given her indifference between playing either of the two pure
strategies (i.e., either “Heads” or “Tails” with certainty) in such circumstances, she
should also be indifferent between playing either of those pure strategies and any
mixed strategy which combines them probabilistically, including her Nash equilibrium
strategy that has her playing “Heads” with a 1/3 probability. Thus, we can say, given
that player 1 plays her Nash equilibrium strategy of “Heads with a 1/3 probability,”
player 2 has no incentive to deviate from her Nash equilibrium strategy of “Heads with
a 1/3 probability” since she does no better for herself by so deviating. (The fact that
she also does no worse is a problem for the theory in that it is essentially an equilibrium
theory rather than a theory for how to play the game ab initio; why, of all those strate-
gies over which she is indifferent, does she feel any compulsion to play the Nash equi-
librium strategy in particular? For an indication of how one game theorist handles this
problem, see Robert J. Aumann, Correlated Equilibrium as an Expression of Bayesian Ra-
tionality, 55 ECONOMETRICA 1 (1987). Aumann describes a correlated equilibrium ap-
proach that does away with the dichotomy usually perceived between the “Bayesian”
and the “game-theoretic” view of the world by synthesizing the two viewpoints and con-
sequently not requiring explicit randomization on the part of the players. Id. at 1.)
And we can also say all this of player 1 if player 2 chooses to play (her Nash equilib-
rium strategy) “Heads with a 1/3 probability.” Thus, the playing of “Heads” with a 1/3
probability by each player is a Nash equilibrium for the game since no player, given the
strategy choice by the other player, can improve her own payoff by adopting an alter-
native strategy.
Michael Bacharach, “We” Equilibria: A Variable Frame Theory of Cooperation
5 June 24, 1997) (unpublished manuscript, on file with author).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1206 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
same sort of thinking in the other player as well. The other sort of
thinking that Bacharach identifies is thinking in a “we” frame.74 The
“we” frame encourages each player to think about the profile S of
strategies (one for each player) that should be adopted by the players
as a group and then identifies the rational strategy for each player as
the one that simply (categorically, nonconditionally) has that player
“doing her part” S, within that overall profile.75 Unlike in the “I/he”
frame, a player in the “we” frame does not have to consider whether
the other players are themselves doing their parts as components of
this profile of strategies in order to justify her strategy choice. Rather,
in response to any question about why she was doing what she was do-
ing, she would only say, “This is simply what we do when we do S (as
best) ,” or, perhaps (to emphasize how the collective understanding
orders the particularity of her individual choice), “This is simply what
I do when we do S (as best),” or even, most provocatively (because
most categorical in tone), “This is simply what it is for us, you and me,
to do S (as best).”
It should be apparent that Bacharach’s “we” frame is closely akin
to the collective understanding that Shafir and Tversky propose as an
explanation for the disjunction effect that they observed in the play of
the prisoner’s dilemma game.76 It will be recalled that there was a
greater propensity for an experimental subject to cooperate when the
strategy choice of the other player in the game was still uncertain. In
that situation, the outcome of the game still had to be collectively de-
termined by the strategy choices of both players, something that put
each player in a more collective (and, it seems, a more cooperative)
frame of mind. On the other hand, when the strategy of the other
player is given, be it to cooperate or not, then the game becomes one
in which the one remaining player chooses to determine the outcome
of the game, something that provides for a more individualistic frame
of mind. This experiment essentially reproduces the “we” frame and
74 Id.
Robert Sugden’s notion of “team reasoning” has a similar structure. See Robert
Sugden, Team Preferences, 16 ECON. & PHIL. 175, 176 (2000) (arguing that “the theory
of choice should allow ‘teams’ of individuals to be decision-making agents and should
allow such teams to have preferences”); Robert Sugden, Thinking as a Team: Towards
an Explanation of Nonselfish Behavior, 10 SOC. PHIL. & PUB. POL’Y 69, 89 (1993) (assert-
ing that “reasons for cooperating do indeed exist, but that these reasons can get a grip
only if we conceive of ourselves as members of a team”).
See supra text accompanying notes 37-45 (noting that the disjunction effect that
occurs depends on whether a player knows if the other player has already chosen her
strategy).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1207
the “I/he” frame for the subjects, and provides some empirical sup-
port for Bacharach’s dualistic account of thinking.
It is important to emphasize that what Bacharach’s account pro-
vides for, and what the Shafir-Tversky experiments support, is the idea
of categorical reason, or the notion that it is a different understanding
that informs choice under the “we” frame. It would be a mistake, for
example, to think that the “we” frame only introduces a different,
more collective, sort of motivation, one that merely identifies “doing
one’s part” with the (more conditional, less categorical) idea that “I
will cooperate if she does.” To begin, the latter idea is not consistent
with what the Shafir-Tversky experiments show; the subjects tended to
be noncooperative almost as frequently when the other player was
known to be cooperating as when the other player was known to be
not cooperating. Second, this sort of conditional cooperation would
do nothing to get the players out of the strategic limbo of the pure
coordination game, a limbo that arises precisely because of an infinite
regress of mutually conditioning conditionals. Third, the idea of do-
ing one’s part as (merely) conditional cooperation would have no im-
pact on the play of the prisoner’s dilemma game, where, in game the-
ory at least, the player has a dominant (not a conditional) strategy not
to cooperate regardless of what the other player does. Rather, what the
Bacharach account provides, and what the experiments support, is an
idea powerful and categorical enough to take us beyond the problem-
atic regress of the pure coordination game and as far as thinking, at
least presumptively,77 that what the other player does, and what one
should do given what the other player does, is not even the right way
to think about strategic choice. The last thought undermines domi-
nance thinking in the prisoner’s dilemma as much as it circumvents
the infinite regress of the pure coordination game.
However, now the worry might be that we have ended up with an
account of rational cooperation that is too unconditional, that is, one
that is implausible precisely because it ignores what the other player
might be doing. Indeed, this is what might be suggested by the very
word categorical, and the somewhat Kantian overtones in the phrase
categorical reason. The Kantian, it is often said, cooperates absolutely,
or just because it is right, and without regard to the contingencies of
what others might choose to do. But, however Kantian that might be,
This is an important qualification, already hinted at supra note 7. The idea is to
allow coordination and cooperation to get started, not to commit to either absolutely.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1208 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151: 1169
it is a mistaken understanding of what is meant here by categorical in
the phrase “categorical reason.”
78
Consider again the earlier etiquette example. That example in-
volved categorical reason because the issue of etiquette was only rele-
vant for those partitions or sets that included the big and small apples
as alternatives for choice. For the other partitions or sets, etiquette
was not relevant at all. In that limited partition-dependent sense, a
sense problematic for WARP, the concern for etiquette was categori-
cal. But it would be a mistake to think that the concern for etiquette
was absolute as compared to, say, the hedonistic interest in eating
larger pieces of fruit. It might be, therefore, that the difference in size
between the large and small apple could become so large that the he-
donistic interest in larger pieces of fruit would overwhelm (even
rightly) the concern for etiquette. In such a situation, with the parti-
tion-dependent effect overwhelmed, there would indeed be transitivity
of preference and no violation of WARP. However, the point of the
example was not to argue that transitivity or WARP never obtain, but
only to suggest that these properties need not always obtain in the way
that rational choice theory suggests. Thus, in this respect, the ex-
treme or uncompromising view is the one offered by rational choice
theory, not the one offered by the theory of categorical reason.
And the same could be said for the theory of cooperation based
on categorical reason. An agent acting under a collective understand-
ing or “we” frame might begin presumptively and categorically with the
thought that she should “do her part” in S because that is what it is for
us, you and me, to do S (as best). But the agent need not think of
herself as absolutely committed to cooperation under strategy S. If too
few others do their part, for example, there may be no “whole” of
which one’s own individual choice can sensibly be construed or un-
derstood as a part. This may call for a rational revision of what it is
that one is doing and allow, therefore, for the possibility of not coop-
erating if others are not cooperating as well.79 However, this should
Supra text accompanying notes 27-28.
For a discussion of conditional or presumptive cooperation as a kind of “revis-
able rationality,” see Chapman, supra note 30, at 472-76. I have argued elsewhere,
Bruce Chapman, Rational Voluntarism and the Charitable Sector, in BETWEEN STATE AND
MARKET: ESSAYS ON CHARITIES LAW AND POLICY IN CANADA 127 (Jim Phillips et al. eds.,
2001), that this account of presumptive cooperation provides a better explanation of
voluntary contributions to public goods, such as in the relief of poverty through chari-
table contributions, than does the theory of rational choice (never cooperate) or Kan-
tian obligation (always cooperate). Moreover, the presumptive cooperation account
can make more sense of the tax treatment of charitable contributions, and the empiri-
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
2003] CATEGORICAL REASON 1209
not be thought of as resurrecting the idea of a purely conditional co-
operation. A purely conditional cooperation is still too much in the
“I/he” frame, and makes no sense at all of a (prior, albeit only pre-
sumptive) collective understanding of one’s action. What categorical
reason rationally requires, therefore, is a defeasible presumption in
favor of cooperation, not an absolute (and thoughtless) commitment
to it.
CONCLUSION
There is a kind of “incompleteness” in the idea of categorical rea-
son that should freely be admitted. An agent who only sees, or under-
stands, alternatives for choice under the aspect of more general con-
cepts, or categories of thought, does not, perhaps, fully appreciate
these alternatives in all their particularity. Any given categorization
need not be crude, of course, but short of reproducing a range of
categories that is as detailed and fine as the particular alternatives it
seeks to organize, it seems inevitable that something must be lost if
choice is to be ordered by categorical reason.80
In rational choice theory, by contrast, the fully rational agent can
compare all possible alternatives for choice. It is true that agents in
the actual world are not thought to be fully rational in this way, but
that is the ideal. Thus, when we say of someone, “She bought the
Volvo because she likes durable cars,” in rational choice theory we
mean to concede that she probably approached the problem of
choice as best she could, but also that, ideally, she would not have lim-
ited herself by these broad generalizations and would have compared
(in detail) all the possible alternatives that she might have chosen.
Rough categorizations and broad rules of thumb are only needed be-
cause an agent must make her way through what would otherwise be
an “incomprehensibly large number of alternatives, most of which
represent unimportant variations on each other.” 8′
cal evidence on how individuals respond to these different tax incentives, than can ra-
tional choice or Kantian theory. See id. at 130 (outlining the complex motivational
structure of homo socioeconomicus).
The incompleteness of choice ordered by categorical reason can easily be ap-
preciated if one reconsiders the criminal trial example, supra text accompanying notes
67-68. While all three alternatives discussed there are available as a final choice, the
“verdict first, sentence afterwards” choice process (which makes outcome I the “not-
between” alternative in the triple) does not permit a (sensible) pairwise comparison
(or choice) between alternative I and either alternative GS or alternative GL. This
noncomparability of certain pairs of alternatives violates completeness.
Doyle & Thomason, supra note 62, at 61.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
1210 UNIVERSITY OFPENNSYLVANIA LAWREVIEW [Vol. 151:1 69
For the rational choice theorist, therefore, something of rational-
ity is lost as we move from a fully particular comparison of all possible
alternatives to a comparison constrained by categorization. But there
must be something gained for rationality as well. For what makes a
comparison of all possible alternatives in their full particularity “in-
comprehensible” is not merely that the set of alternatives is “large.” It
is also that, without some such categorization, the particularity of
choice would literally be “unthinkable.” We think through to the par-
ticulars of our world, after all, only under the aspect of more general
concepts or categories of thought.
So we should not be surprised that there is a notion of ideal ra-
tionality that competes with the full rationality of rational choice the-
ory and pulls us in an opposite direction, that is, from particular to
general rather than from general to particular. However, I hope to
have shown in this Article that the tendency to reduce the general to
the particular, all in the name of a more fully rational choice, contin-
ues to plague rational choice theory. Sometimes this tendency shows
up as the temptation to see alternatives for choice in a partition-
independent way, as if features of the alternatives themselves were all
that mattered for choice and never features shared with other alterna-
tives in the choice set. This is what our discussion of the choice con-
sistency condition WARP revealed. At other times, the propensity for
particularity is manifested in the tendency to reduce what is attractive
in a whole to what is attractive in its parts. But, as we saw in our dis-
cussion of the sure thing principle (or strong independence condi-
tion), our understanding of a choice situation and, therefore, what we
should rationally do under that understanding, varies according to
whether the choice is seen as a whole or as a disjunction of its parts.
These different notions of rationality have been in play in the be-
haviorists’ experiments on choice for some years now. But, for the
most part, the results of these experiments have not been organized
under an alternative conception of rational choice. This Article has
tried to suggest that the alternative conception that is required, one
based on categorical reason, is part of a long-standing theoretical tra-
dition, and that rational choice theorists would do well to look to this
tradition to solve some systematic difficulties that they confront in so-
cial choice theory and the theory of games. That law forms part of
this tradition of categorical reason suggests further, perhaps, that le-
gal theorists have a special obligation to show them the way.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:05:46 +00:00�������������
All use subject to https://about.jstor.org/terms
- Contents
- Issue Table of Contents
p. 1169
p. 1170
p. 1171
p. 1172
p. 1173
p. 1174
p. 1175
p. 1176
p. 1177
p. 1178
p. 1179
p. 1180
p. 1181
p. 1182
p. 1183
p. 1184
p. 1185
p. 1186
p. 1187
p. 1188
p. 1189
p. 1190
p. 1191
p. 1192
p. 1193
p. 1194
p. 1195
p. 1196
p. 1197
p. 1198
p. 1199
p. 1200
p. 1201
p. 1202
p. 1203
p. 1204
p. 1205
p. 1206
p. 1207
p. 1208
p. 1209
p. 1210
University of Pennsylvania Law Review, Vol. 151, No. 3 (Jan., 2003) pp. i-x+707-1290
Front Matter [pp. i-x]
Symposium: Preferences and Rational Choice: New Perspectives and Legal Implications
Introduction [pp. 707-715]
The Domain of Preference [pp. 717-746]
Paradoxes of the Safe Society: A Rational Actor Approach to the Reconceptualization of Risk and the Reformation of Risk Regulation [pp. 747-786]
Takeover Defense When Financial Markets Are (Only) Relatively Efficient [pp. 787-824]
Will as Intertemporal Bargaining: Implications for Rationality [pp. 825-862]
Before and after: Temporal Anomalies in Legal Doctrine [pp. 863-885]
Can Utilitarianism Justify Legal Rights with Moral Force? [pp. 887-915]
Prudence and Constitutional Rights [pp. 917-961]
Is Risk a Harm? [pp. 963-1001]
Beyond the Precautionary Principle [pp. 1003-1058]
Trust, Guilt, and Securities Regulation [pp. 1059-1095]
The Jurisprudence of Greed [pp. 1097-1133]
Value Analysis of Political Behavior. Self-Interested: Moralistic:: Altruistic: Moral [pp. 1135-1167]
Rational Choice and Categorical Reason [pp. 1169-1210]
Regulation for Conservatives: Behavioral Economics and the Case for “Asymmetric Paternalism” [pp. 1211-1254]
The Puzzle of “Ex Ante Efficiency”: Does Rational Approvability Have Moral Weight? [pp. 1255-1290]
Back Matter
Do Lawyers Improve the Adversary System? A General Theory of Litigation Advice and
Its Regulation
Author(s): Stephen McG. Bundy and Einer Richard Elhauge
Source: California Law Review , Mar., 1991, Vol. 79, No. 2 (Mar., 1991), pp. 313-420
Published by: California Law Review, Inc.
Stable URL: https://www.jstor.org/stable/3480690
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
is collaborating with JSTOR to digitize, preserve and extend access to California Law Review
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
https://www.jstor.org/stable/3480690
Do Lawyers Improve the Adversary
System? A General Theory of
Litigation Advice and Its
Regulation
Stephen McG. Bundy
Einer Richard Elhauge
I. The Basic Account of How Parties and Lawyers Influence
the Information Reaching Tribunals in an Adversary
System …………….. ………………………….. 320
A. The Parties, the Tribunal, and the Litigation Process 320
B. Party Access to Information and the Role of
Evidentiary Sanctions …………………………….. 323
C. Decisionmaking About Investigation, Presentation, and
Influencing Evidentiary Sanctions ………………. 326
D. Accounting for the Contribution of Lawyers ……… 332
E. Distributions of Access, Error, and Information ……. 335
II. The Basic Account of the Informational Effects of
Litigation Advice ……………….. ………………. 335
A. Legal Advice That Helps Parties Categorize
Information …………………………………… 337
B. Legal Advice That Alters Investigation ……………. 343
C. The Effects of Legal Advice on Effective Evidentiary
Sanctions ……………………… …………… 348
D. Summary of the Basic Account ………………….. 357
III. The Generality of the Basic Account Conclusions ………. 361
A. Cost-Related Factors …………………………… 362
B. Different Motivations ……………. … ……… 366
C. Information That Influences Tribunals in an
Undesirable Way …………………………….. 370
D. Skewed Distributions of Information, Access,
or Error ……………………………………… 372
E. Conclusion …………………………… …… 379
IV. The Social Desirability of the Informational Effects of
Litigation Advice ………… . …………………. 380
A. Deterrence Theories……………………………. 380
B. Nondeterrence Theories…………………………. 392
313
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
314 CALIFORNIA LAW REVIEW [Vol. 79:313
C. Conclusion ……………………………………. 395
V. Regulating the Provision of Litigation Advice …………. 396
A. Measures That Encourage or Discourage the Use of
Lawyers ……………………………….. 396
B. Attorney-Client Confidentiality …………………. 401
C. Restrictions on Advice About Responding to
Investigation ………………………………….. 413
Conclusion ………………………………. …………… 419
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
Do Lawyers Improve the Adversary
System? A General Theory of
Litigation Advice and Its
Regulation
Stephen McG. Bundyt
Einer Richard Elhauget
This Article presents a rational actor account of how litigation advice
influences the information that reaches the tribunal in an adversary sys-
tem. The authors conclude that although advice has some disturbing or
ambiguous informational effects, on balance providing advice to one or
both parties will generally improve the capacity of tribunals to determine
who should be sanctioned and who should not. They also identify condi-
tions under which advice is likely to be socially undesirable. The authors
apply their account to analyze a range of policy issues, including whether
to award fees to private attorneys general, whether to guarantee representa-
tion to criminal defendants under the sixth amendment, and whether the
due process clause should require access to counsel for claimants in govern-
ment benefits cases. They also analyze various doctrines of attorney-client
confidentiality and rules restricting access to advice about responding to
investigations or other questioning.
The lawyer’s role in the adversary system frequently calls for con-
duct that appears to thwart truthful or just outcomes. Acting as an advi-
sor, the lawyer certainly may, and arguably must, provide her clients
with complete and accurate advice, even when she reasonably believes
that doing so will cause them to withhold or suppress evidence.’ In
t Acting Professor of Law, Boalt Hall School of Law, University of California, Berkeley.
A.B. 1973, Harvard College; J.D. 1978, Boalt Hall School of Law, University of California,
Berkeley.
T Acting Professor of Law, Boalt Hall School of Law, University of California, Berkeley.
A.B. 1982, Harvard College; J.D. 1986, Harvard Law School.
We wish to express our gratitude for the generous support of the Boalt Hall Fund and for the
helpful comments of McGeorge Bundy, Evan Caminker, James Crawford, William Fletcher,
Kenneth Graham, Henry Hecht, Tom Jorde, Louis Kaplow, Lewis Kornhauser, John Leubsdorf,
Paul Mishkin, Bob Mnookin, Rachel Moran, Andrea Peterson, Robert Post, Dan Rubinfeld, Steven
Shavell, Jan Vetter, and participants in seminars at Boalt Hall School of Law, Harvard Law School,
Stanford Law School, and the University of Virginia School of Law.
1. The initial discussion draft of the Model Rules would have forbidden lawyers from giving
advice that “the lawyer can reasonably foresee will . . . be used by the client to further an illegal
315
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
316 CALIFORNIA LAW REVIEW [Vol. 79:313
litigation, the lawyer’s obligations of zeal2 and confidentiality3 require or
permit her to engage in a host of dubious activities: withholding evi-
dence, even when the resulting record is radically incomplete;4 presenting
documents or testimony that she believes, based on information unavail-
able to the tribunal, to be false;5 discrediting through cross-examination
witnesses she knows to be truthful;6 and arguing for inferences from the
evidence that she knows are unwarranted.7
Questions about the social desirability and moral standing of this
troubling role have long divided those who think and write about adjudi-
cation and lawyers. Defenders argue that lawyer-aided adversary
factfinding is more likely than the alternatives to produce an accurate
decision because it leads to a better evidentiary record. Delegation to the
parties and their lawyers harnesses their competing interests (and the
lawyers’ skill) in the service of the tribunal’s interest in receiving a com-
plete account of the dispute.8 Competition between lawyers is also said
course of conduct.” MODEL RULES OF PROFESSIONAL CONDUCT Rule 2.3 (Discussion Draft,
January 30, 1980). Later versions dropped this language in favor of a prohibition against offering
knowing counsel or assistance in a crime or fraud in Model Rule 1.2. MODEL RULES OF
PROFESSIONAL CONDUCT Rule 1.2 (1989) [hereinafter MODEL RULES]. The comment to Model
Rule 1.2 states that a “lawyer is required to give an honest opinion about the actual consequences
that appear likely to result from a client’s conduct,” and that “[t]he fact that a client uses advice in a
course of action that is criminal or fraudulent does not, of itself, make a lawyer a party to the course
of action.” Id. Rule 1.2 comment. The strong, though not inevitable, inference is that a lawyer is
not free to withhold advice simply on the ground that it may lead to wrongdoing.
2. See MODEL CODE OF PROFESSIONAL RESPONSIBILITY EC 7-1 (1980) [hereinafter MODEL
CODE] (“The duty of a lawyer, both to his client and to the legal system, is to represent his client
zealously within the bounds of the law, which includes Disciplinary Rules and enforceable
professional regulations . . . .”); MODEL RULES, supra note 1, Rule 1.3 comment (a lawyer “may
take whatever lawful and ethical means are required to vindicate a client’s cause or endeavor” and
“should act with commitment and dedication to the interests of the client and with zeal in advocacy
upon the client’s behalf”).
3. MODEL CODE, supra note 2, DR 4-101; MODEL RULES, supra note 1, Rule 1.6.
4. See C. WOLFRAM, MODERN LEGAL ETHICS 639-40, 640 n.31 (1986) (citing cases).
5. See MODEL CODE, supra note 2, DR 7-102(A)(4) (attorney only barred from presenting
evidence known to be false); C. WOLFRAM, supra note 4, at 656. The Model Rules contain a
provision that allows, but does not require, a lawyer to refuse to present evidence that would be
persuasive to the tribunal but that she reasonably believes is false. MODEL RULES, supra note 1,
Rule 3.3(c). The Model Code has no comparable provision. Moreover, a criminal defense lawyer
may fail to provide the effective representation constitutionally required if she refuses to present
exculpatory evidence that she believes but does not “know” beyond a reasonable doubt to be false.
Cf Johns v. Smyth, 176 F. Supp. 949, 953 (E.D. Va. 1959) (holding that a prisoner’s due process
rights were violated when his attorney refused to argue for acquittal based on a client statement he
believed to be false). Furthermore, where the evidence consists of the client’s testimony, a rule
giving criminal defense counsel discretion not to present evidence might well violate the client’s fifth
amendment right to testify. To our knowledge, no court has yet adjudicated the constitutionality of
Model Rule 3.3 in criminal cases.
6. See United States v. Wade, 388 U.S. 218, 257-58 (1967) (White, J., concurring in part and
dissenting in part); see also C. WOLFRAM, supra note 4, at 650-51 (quoting Wade).
7. See generally Subin, The Criminal Defense Lawyer’s “Different Mission”. Reflections on the “Right” to Present a False Case, 1 GEO. J. LEGAL ETHICS 125 (1987).
8. See G. HAZARD, JR., ETHICS IN THE PRACTICE OF LAW 121 (1978); Luban, The
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 317
to improve the quality of information presented. Because each lawyer
will vigorously attack unreliable information that the opposing counsel
presents, such information will either be withheld or promptly discred-
ited. In this account, confidentiality is the handmaiden of effective fact-
development: if the law instead compelled lawyers to disclose the results
of their investigations, it would weaken the incentive to investigate.9
Critics of adversary advocacy contend that lawyers make no contri-
bution to truth.’0 They point to the lack of hard evidence that competi-
tion between legally advised partisans improves outcomes.” They also
sharply question the underlying logic of the defenders’ position. Given
the lawyer’s freedom or obligation to make partial and even affirmatively
misleading evidentiary presentations, critics argue, why should we
believe competition has an inherent tendency to enlighten the tribunal?
Competitive presentations “may simply pile up … confusion.”‘2 And
when one party has superior access to advice, delegation to private initia-
tive seems more likely to skew the quest for truth than to further it.’3
The emerging literature that applies the methods of economics to
the study of adversary adjudication echoes the traditional debate. Some
economic accounts assume that improved legal efforts enhance a party’s
chances of victory and increase the amount of information reaching
tribunals.’4 Lawyers enable parties to, among other things, investigate,
organize, and present information in digestible form, and expose inaccu-
racies in the opponent’s information.” The lawyer’s zeal ensures that
motivation is high, and competition from the opposing counsel checks
potential excesses and ensures that each party’s increased effort enhances
Adversary System Excuse, in THE GOOD LAWYER 83, 94 (D. Luban ed. 1984); Saltzburg, Lawyers,
Clients, and the Adversary System, 37 MERCER L. REV. 647, 656 (1986); cf Fuller, The Adversary
System, in TALKS ON AMERICAN LAW 30, 31 (H. Berman ed. 1961) (discussing the importance of
“partisan zeal” in achieving a fair outcome).
9. See, e.g., M. FREEDMAN, LAWYERS’ ETHICS IN AN ADVERSARY SYSTEM 27, 30 (1975).
10. See J. FRANK, COURTS ON TRIAL (1947) (adversary system retards discovery of truth); M.
FRANKEL, PARTISAN JUSTICE (1980) (same); Pound, The Causes of Popular Dissatisfaction With the
Administration of Justice, 29 ABA REPORTS 395, 404-06 (1906) (decrying “the sporting theory of
justice”).
11. See Hazard, Rules of Ethics: The Drafting Task, 36 REC. A.B. CITY N.Y. 77, 93 (1981); see
also D. LUBAN, LAWYERS AND JUSTICE: AN ETHICAL STUDY 74 n.14 (1988) (citing Hazard);
Rhode, Ethical Perspectives on Legal Practice, 37 STAN. L. REV. 589, 596 (1985) (same).
12. See Luban, supra note 8, at 94; accord M. FRANKEL, supra note 10, at 16; Simon, Ethical
Discretion in Lawyering, 101 HARV. L. REV. 1083, 1140 (1988).
13. A classic statement of this view is contained in J. FRANK, supra note 10, at 93-99; see also
M. FRANKEL, supra note 10, at 19 (noting economic constraints on parties’ ability to litigate);
Simon, The Ideology of Advocacy: Procedural Justice and Professional Ethics, 1978 WIs. L. REV. 29,
49-50 (discussing differential access to legal services).
14. See generally Cooter & Rubinfeld, Economic Analysis of Legal Disputes and Their
Resolution, 27 J. ECON. LITERATURE 1067, 1071-75, 1087 (1989) (reviewing literature).
15. R. POSNER, ECONOMIC ANALYSIS OF LAW ? 19.1, at 492 (3d ed. 1986).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
318 CALIFORNIA LAW REVIEW [Vol. 79:313
rather than impairs the accuracy of adjudication.16
Other economic accounts of legal advice tend to support the critical
perspective. One leading recent account analyzes advice on the assump-
tion that each party has exclusive access to its own information and that
the sole function of lawyers is to assist parties in selecting which informa-
tion to present.17 Based on those assumptions, the account reasons that
legal advice has no tendency to increase the amount of information
reaching the tribunal, and that its only effect is to skew the mix of infor-
mation presented in favor of the party receiving advice.18 Not surpris-
ingly, the account concludes that litigation advice has an ambiguous
effect both on the accuracy of adjudication19 and on the behavior of per-
sons who decide how to act based on adjudicated outcomes,20 and there-
fore has “questionable social value.”21
We share with prior scholars the belief that rational actor analysis
can help illuminate the lawyer’s role in an adversary system. But we
believe that prior rational actor accounts have been partial and simplis-
tic. Our aim is to provide a comprehensive and accurate examination,
within a rational actor framework, of the considerations that determine
the informational effects and social value of litigation advice. Our
account suggests that defenders of litigation advice give insufficient atten-
tion to its undeniable dubious or adverse consequences. Specifically, pro-
viding advice to a previously unrepresented party may increase his
capacity to withhold and suppress information, to inflict costs on his
opponent, and to present false or prejudicial information. Moreover,
competitive presentation does not invariably prevent or correct inaccu-
racy, especially where one party has more access to legal advice or can
control or suppress more information. Defenders ignore or minimize
these effects by stressing cases in which both parties have similar access
to advice and to information.
Critics are, however, too quick to dismiss several general effects of
advice that increase the amount of information reaching the tribunal.
Advice enhances competition in the presentation of information available
to both parties, and it enables parties to sort relevant from irrelevant
information. By improving investigation, it increases the amount of
information identified and presented. And it introduces into the process
a second actor, the lawyer, whose personal susceptibility to sanctions for
misconduct in withholding or suppressing information will sometimes
16. See id.
17. See Kaplow & Shavell, Legal Advice About Information to Present in Litigation: Its Effects
and Social Desirability, 102 HARV. L. REV. 565, 568, 571-75 (1989).
18. Id. at 568, 577, 581, 595.
19. Id. at 569, 581, 603.
20. Id. at 568, 586, 596, 614.
21. Id. at 597, 614.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 319
cause the tribunal to receive information that an unadvised party would
have withheld or suppressed. Critics avoid these positive aspects of
advice by stressing cases in which there is little commonly available
information, access to advice or to information is unbalanced, or lawyers
are insensitive to sanctions.
We develop our account of legal advice in several stages. Part I
presents our basic account of how parties in an adversary system influ-
ence the information reaching tribunals. Part II then applies this
account to reach initial conclusions about the general informational
effects of legal advice. We conclude that, while advice has many dis-
turbing or ambiguous informational effects, on balance providing litiga-
tion advice to one or both parties will generally increase the information
reaching tribunals and improve the capacity of tribunals to determine
who deserves to be sanctioned and who does not.
Although our basic account is more complex and comprehensive
than those offered in the prior literature, it nonetheless abstracts from
several important complications. We reintroduce those complications in
Part III, and use them to test our initial conclusions and to identify con-
ditions where those conclusions will not hold and where litigation advice
may thus have ambiguous or harmful informational effects. Part IV then
normatively evaluates the informational effects of legal advice under vari-
ous conditions. We there conclude that, under the conditions most likely
to describe its general effects, litigation advice is socially desirable under
a variety of instrumental and noninstrumental standards.
Ultimately, conclusions about the desirability of advice turn on
empirical questions. We do not purport to answer all those questions,
though we offer plausible intuitions about many. We do, however, claim
to have isolated the right set of empirical questions. If we are correct in
that claim, our account does more than focus the general academic
debate about the desirability of litigation advice on the right set of issues.
It also identifies the conditions likely to make particular forms of advice
sufficiently harmful to merit legal restriction or sufficiently desirable to
merit legal encouragement. Our analysis is thus relevant to a variety of
real-world legal issues, some of which are discussed in Part V.
Section V(A) presents our analysis of measures generally intended to
encourage or discourage access to advice. These include rules wholly
barring the use of lawyers, such as those that sometimes apply in small
claims or alternative dispute resolution procedures. They also include
measures that encourage or discourage retention of a lawyer, such as
rules allowing fee awards to lawyers for parties who act as private attor-
neys general, subsidizing representation for criminal defendants under
the sixth amendment, or restricting the sum a private citizen can pay for
advice. Our analysis also has implications for narrower, special purpose
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
320 CALIFORNIA LAW REVIEW [Vol. 79:313
rules that seek to increase the information reaching tribunals. Section
V(B) illustrates the implications of our analysis for proposals to reform
attorney-client confidentiality, and Section V(C) applies our analysis to
rules that restrict access to legal advice about responding to investigation
or other questioning.
I
THE BASIC ACCOUNT OF How PARTIES AND LAWYERS
INFLUENCE THE INFORMATION REACHING
TRIBUNALS IN AN ADVERSARY SYSTEM
To provide an adequate analysis of the informational effects of legal
advice, we first need an account of how parties and lawyers generally
influence the information reaching tribunals in an adversary system. For
clarity and emphasis, we initially adopt a stylized account of litigation
that abstracts from certain important complications about costs and set-
tlement, party and lawyer motivation, the dubious probative value of
some information, and skewed distributions of informational access, liti-
gation error, and favorable information. After using this basic account
to derive initial conclusions about the informational effects of legal
advice in Part II, we reintroduce these complications in Part III and
explore whether and how they might alter our conclusions.
Although necessarily simplified, our basic account is nonetheless
complex in that it considers both relevant and irrelevant information,
information that is available to both parties, decisions about investiga-
tion, withholding, and suppression, and sanctions for withholding and
suppressing evidence. Section I(A) states our basic assumptions about
party motivations, the tribunal, and the activities that influence the infor-
mation reaching the tribunal. Section I(B) provides a brief account of
the pertinent types of information, in terms of both their effect on the
tribunal and their availability to the parties, and explains how sanctions
for withholding and suppression determine the availability of informa-
tion to each party. We describe in greater detail in Section I(C) how
parties make decisions about investigating, presenting information, and
influencing the level of evidentiary sanctions. Section I(D) makes some
initial assumptions about how lawyers influence litigation conduct.
Finally, Section I(E) states initial assumptions about the distribution of
informational access, litigation error, and favorable and unfavorable
information.
A. The Parties, the Tribunal, and the Litigation Process
In our account, two parties litigate disputed questions of fact before
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 321
a tribunal.22 The tribunal is completely passive.23 It does not conduct its
own investigation. Nor can it recognize, without the parties’ assistance,
all information embodied in the evidence placed before it. Instead, it
determines the appropriate legal sanctions to impose based solely on the
information identified and presented by the parties.24 Parties are sanc-
tioned primarily for the conduct that gives rise to the litigation, which we
call “primary conduct.” Such sanctions include damages, fines, injunc-
tions, and imprisonment.
But the tribunal also imposes “evidentiary sanctions” to regulate lit-
igation conduct that influences the information reaching the tribunal,
such as withholding or suppressing information. Formal evidentiary
sanctions include discovery sanctions and penalties for perjury or refus-
ing to testify.25 But many evidentiary sanctions are not codified or
explicit. For example, the most effective sanction for false testimony may
not be the criminal sanction for perjury, but the adverse inference the
tribunal might draw about a party’s entire case if she offers testimony
that appears to be false.26
Our stylized account also assumes that each party is a rational, risk-
neutral actor who seeks the combination of her litigation costs and pri-
mary and evidentiary conduct sanctions that most advantages herself.
Defendants seek to minimize the sum of total expected sanctions and
costs. Plaintiffs seek to maximize the net value of primary sanctions, less
evidentiary sanctions and costs. For example, where a plaintiff has unfa-
vorable information that will cost her $1,000 to produce and, if pro-
22. Although real litigation often involves three or more parties, there are usually two sides to
any factual issue, and our analysis is unaffected if there is more than one party to a side. The
tribunal could be a court, jury, arbitrator, administrative adjudicator, or any other body that resolves
disputes by adjudicating facts.
23. Our basic assumption of a passive tribunal reflects the traditional ideal of the adversary
system. See, e.g., Feeley, The Adversary System, in 2 ENCYCLOPEDIA OF THE AMERICAN JUDICIAL
SYSTEM 753 (R. Janosik ed. 1987) (describing adversarial factfinding). Although many tribunals are
not as passive as we describe, see, e.g., Resnik, Managerial Judges, 96 HARV. L. REV. 374, 376
(1982), we assume passivity because our interest is in analyzing the effect of legal advice on those
elements of adversary factfinding that are party-orchestrated. Our argument therefore does not
speak directly to the question of whether the tribunal should assume a more active or “inquisitorial”
role in factfinding, although some of our discussion in Part IV of adjusting sanctions bears on that
question.
24. The determination of the appropriate sanction involves conclusions about both liability and
remedies. We do not examine the rule of decision by which the tribunal determines which
information warrants what sanction, other than to assume that this rule bears some correlation to
the social desirability of the conduct in question when the information is true, relevant, and
nonprejudicial.
25. See, e.g., 18 U.S.C. ?? 1510-1513 (1984) (criminal sanctions for obstruction of justice); 18
U.S.C. ? 1621 (1984) (criminal sanctions for perjury); FED. R. CIV. P. 37 (discovery sanctions).
26. Other evidentiary sanctions are explicitly incorporated into the rules by which the tribunal
determines the outcome on the merits. For example, a tribunal may punish the failure to produce
certain information in discovery by allowing or requiring the factfinder to draw an inference adverse
to the party failing to produce. See, e.g., FED. R. CIV. P. 37(b)(2)(A).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
322 CALIFORNIA LAW REVIEW [Vol. 79:313
duced, will reduce her expected judgment by $9,000, she will nonetheless
produce it if the sum of her costs of withholding and expected eviden-
tiary sanctions for withholding exceeds $10,000.27 We will call such
parties “sanction-optimizers.”
Our assumption that parties are sanction-optimizers necessitates
two further simplifying assumptions: that every case is litigated to judg-
ment, and that parties never engage in litigation conduct to drive up their
opponents’ litigation costs.28 This allows us initially to avoid the com-
plex effects on decisions about investigation and presentation that flow
from the constant possibility that a party might settle or abandon a suit.
It also allows us initially to avoid some of the strategic aspects of litiga-
tion explored in the game theory literature.29
Although we reintroduce more realistic assumptions about all these
features in Part III, the basic assumption of purely self-interested party
behavior is perhaps sufficiently fundamental and controversial to merit
some attention up front. We adopt this working assumption for two rea-
sons. First, prior studies of civil and criminal litigation30 and our per-
sonal observations give us some confidence that the assumption captures
the motivations of a substantial number of litigants. Second, our intui-
tive sense is that, as a uniform assumption about party motivation, the
assumption that parties respond solely to the threat of sanctions provides
a good “worst case” test of legal advice in an adversary system.31
We recognize, however, that not all litigants are rational actors, and
those who are do not invariably have optimizing sanctions as their exclu-
sive (or even first) priority. Some litigants, whom we call “law-abiding,”
may feel obligated to comply with the law governing the production of
information even if the sanctions for violating the law are alone insuffi-
cient to compel compliance. Others, whom we call “truth-tellers,” may
feel obligated to tell the “whole truth” even if the law does not require
them to do so. Government litigants, who have a duty to pursue justice
and a relative lack of personal interest, may seem particularly likely to
deviate from the sanction-optimizing assumption. Section III(B)
27. Our account does not, however, assume that the only values at stake are monetary or
monetizable. Parties may also take into account the nonmonetary benefits of litigation conduct
(such as the opportunity to have their “day in court”), the nonmonetary costs (such as time or
reputation), and nonmonetary sanctions (such as loss of liberty).
28. Parties may, in deciding on their own conduct, take into account the opponent’s likely
litigation conduct. See infra Section I(B). But we initially assume that they take their opponent’s
likely conduct as essentially unaltered by their own litigation conduct.
29. See, e.g., Cooter & Rubinfeld, supra note 14, at 1072-73; Sobel, An Analysis of Discovery
Rules, 52 LAW & CONTEMP. PROBS. 133 (Winter 1989).
30. See, e.g., K. MANN, DEFENDING WHITE COLLAR CRIME passim (1985); Brazil, The
Adversary Character of Civil Discovery, 31 VAND. L. REV. 1295, 1312-32 (1978).
31. We will see in Section III(B) that, where party motivations are not uniform, the “worst
case” is providing legal advice to a sanction-optimizing party who faces a truth-teller.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 323
describes how our account might change when parties are law-abiding or
truth-telling.
Under our initial assumption of sanction-optimizing behavior, each
party seeks (within limits set by litigation costs and evidentiary sanc-
tions) to influence sanctions by influencing the information that reaches
the tribunal. Whether the party wants the information to reach the tri-
bunal will depend on the party’s judgment about whether the informa-
tion is “favorable” or “unfavorable”: that is, whether it would tend to
cause the tribunal to reach a factual conclusion favorable or unfavorable
to the presenting party.32 Because parties are opposed on each disputed
factual issue, an item that is favorable for one is unfavorable for the
other.
Initially, we assume that the tribunal’s response to information actu-
ally reflects that information’s genuine tendency to prove or disprove a
contested proposition of fact. Thus, our initial account excludes infor-
mation that has a favorable or unfavorable impact but is false, appeals to
prejudice, or achieves its effect by drowning the tribunal in redundant or
irrelevant information. We reintroduce such kinds of information in Sec-
tion III(C). In our initial account, however, irrelevant information refers
to all information that is neither favorable nor unfavorable. Parties in
our initial account thus have no desire to identify or present it.
Four basic activities influence the information reaching the tribunal.
First, each party seeks to identify information through investigation.
Second, each party seeks to prevent the opponent from identifying infor-
mation by withholding or suppressing it. Third, from the information
she has identified, each party selects information to present to the tribu-
nal. Fourth, the parties seek to lower their own evidentiary sanctions
and raise those applicable to their opponent.
B. Party Access to Information and the Role of Evidentiary Sanctions
When a dispute arises, each party knows of some information that
might be presented to the tribunal. “Investigation” is how a party identi-
fies additional presentable information. Investigation typically involves
interviewing prospective witnesses and inspecting documents and other
physical evidence. A party may investigate information that she pos-
32. This definition parallels that in Kaplow & Shavell, supra note 17, at 577-82, 595-96. Our
discussion assumes that individual items of favorable and unfavorable information are discrete. In
practice, of course, favorable and unfavorable information often comes in aggregate lumps, so that
decisions to produce favorable and unfavorable items are inextricably linked. For example, a
criminal defendant’s decision to testify in order to present a favorable alibi must also take account of
the unfavorable information about the defendant’s prior criminal record that cross-examination will
elicit. Aggregation generally does not affect our conclusions so long as parties continue to make
decisions based on the net balance of favorable and unfavorable information in the relevant lump.
An exception is discussed at infra note 71.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
324 CALIFORNIA LAW REVIEW [Vol. 79:313
sesses, may conduct informal and voluntary inquiries of cooperative
sources, or (with the opponent and other unwilling sources) may proceed
through formal discovery or examination at trial. Formal investigation
carries with it the threat of evidentiary sanctions.
1. Party-Controlled and Discoverable Information
Each party possesses some information when a dispute arises.33 If a
party believes that information in her possession is unfavorable, she will
want to keep it from the opponent and the tribunal. A party “with-
holds” information in her possession by failing to volunteer or produce it
to the opponent or the tribunal. Leaving aside the direct costs of the
actions needed to withhold or produce information, a party will withhold
information she believes is unfavorable when the expected evidentiary
sanction for withholding is less than the expected adverse effect on the
judgment from making the information available to the opponent or the
tribunal.
Some information in each party’s initial possession can be identified
by the opponent in discovery or at trial: if the opponent makes a demand
for production, and if the perceived evidentiary sanctions34 (coupled with
the direct costs of withholding) exceed the perceived harm of production
(including the direct costs of production), the party will produce the
information. We will call such information the party’s “discoverable
set.” For the remaining information in each party’s possession, sanctions
will not be enough to compel production even if a proper demand is
made. Sometimes this will occur because the party can get away with
improper withholding, but sometimes the law expressly permits with-
holding, for example, because of a valid evidentiary privilege. We will
call information in a party’s possession whose disclosure cannot be com-
pelled the party’s “controlled set.”
2. Commonly Available Information
Both parties potentially can identify some of the information that
neither party possesses at the outset of litigation.35 Such information will
include information in the public domain and information from sources
33. A party “possesses” information when an opponent can only obtain that information if the
party volunteers it or is compelled to produce it by the threat of evidentiary sanctions. Possession
thus encompasses not only physical possession of documents or other evidence but also a party’s
control over her own recollection.
34. Perceived sanctions represent the party’s subjective estimate of actual expected sanctions.
The expected evidentiary sanction is the product of the actual probability that the act will be
detected and sanctioned by the tribunal and the actual magnitude of the sanction that will be
imposed.
35. We exclude from consideration information that neither party could identify (with or
without advice), because legal advice cannot affect the presentation of such information.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 325
not aligned in interest with either side. For example, where the plaintiff
brings suit for her injuries in an automobile accident case, information
potentially available to both parties includes the site of the accident,
police and hospital records, the recollections of nonparty eyewitnesses or
investigating officers, and the published literature on and available
experts concerning any disputed technical or scientific issues.36 We will
call the set of information that is initially available to both parties the
“commonly available” set.
3. Suppressible Information
Just as a party in possession of information can seek to withhold it, a
party can sometimes attempt to prevent her opponent from identifying or
presenting information that is commonly available or in the opponent’s
possession. A party may, for example, destroy a document in another’s
files. We call such conduct “suppression.”37 Both physical and testimo-
nial evidence can be suppressed. Sometimes the law encourages suppres-
sion because it furthers some other social policy, such as respecting
privacy or preventing government coercion. For example, criminal
defendants frequently move to suppress evidence on the ground that it
was obtained in violation of the fourth or fifth amendments. Other forms
of suppression are less reputable. Skillful, suggestive interviewing can
alter a witness’ recollection, so that, without realizing it, she represses or
forgets information unfavorable to the interviewer.38 Prospective sources
of information can also be subjected to legitimate or illegitimate incen-
36. Investigation of information held by such sources will often be cooperative, since the
sources are not aligned with one party or the other. However, because nonaligned sources may not
be fully compensated for the cost of providing information, it is sometimes necessary to invoke
sanctions to obtain their information. The level of sanctions required to compel production need
not, however, vary depending on whether the information is favorable or unfavorable to the party
seeking it, because the source has no stake in the outcome.
Sources who share some but not all interests with one of the parties can be placed on a
continuum between nonaligned sources and sources whose interests are perfectly aligned with a
party. Information possessed by nonaligned sources is commonly available. Information possessed
by perfectly aligned sources will be analytically the same as information possessed by a party, so we
will treat perfectly aligned sources as one of the parties for the purposes of this Article. Information
possessed by partially aligned sources has some characteristics of commonly available information,
but it could also be said to be partially possessed by the party who shares interests with the source
(because she will have an advantage in investigating, gathering, and suppressing that information).
37. The difference between suppression and withholding is thus the source of the information
that someone prevents from reaching the tribunal. A party can only withhold information she
possesses but can suppress information possessed by another.
38. See, e.g., K. MANN, supra note 30, at 162 (explaining the “basic working principle that he
who arrives first gets a better interpretation of the facts … [because] the respondent’s perceptions
have been influenced by the interrogator’s construction of questions”); see also M. FREEDMAN,
supra note 9, at 68-89 (discussing the influence of interviewers on the information the interviewees
recalled). In addition, a party may use the record of an interview to suppress information. For
example, police officers who have interrogated a suspect may prepare a written statement that omits
exculpatory information volunteered by the defendant. If the suspect signs the statement, the
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
326 CALIFORNIA LAW REVIEW [Vol. 79:313
tives or threats. For example, a party may advise a nonparty witness of
the risk that she will incriminate herself if she testifies and of her fifth
amendment right to decline to answer incriminating questions.” Or a
party may threaten the potential witness with physical or financial injury
if she testifies.
Some suppression is subject to sanctions. Where suppression is pos-
sible, a party decides whether to suppress by determining whether the
information is unfavorable and then weighing the perceived value of sup-
pressing the information against both perceived sanctions for suppression
and any associated financial costs.40 Where perceived evidentiary sanc-
tions are sufficient to deter suppression, we call the information “nonsup-
pressible.” Such information can be identified and presented by both
parties. For each party, however, there will remain some information in
the commonly available set or in her opponent’s possession that she will
be able and willing to suppress, given the perceived costs and benefits of
doing so. We define information as “suppressible” if one of the parties
would be willing and able to suppress it under some circumstances.
C. Decisionmaking About Investigation, Presentation, and Influencing
Evidentiary Sanctions
1. Decisions About Investigation
A party will conduct an investigation only if the investigation’s per-
ceived value exceeds its cost.41 For an investigation to have positive
value, it must have some possibility of leading, directly or indirectly, to
prosecutor can later use it to cast unwarranted doubt on the exculpatory information when the
defendant offers it at trial.
39. Courts have held that a lawyer who gives such advice to a nonparty with the “corrupt
motive” of protecting her client may be convicted of obstruction of justice even though the advice
given is both accurate and apt. See United States v. Fayer, 523 F.2d 661, 663 (2d Cir. 1975); C.
WOLFRAM, supra note 4, at 647. To our knowledge, however, the Supreme Court has never
considered the first, fifth, or sixth amendment implications of this rule.
40. Often a significant factor in determining the expected sanctions for suppression is who
identifies the information first. A party who can identify information before her opponent identifies
it enhances her ability to suppress the information and reduces her expected sanctions for doing so
because her opponent is less likely to detect the suppression. At the same time, she reduces her
opponent’s ability to suppress the information and increases her opponent’s expected sanctions for
attempting to suppress it. This factor, however, will not always be determinative. Some criminals,
for example, may have no compunction about using threatened or actual physical harm to silence
witnesses who are already providing information to the prosecutor. Witness protection programs are
designed to combat this problem.
41. The perceived value of an investigation is the party’s subjective estimate of its expected
value. One calculates an investigation’s expected value by identifying the values of the
investigation’s possible outcomes, multiplying each outcome by the probability it will occur, and
summing those products over all possible outcomes. H. RAIFFA, DECISION ANALYSIS:
INTRODUCTORY LECTURES ON CHOICES UNDER UNCERTAINTY 9 (1968).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 327
relevant information.42 Investigations clearly have positive value to the
extent they are expected to yield favorable information. But many inves-
tigations that are expected to yield only unfavorable information may
also have positive value. Knowing about unfavorable information can
assist in forming one’s theory of the case, understanding the favorable
and unfavorable character of one’s own information, preparing to chal-
lenge the opponent’s case, and gathering additional favorable informa-
tion to present to the tribunal. Even irrelevant information can have
positive value if it provides investigative leads to helpful relevant infor-
mation. Investigations for relevant suppressible information will have
positive value whether the information is unfavorable or favorable
because such investigations enhance the likelihood of both suppressing
unfavorable information and preventing the suppression of favorable
information.
The potential negative informational effect of investigation is that it
can increase the likelihood that unfavorable information will reach the
tribunal.43 This can happen in two ways. First, investigation sometimes
requires a party to divulge information directly to the opponent or to
persons who are more likely than the party to disclose it to the opponent
or the tribunal.” Second, a party’s identification of unfavorable informa-
tion in investigation may make that information more accessible to the
opponent. For example, a party may expend great effort to locate and
interview a witness whom the opponent would not have found through
independent investigation and whose testimony turns out to be damag-
ing. Once the party has gathered the information, however, the oppo-
nent may be able to obtain it easily through a discovery demand directed
42. Given our assumptions in Section I(A), an investigation cannot have value simply because
it burdens the opponent or leads him to believe that he must spend money to counteract it.
43. Investigation can have adverse consequences other than identifying or divulging
unfavorable information. Sometimes the information that can lead the opposition to unfavorable
information is itself favorable. For example, a party may wish to prevent the opposing party from
obtaining favorable information because, if alerted to it, the opposing party will then identify
additional information to counter it. See Brazil, supra note 30, at 1317-18 (arguing that a party
anticipating litigation may decide not to depose a favorable witness in order to avoid having to give
the opponent an opportunity to respond to the witness’ information at trial).
44. For example, witness interrogation may itself reveal information held by the questioner, as
occurred in the infamous case of Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263 (2d Cir.
1979), cert. denied, 444 U.S. 1093 (1980). There, a nonparty witness for Kodak had prepared, but
subsequently forgotten, a very damaging document that had been improperly withheld during
discovery. On the eve of the witness’ trial testimony, a lawyer for Kodak investigating to obtain
information about the document showed it to the witness. This refreshed the witness’ recollection.
When asked about the document on cross-examination the following day, the witness reported its
existence, causing its belated production. See Kiechel, The Strange Case of Kodak’s Lawyers,
FORTUNE, May 8, 1978, at 192; see also Brazil, supra note 30, at 1321 (arguing that “the drafting
litigator is constantly under pressure not to initiate discovery that could help educate the
opponent”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
328 CALIFORNIA LAW REVIEW [Vol. 79:313
at the party.45
Where cost-effective, a party should generally have strong overall
incentives to investigate. A party always wants to identify information
expected to be favorable. If the party expects the information to be unfa-
vorable, the incentive to investigate depends in part on the opponent’s
access to that information. If the unfavorable information is suppress-
ible, opponent access gives the party strong incentives to identify the
information so she can suppress it. If the unfavorable information can-
not be suppressed, it might seem that the party has no incentive to search
for it, especially since the opponent might be able to “feed off” of the
party’s search. But the opponent’s ability to identify the information
independently may also strengthen the incentive to investigate for two
reasons.
First, it is valuable to know what the opponent’s case might contain.
Second, the likelihood of independent opponent identification decreases
the likelihood that the party’s investigation will cause information to
reach the tribunal that would not have reached it anyway. Without
knowing the relative likelihoods that the opponent will identify informa-
tion independently or as a derivative consequence of the party’s investi-
gation, one cannot determine the strength of the incentive (or
disincentive) to seek out unfavorable nonsuppressible information. But
the overall incentives to investigate (where cost-effective) should gener-
ally be high in all information sets, because the incentive to identify
favorable information is invariably strong and the incentive to identify
unfavorable information is at worst mixed.
A well-chosen investigation strategy can increase both the favorable
information that a party identifies and the unfavorable information iden-
tified by her opponent. A party decides whether to conduct an investiga-
tion by determining whether it has net positive expected value.46 But the
net expected value of an investigation can be positive, even though some
of its expected informational consequences are harmful. Many investiga-
45. See FED. R. CIV. P. 26(b)(1) (permitting discovery of all relevant nonprivileged
information); cf Shapiro, Some Problems of Discovery in an Adversary System, 63 MINN. L. REV.
1055, 1064-77 (1979) (arguing that a party whose lawyer uncovers extremely unfavorable
information from a nonparty witness is legally obliged to produce the information in response to a
properly framed interrogatory notwithstanding the work product doctrine). Investigation of
information in one’s own possession can also help an opponent identify that information. For
example, in responding to interrogatories, a civil litigant must conduct a “reasonable” investigation
of information she possesses. See J. FRIEDENTHAL, M. KANE & A. MILLER, CIVIL PROCEDURE
? 7.9, at 401 (1985). Parties anticipating interrogatories may therefore avoid conducting
investigations more extensive than “reasonable” for fear they may identify (and thus be obliged to
produce) unfavorable information they would not have identified in the compulsory “reasonable”
investigation.
46. Where a party believes that alternative investigations with positive expected value will
uncover the same favorable or helpful information, she will choose the investigation with the highest
value.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 329
tions will identify multiple items of evidence, some helpful and some
harmful.47 Moreover, a defining characteristic of investigations, given
that they are motivated by a party’s lack of knowledge about informa-
tion, is uncertainty about what their fruits will be.48 The most carefully
tailored investigation strategy will sometimes collect information of types
or from sets different from those sought, and thus cause unfavorable
information to reach the opponent or the tribunal that otherwise would
not have done so.
2. Decisions About Presenting Information and the Role of Advocacy
When the parties have completed their investigations, each will have
a pool of information that she has identified and can present.49 Each
must select the information to present from this pool. Subject to cost
constraints and evidentiary sanctions, each will seek to present all infor-
mation she believes is favorable and to withhold or suppress information
she believes is unfavorable or irrelevant.
Presenting information encompasses more than the mechanical
tasks of placing selected physical objects in evidence and eliciting oral
testimony. Getting information admitted into evidence makes it avail-
able to both sides but may not be sufficient to alert the tribunal to its
favorable or unfavorable character. Presentation thus often involves
advocacy, in which each side shapes the tribunal’s interpretation of the
admitted evidence by pointing out the favorable factual inferences that
can be drawn from the evidence and minimizing the unfavorable infer-
ences. The more passive, obtuse, overburdened, or inexperienced the tri-
bunal, the more advocacy is needed to ensure that the tribunal recognizes
all the information embodied in admitted evidence.
An example may illustrate the process of presenting and withhold-
ing information. Suppose that the defendant in a contract dispute
believes that her truthful account of conversations with the plaintiff will
assist her case. Accordingly, she testifies to those conversations and
points out the favorable inferences that can be drawn from them. The
47. Even where a single item is involved, it may have both helpful and harmful aspects. For
example, an item of unfavorable information may provide valuable investigative leads to favorable
information, but gathering it may greatly increase the likelihood that the opponent will obtain and
present it.
48. Indeed, the value of information will generally be uncertain even after the party has
investigated. For example, the party may not know for certain whether: (1) the opponent will
obtain (or would have obtained) the information independently; (2) the opponent will make a
discovery request for the collected information, and what perceived sanctions for nonproduction will
be if he does; (3) the information will lead to further positive value investigations; or (4) additional
information will arise that alters the information’s favorable or unfavorable character.
49. A party can directly present information it possesses. If the opponent or nonparties
possess the identified information, the party may have to mobilize sanctions to compel its
presentation. Sometimes the party may identify information but be unable to present it.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
330 CALIFORNIA LAW REVIEW [Vol. 79:313
same defendant’s correspondence file contains copies of several letters
supporting inferences unfavorable to the defendant. Because the oppo-
nent does not demand those copies, the party says nothing about them,
and the contents of her file are never introduced into evidence.50 The
defendant’s failure to produce the documents is a form of withholding.
The opponent finally offers in evidence a copy of one of the letters and
points out an inference supported by the letter that is unfavorable to the
defendant, but fails to point out several other even more damaging infer-
ences supported by the letter. The defendant, who perceives those infer-
ences, elects to remain silent about them. By failing to point out the
additional inferences to the tribunal, both the plaintiff (mistakenly) and
the defendant (correctly) have withheld that information from the
tribunal.
3. Party Efforts to Influence Evidentiary Sanctions and Secondary
Information
At the same time that parties are seeking to identify and present
information that will influence the tribunal in imposing sanctions for pri-
mary conduct, they are also engaged in a “litigation within a litigation”
whose subject is the level of sanctions for withholding and suppression.
In this litigation, each side seeks to ensure that the information reaching
the tribunal about its own evidentiary conduct is as favorable as possible
and that the information reaching the tribunal about the opponent’s evi-
dentiary conduct is as unfavorable as possible. Thus, a party will seek to
present her own information in a way which suggests that her presenta-
tion is complete and candid, while endeavoring at the same time to pres-
ent information which suggests that her opponent’s presentation is
incomplete and dishonest. The information the parties present on these
issues is in part the same information that is relevant to the tribunal’s
determination of the merits. But each party may also be able to present
some information that goes solely to the likelihood that information is
being improperly suppressed or withheld. We will call information that
influences the level of evidentiary sanctions “secondary information.”
4. An Illustrative Case
We can illustrate some of the factors that inform the process of mak-
ing decisions under uncertainty about investigation and presentation by
examining the traditional lawyers’ rule of thumb that in cross-examining
a hostile witness at trial, “a lawyer should never ask a question to which
50. This example is modeled on Professor Samuel Williston’s account of a case in which he
participated as a young lawyer, as recounted in Curtis, The Ethics of Advocacy, 4 STAN. L. REV. 3,
9-10 (1951).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 331
she does not already know the answer.”” The lawyer’s lack of knowl-
edge about the answer she will receive means that, although the inquiry
takes place before the tribunal, it involves a significant element of investi-
gation. Her ignorance indicates that she has no strong reason to believe
the question will produce favorable information. What is more, if the
witness withholds favorable information, the lawyer’s ignorance makes it
unlikely that she can bring evidentiary sanctions to bear by impeaching
the witness’ testimony.
If the information identified in cross-examination proves unfavora-
ble, the consequences are likely to be particularly harmful. The circum-
stances make clear that the opponent would not have independently
presented the information. Further, unlike information disclosed to the
opponent during out-of-court investigation, answers given during cross-
examination may reach the tribunal whether or not the opponent makes
a conscious decision to present the information. Nor is the party likely
to realize the benefits that normally flow from identifying unfavorable
information prior to trial. Because of the timing of the investigation, any
unfavorable information the party unearths is unlikely to aid in case
preparation or subsequent investigation.
In short, blind cross-examination involves a higher risk of adverse
consequences and a lower chance of positive payoff than normal investi-
gation.52 This does not mean that it will always be disadvantageous.
When opportunities for out-of-court investigation to identify favorable
information are limited, as they are in systems without pretrial discovery,
such inquiries are more likely to have positive expected value notwith-
standing the risks involved. One would expect parties in such a system to
take those risks more frequently than in a system allowing pretrial
discovery.53
51. See, e.g., T. MAUET, FUNDAMENTALS OF TRIAL TECHNIQUES 215 (2d ed. 1988).
52. Although statements of the rule of thumb typically restrict its application to cases of cross-
examination, much of the analysis here also applies to direct examination of a friendly witness in
circumstances where the examiner does not “know” the answer. Friendly witnesses should be more
likely than hostile witnesses to have favorable information. But direct examination typically will
have elicited the lion’s share of the unfavorable information possessed by the hostile witness.
Therefore, the mix of favorable and unfavorable information that has not been elicited from the
hostile witness may not differ from the mix of information that is possessed by a friendly witness but
was not identified in pretrial investigation.
53. Thus, Francis Wellman, writing near the turn of the century about lawyers in a system of
limited discovery, noted the maxim against cross-examination where one was uncertain of the
answer, but nevertheless stated that “[t]he most cautious cross-examiner will often elicit a damaging
answer.” F. WELLMAN, THE ART OF CROSS-EXAMINATION 13 (3d ed. 1923). This judgment
presumably reflects the fact that, in a regime without pretrial discovery, cross-examination often
might represent a party’s only opportunity to identify and present favorable information possessed
by a hostile witness.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
332 CALIFORNIA LAW REVIEW [Vol. 79:313
D. Accounting for the Contribution of Lawyers
An account of legal advice requires that we make certain assump-
tions about what lawyers know, what their advice costs, what their moti-
vations are, and whether their advice has symmetric or asymmetric
effects on the litigating parties. We assume that lawyers’ expertise assists
the full range of activities described above: investigation, categorizing
and presenting information, withholding, and suppression. Lawyers
know more than clients about whether information about primary con-
duct is favorable, unfavorable, or irrelevant. Lawyers also know more
about the available procedures and techniques for investigation, presenta-
tion, withholding, and suppression, about the sanctions applicable to cli-
ents’ evidentiary conduct, and about whether information is favorable,
unfavorable, or irrelevant on the issue of whether evidentiary sanctions
should be imposed. For example, lawyers know more about the rules
governing the compelled production of information, including the proper
form of demands for production, the ways of resisting such demands, and
the sanctions for failure to produce. As a result, lawyers can make more
accurate predictions about whether information is party-controlled, dis-
coverable, commonly available, or suppressible. In each of these areas of
professional expertise, we assume that lawyers’ knowledge is imperfect
but better than their clients’.54
We also assume, for the sake of simplicity, that legal advice is
costless. In consequence, in this simplified account parties have no
incentive to economize on legal advice, and lawyers have no incentive to
increase their profit by working too much or too little. We consider the
consequences of relaxing this important assumption in Section III(A).
We do not, however, assume that lawyers are perfectly loyal to their
clients’ interests. Sometimes lawyers will refuse to assist or engage in
evidentiary misconduct even though such misconduct would optimize
their clients’ sanctions. In our basic account, we assume that lawyers
behave this way only out of a self-interested concern for their own sus-
ceptibility to evidentiary sanctions for engaging in or assisting unlawful
withholding and suppression of evidence.”5 This assumption implies,
among other things, that evidentiary sanctions applicable to lawyers
sometimes abrogate lawyer-client confidentiality. We reserve until Sec-
54. If lawyers were unavailable during litigation, parties might be able to replace some of the
lost legal knowledge by self-study or by consulting lawyers before litigation. But there would be
limits on the ability of parties to substitute such sources of knowledge for lawyers. Accord Kaplow
& Shavell, supra note 17, at 599, 600-01. To the extent parties can obtain legal knowledge without
being represented by lawyers during litigation, that would affect only the overall extent, and not the
nature or direction, of the effects we predict under our model.
55. Section II(C)(2) describes some of these sanctions in greater detail.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 333
tion V(B) consideration of the comparative informational effects of legal
regimes that protect confidentiality more or less than the current regime.
Although many lawyers may refuse to participate in evidentiary
misconduct for other reasons, the assumption that lawyers avoid eviden-
tiary misconduct only because they fear sanctions allows us to address
the apparent “worst case” argument against legal advice. In Section
III(B)(3), we reconsider our results under the assumption that some law-
yers refuse to become involved in evidentiary misconduct because of pro-
fessional or moral obligations.
In much of our discussion of the effects of advice on party behavior,
we adopt a narrative convention which implies that lawyers give legal
advice and that clients then execute the acts of investigation, presenta-
tion, suppression or withholding that the lawyers have recommended.
Although this may sometimes happen, in fact clients often delegate exe-
cution of these tasks to lawyers, particularly when the tasks involve
investigation or appearances before the tribunal. Indeed, often a party
cannot obtain the full benefit of advice about investigation or presenting
information without delegating the execution of those tasks to her law-
yer. Where delegation occurs, lawyers’ increased personal susceptibility
to evidentiary sanctions may make them unwilling to allow or participate
in efforts to suppress or withhold information. This effect is important to
our conclusions and we address it in Section II(C)(2). Our narrative con-
vention is not meant to deny this important effect of delegation but
rather to simplify the exposition.
A final important issue in developing an account of the informa-
tional effects of litigation advice is whether to measure those effects under
a bilateral or unilateral account. A bilateral account compares a regime
in which both parties receive comparable litigation advice with a regime
in which neither side has any advice. A unilateral account holds the
advice given one party constant and compares the results when the oppo-
nent receives advice with the results when it does not.
Each approach has its virtues. A bilateral account more closely
approximates the traditional understanding of the adversary system of
factfinding, in which both parties operate with similar resources and
under similar rules governing access to advice. More important, if the
account’s predicted effects are to provide useful guidance for legal policy
analysis-rather than for individual decisions to retain counsel-the
account must measure the effects of the societal decision to employ or
reform rules that require, encourage, permit, discourage, or forbid litiga-
tion advice. Such societal decisions will often have bilateral effects.
Indeed, in civil cases between private parties, society may have little
choice but to allow litigation advice either to both parties or to neither.
Allowing litigation advice only to plaintiffs or only to defendants would
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
334 CALIFORNIA LAW REVIEW [Vol. 79:313
make little sense functionally because either side can benefit from slant-
ing the mix of information presented to tribunals and might be tempted
by improved prospects in litigation to engage in undesirable primary con-
duct. Such a distinction would also be administratively unworkable in a
system where almost any defendant can, given declaratory judgments,
counterclaims, and cross-claims, paint herself as a plaintiff.”56 When the
government opposes a private citizen, distinctions may be easier to draw,
both for functional reasons (the government’s duty to serve justice and,
in criminal cases, its distinctive legal disclosure obligations might make it
less likely to present evidence selectively) and for administrative reasons
(the government can be readily identified in every case). Other social
values, however, might still limit society to choosing between allowing
legal advice to both parties or neither.
On the other hand, a unilateral account has important applications.
In a troubling class of civil cases, litigation advice may be nominally
available to both sides but in practice available to only one side because
the other cannot afford adequate legal counsel.57 If the legal regime
could identify with confidence the cases falling in this category, one
could employ a unilateral account to describe the informational effects,
in this class of litigation, of a societal decision to either deny counsel to
both sides or subsidize counsel for the party who otherwise could not
afford it. The former solution is sometimes adopted in small claims
court; the latter underlies many legal aid programs.
In another class of cases that appear nominally bilateral, the quality
of legal advice available to both sides is adequate, but because one lawyer
is better than the other or because one party initially knows much less
law than the opponent, legal advice alters one party’s litigation behavior
more than it alters her opponent’s. Where the bilateral provision of legal
advice causes one party’s rate of error to decrease more than her oppo-
nent’s, the effects are in part bilateral, but the differential advantage flow-
ing from advice is more appropriately evaluated as a unilateral increase
in advice.
Finally, because it is easier to draw formal and functional distinc-
tions between the government and other parties, and because the govern-
56. Kaplow and Shavell argue that one can permit or disallow legal advice to only one side,
citing the contrasting roles of criminal prosecutors and defense attorneys. Kaplow & Shavell, supra
note 17, at 597 n.76. However, they offer no explanation of how to make distinctions useful for civil
litigation between two private parties.
57. These include cases where the expected value to one side does not merit the necessary
investment in legal counsel. They also include cases where one side is too poor to hire adequate
counsel and either cannot win a monetary recovery or cannot (due to legal restrictions or market
imperfections) finance the litigation by borrowing against (or selling or assigning a portion of) the
potential recovery.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE AD VERSAR Y SYSTEM? 335
ment is a repeat player,58 a unilateral account may be valuable in
analyzing the informational effects of measures that regulate the legal
advice received by government opponents. For example, a unilateral
account can help analyze the informational effects of forfeiture statutes
that deny suspected criminals the ability to use the alleged proceeds of
their offenses to hire expensive counsel.” We will therefore present both
unilateral and bilateral accounts.
E. Distributions of Access, Error, and Information
Our basic account adopts a number of further assumptions about
the distribution of informational access, litigation error, and favorable
versus unfavorable information. First, at the beginning of litigation, each
party has equal access to relevant information. To the extent that rele-
vant information is party-controlled or suppressible, it is equally con-
trolled or suppressible by each party. Second, the errors that unadvised
parties may make about litigation matters-such as how to categorize
information as favorable or unfavorable and about the level of eviden-
tiary sanctions-are not systematically biased. Parties are as likely to
mistake favorable information for unfavorable as vice versa and are as
likely to overestimate expected sanctions as underestimate them. Third,
in each set of information available to one or both sides, the relevant
information is equally divided between information favorable to one side
and information favorable to the other. In other words, the mix of
favorable and unfavorable information is fifty-fifty in each of the con-
trolled, discoverable, commonly available, and suppressible sets.
Obviously none of these assumptions of perfectly equal distribution
is likely to be true in any particular case. They may, however, accurately
capture overall distributions in a general class of litigation and thus pro-
vide an adequate basis for deriving the general effects of litigation advice.
We thus derive our initial conclusions about the general effects of advice
under assumptions of equal distribution. We will then examine in Part
III whether any of these distributions might be systemically skewed
either generally or in particular classes of litigation and what the conse-
quences of such skewed distributions might be.
II
THE BASIC ACCOUNT OF THE INFORMATIONAL EFFECTS OF
LITIGATION ADVICE
In this Part, we derive the informational effects of the legal advice
58. It may be impossible to effectively deny legal advice to government authorities, since as
repeat players they would develop legal sophistication whether or not they were represented by
lawyers.
59. See infra text accompanying notes 203-04.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
336 CALIFORNIA LAW REVIEW [Vol. 79:313
that is actually provided in an adversary system in which clients and law-
yers both face evidentiary sanctions and in which lawyer-client confiden-
tiality is imperfect. We postpone to Part V consideration of how
evidentiary sanctions applicable to parties and their lawyers can some-
times prevent more legal advice from being given. Thus we do not here
measure the informational effects of restrictions on advice, which would
require comparing a legal regime with advice and the legal restrictions
against a regime with advice but without the restrictions. Rather, this
Part measures the effects of legal advice against a baseline of parties liti-
gating with no advice.6
Even within our stylized basic account, the informational effects of
the legal advice provided to parties in an adversary system are diverse
and complicated. To clarify our exposition, we discuss separately the
effects of advice on different types of evidentiary conduct. In Section A,
we trace the informational effects of advice about how to categorize (as
irrelevant, favorable, or unfavorable) the information the parties have
identified. We subsequently consider the effects of legal advice on inves-
tigation in Section B and the effects of legal advice on the perceived and
actual sanctions for withholding or suppressing unfavorable information
in Section C. In each instance, we present the effects of unilateral litiga-
tion advice first and the effects of bilateral advice second.
Several informational effects of litigation advice are unambiguously
positive. First, legal advice helps parties sort the relevant information
from the irrelevant. Second, whenever the same information is available
to both parties (so that both can compete in presenting it to the tribunal),
legal advice to either party about how to categorize information as
favorable or unfavorable will normally increase the amount of favorable
information that the advised party presents more than it decreases the
amount of unfavorable information she presents. Third, legal advice
increases investigation, and thus the presentation of information that
would not have been identified without legal advice. Finally, for any
legal advice provided, lawyers’ personal susceptibility to evidentiary
sanctions will sometimes decrease withholding and suppression of unfa-
vorable information.
Other informational effects of advice, such as those flowing from
improved categorization of party-controlled and suppressible informa-
tion, are disturbingly ambiguous: they skew the mix of information in
favor of the party receiving advice, but they neither increase nor decrease
the total relevant information reaching the tribunal. But because the
unambiguously positive effects of advice are not offset by any clearly neg-
60. The distinction is important because, as Section V(B) demonstrates, a conclusion that
litigation advice is desirable, ambiguous, or undesirable does not alone dictate any conclusion about
whether confidentiality intended to encourage such advice is desirable or undesirable.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 337
ative effects, we conclude that in our stylized account litigation advice
has the following general effects. It decreases the proportion (and per-
haps total amount) of irrelevant information presented under both the
unilateral and bilateral accounts. Unilateral litigation advice increases
the presentation of information favorable to the party receiving advice
more than it decreases the presentation of unfavorable information, and
thus increases the total relevant information presented. Bilateral advice
increases the favorable and unfavorable information presented about
both sides.
A. Legal Advice That Helps Parties Categorize Information
In this Section, we assume that each party has a pool of potentially
relevant information that she has identified,6″ which is either nonsup-
pressible or suppressible.62 Unadvised parties may mistakenly categorize
such information in four ways: (1) irrelevant as relevant; (2) relevant as
irrelevant; (3) favorable as unfavorable; and (4) unfavorable as favorable.
Legal advice correcting these errors helps a party decide which informa-
tion to present, withhold, suppress, or save from suppression.
1. Effects of Unilateral Advice
a. Nonsuppressible Irrelevant Information
We consider nonsuppressible information first. Some nonsup-
pressible information that a party has identified will be irrelevant. With
legal advice, a party will be less likely to categorize such information
mistakenly as relevant and favorable. Correcting these errors reduces the
amount of irrelevant information presented.
b. Nonsuppressible Information Unadvised Parties Would
Miscategorize as Irrelevant
The remaining nonsuppressible information will either be favorable
or unfavorable. When legal advice corrects a party’s mistaken belief that
this nonsuppressible information is irrelevant, it increases that party’s
presentation of relevant information. Because lawyers are more likely to
present favorable information than unfavorable, the additional favorable
information presented out of the set that otherwise would have been mis-
61. This analysis applies only to information that a party has identified and can present.
Although normally a party should be able to present or compel the presentation of any potentially
relevant information she identifies, there can be information that a party can and does identify but
cannot present, such as demeanor information about a witness who is beyond the subpoena power of
the court. For purposes of analyzing advice about which information to present, such information is
analytically the same as information the party has not identified at all.
62. We discuss in Sections II(B) and II(C) the changes in the information identified by the
party or her opponent because of the effects of advice on investigation and on sanctions for
withholding and suppression.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
338 CALIFORNIA LAW REVIEW [Vol. 79:313
categorized as irrelevant (and consigned to oblivion) should outweigh the
additional unfavorable information mistakenly presented out of that set.
c. Correcting Mistakes About Which Side Nonsuppressible Information
Favors
Unilateral legal advice also reduces party errors about which side is
favored by nonsuppressible information that an unadvised party would
correctly categorize as relevant. These corrections increase the presenta-
tion of favorable information and decrease the presentation of unfavora-
ble information. The relative magnitudes of these effects depend on
whether the information has been identified by the advised party alone or
by both parties.
i. Separately Identified Nonsuppressible Information
Information identified by the advised party alone includes informa-
tion in the party’s controlled set, such as a party’s recollections concern-
ing her own mental state, as well as information in the discoverable and
commonly available sets that the opponent has not requested. We will
call such information “separately identified.” Because the opponent can-
not present separately identified information, whether it reaches the tri-
bunal will depend solely on the party’s decision whether to present it.63
Legal advice decreases the probability that a party will either withhold
favorable information or present unfavorable information. The increase
in favorable information presented will be the product of the change in
the probability of mistaken withholding of favorable information and the
amount of favorable information being considered for presentation.64
The decrease in unfavorable information presented will be the product of
the change in the probability of mistaken presentation of unfavorable
information and the amount of unfavorable information being considered
for presentation.
Given our simplifying assumptions, any decrease in the probability
of mistaken withholding will equal the decrease in the probability of mis-
taken presentation. We will see, after analyzing investigation, that there
is reason to believe that the mix of separately identified information being
considered for presentation tends to be more favorable than unfavora-
ble.65 This will further bolster our conclusion that legal advice about
information categorization increases the favorable information reaching
63. In effect, advice about whether to present this class of information is the only class of
advice that Kaplow and Shavell model. See Kaplow & Shavell, supra note 17, at 568-75.
64. We define the change in the probability of mistaken withholding or presentation as the
difference between the probability a piece of identified information will mistakenly be withheld (or
presented) without legal advice, and the probability it will mistakenly be withheld (or presented)
with advice.
65. See infra Section II(B)(2)(b).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 339
the tribunal more than it decreases unfavorable information. At this
juncture, however, we assume only that the mix of separately identified
information is equally favorable and unfavorable.66 On this working
assumption we conclude that the increase in favorable separately identi-
fied information presented because of legal advice about which side the
information favors should equal the decrease in unfavorable separately
identified information presented. The result is no net gain or loss of rele-
vant information reaching the tribunal, but a mix of information skewed
in favor of the advised party.
ii. Commonly Identified Nonsuppressible Information
We will call information “commonly identified” if both parties have
identified and correctly categorized the information as relevant. Advice
about whether commonly identified nonsuppressible information is
favorable or unfavorable will increase the information reaching the tribu-
nal. This effect is easiest to recognize in the limiting case where the
opponent has perfect legal knowledge and therefore presents all com-
monly identified information unfavorable to the party receiving advice.
Against such an opponent, providing a party with advice cannot decrease
the unfavorable commonly identified information presented since the
opponent presents it in any case; advice will instead tend only to increase
the favorable information presented.
The effect is, however, more general: whenever the party’s opponent
is more than minimally competent,67 unilateral advice will tend to
increase the commonly identified information presented. A minimally
competent opponent is more likely to present unfavorable information
(from the advised party’s perspective) than favorable information.68
Accordingly, the commonly identified information that the opponent will
not present (and whose presentation can thus be affected by the quality of
the decisions made by the advised party) will tend to contain more
favorable than unfavorable information.69 In selecting from commonly
66. This assumption is reasonable given our initial assumption that each side controls or can
suppress an equal amount of relevant information, and that each set of information available to
either party contains equal amounts of favorable and unfavorable information. See supra Section
I(E).
67. A minimally competent opponent is one who has knowledge sufficient to make his ability
to distinguish favorable from unfavorable information better than random. We expect that almost
all opponents will be minimally competent by this standard, whether or not they have retained a
lawyer. Even if there are isolated exceptions, we are describing the general effects of advice, and
therefore find it sufficient for that purpose to assume that in general opponents will tend to be better
than random selectors.
68. Throughout this Section, we use “favorable” to refer to information that benefits the
advised party, even if it actually is presented by her opponent (to whom it is unfavorable).
69. To express this mathematically, suppose that the set of commonly identified information
contains F items of information favorable to the party and U of unfavorable. The opponent has a
probability P of selecting information unfavorable to the party receiving advice, and a probability
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
340 CALIFORNIA LAW REVIEW [Vol. 79:313
identified information, the party thus must make more decisions about
whether to present favorable information that otherwise would not reach
the tribunal than about whether to present unfavorable information that
otherwise would not reach the tribunal. Legal advice that increases the
probability of correct decisions about whether to present information
should therefore increase the favorable information reaching the tribunal
more than it decreases the unfavorable information reaching the tribu-
nal.70 And as the extreme case illustrates, the more competent the oppo-
I-P of selecting information favorable to the party receiving advice. Thus after the opponent has
selected, the amount of information remaining for the party to select is (P)F and (1-P)(U). Where F
and U are equal, the amount of favorable information remaining after the opponent has selected will
exceed the amount of unfavorable information remaining whenever P > 1-P, that is, whenever P is
greater than .5, the condition of minimal competence.
70. We are grateful to Louis Kaplow for pointing out that this effect may be offset because an
unadvised party facing a more competent opponent might deduce that her best strategy is simply to
present all commonly identified information unpresented by the opponent, figuring that the
inferences she can draw from her opponent’s nonpresentation are better than the inferences she can
draw from her independent categorization. Suppose, for example, the opponent is 80% competent
in categorizing information, and the unadvised party is 60% competent. Assume also that the mix
of commonly identified information unpresented by the opponent contains 100 items. Given the
opponent’s competence, 80 of these items should be favorable to the unadvised party, and 20 should
be unfavorable. The unadvised party could thus present a mix of information that is 80% favorable
by blindly presenting all 100 items. This 80% success with blind presentation might sometimes be
more advantageous than a selective presentation in which she applies her own 60% competence.
Where this logic would lead an unadvised party to present blindly, advice will tend to make the
party’s presentation more selective, thus decreasing the party’s presentation of commonly identified
information.
However, although a strategy of unadvised blind presentation is sometimes theoretically
advantageous, often it is not. Any party who is more than minimally competent can, through
selective presentation, almost always make the mix of commonly identified information she presents
more favorable than it would be if she followed a strategy of blind presentation. (The one exception
is that selective presentation cannot improve the mix when the opponent has presented 100% of the
favorable information.) To continue the last example, if the unadvised party applies her 60%
competence to selecting among the information unpresented by her opponent, she will present 60%
of the favorable items (or 48 favorable items) and only 40% of unfavorable items (or 8 items).
Selective presentation will thus make the mix of information she presents 85.7% (48/56) favorable,
an improvement from blind presentation. Depending on the sanctioning function and the mix of
information presented by her opponent, an 85.7% favorable mix of 56 selected items can be more
advantageous than an 80% favorable mix of 100 blindly presented items. Where it is advantageous,
the unadvised party will be selective and present only 56 items of information. If her attorney is
80% competent, with advice she will instead present 64 favorable items and only 4 unfavorable
items. Advice will then increase the items she presents from 56 to 68.
Moreover, even when theoretically advantageous, the practical extent of the incentive for
unadvised blind presentation seems small because of various obstacles. First, the unadvised party
cannot use the strategy unless she has the second move, so that she can observe the opponent’s
selection first. Second, the strategy is so complex that it is doubtful it would occur often to an
unadvised party. Third, the strategy has no potential application unless the unadvised party believes
her opponent is more competent. Finally, and perhaps most important, the unadvised party may not
know (and may have little way of knowing) whether the information unpresented by her opponent
has been commonly identified as relevant. Without such knowledge, the unadvised party has little
reason to infer from the opponent’s nonpresentation that information is favorable. The opponent’s
nonpresentation may instead mean that the opponent categorized the information as irrelevant or
never identified it all.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 341
nent the more the increase in favorable information presented because of
advice will exceed the decrease in unfavorable information presented.
d. Suppressible Information
Some portion of the information that a party identifies will be sup-
pressible. Advice enabling parties to categorize suppressible information
as irrelevant, favorable, or unfavorable has at least one positive effect.
When a party mistakenly categorizes irrelevant suppressible information
as relevant, advice causes the party to reclassify as irrelevant some infor-
mation that she would mistakenly have categorized as favorable and
saved from suppression or presented. This reduces the amount of irrele-
vant information reaching the tribunal.
Other effects are more ambiguous. When a party mistakenly catego-
rizes relevant suppressible information as irrelevant, legal advice some-
times results in its recategorization as relevant and either favorable or
unfavorable. A party with advice is more likely to prevent opponent sup-
pression and to avoid her own mistaken suppression of favorable infor-
mation. This increases the total amount of favorable information
reaching the tribunal. The advised party is also more likely to suppress
from her opponent and to avoid her own mistaken presentation of unfa-
vorable information. This reduces the amount of unfavorable informa-
tion reaching the tribunal. Assuming as we do that mistakes in
categorizing favorable and unfavorable information occur with equal fre-
quency, and that the amounts of favorable and unfavorable suppressible
information are equal, such advice should have no impact on the total
amount of relevant information reaching the tribunal. For similar rea-
sons, advice correcting party errors about which side is favored by rele-
vant suppressible information increases the favorable information
reaching the tribunal and decreases the unfavorable, with no net impact
on the total amount of information presented. In both cases, though, the
mix of information is skewed in favor of the advised party.
e. Summary
Unilateral litigation advice about how to categorize information has
several positive effects. For both nonsuppressible and suppressible infor-
mation, such advice decreases the amount of irrelevant information
presented. For nonsuppressible information, advice that corrects mis-
taken beliefs that information is irrelevant increases the presentation of
both favorable and (less strongly) unfavorable information. For nonsup-
pressible commonly identified information, advice that corrects party
errors about which side is favored by the information tends to increase
the presentation of favorable information more than it decreases the pres-
entation of unfavorable information, with a net increase in total relevant
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
342 CALIFORNIA LAW REVIEW [Vol. 79:313
information presented. Advice about categorization also has some
ambiguous effects. For relevant suppressible information and (under our
working assumption) for nonsuppressible separately identified informa-
tion, advice causes an increase in the amount of favorable information
reaching the tribunal and an equivalent decrease in unfavorable informa-
tion. The overall effect of advice about categorization is therefore a
decrease in the presentation of irrelevant information and an increase in
favorable information greater than the decrease in unfavorable informa-
tion. This increases the total relevant information reaching the tribunal.
2. Effects of Bilateral Advice
In a bilateral account, some effects of advice remain ambiguous.
For relevant suppressible information and nonsuppressible separately
identified information, advice causes each party to present more
favorable and less unfavorable information, with no net effect on the total
amount of favorable information or unfavorable information reaching
the tribunal about either side. Other effects remain just as positive as in
the unilateral account. Providing advice to both parties decreases the
amount of irrelevant information each side presents, and advice cor-
recting mistaken judgments that relevant nonsuppressible information is
irrelevant increases the amount of favorable and unfavorable information
each side presents. Finally, bilateral advice actually enhances the posi-
tive effect of advice on the presentation of commonly identified nonsup-
pressible information. As in the unilateral account, legal advice
improves each party’s ability to present favorable information from this
set more than it improves her ability to reduce the presentation of unfa-
vorable information. The tendency for the increase in favorable informa-
tion presented to exceed the decrease in unfavorable will, however, be
greater than in the unilateral account because bilateral advice will
improve the opponent’s competence.71
In sum, when both parties receive advice about categorizing infor-
mation, each party presents less irrelevant and more relevant informa-
tion, because advice increases her presentation of favorable information
more than it decreases her presentation of unfavorable information.
Because information favorable to one party is unfavorable to the other,
the result of bilateral advice is an increase in the presentation of favorable
and unfavorable information about each party.
71. See supra notes 67-70 and accompanying text (discussing the relation between unilateral
effects and opponent competence). There may be an offsetting effect to the extent evidence comes in
aggregate lumps of favorable and unfavorable information rather than the discrete units we assume.
See supra note 32. Increased opponent competence may decrease a party’s incentive to present such
“lumpy” evidence, because it increases the likelihood that the opponent will recognize and present
(through advocacy) the unfavorable informational aspects of the evidence.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 343
B. Legal Advice That Alters Investigation
1. Effects of Unilateral Advice on Investigation
Legal advice helps parties conduct more and better investigations in
a variety of ways. First, advice helps parties design or conceive of a
wider range of feasible investigations. Lawyers better understand the full
range of potentially helpful information, the possible modes of inquiry,
and the art of framing investigations to maximize the expected value of
the response. Lawyers also better understand the significance for the
design of investigations of whatever information is already at hand. An
advised party should thus identify more useful persons to interview or
depose, more documents to demand, and more questions to pose in for-
mal or informal pretrial investigation or at trial.72 In consequence, an
advised party can effectively choose from more investigative options-
having both positive and negative expected value-than an unadvised
party. A skilled cross-examiner, for example, effectively expands a
party’s investigative options by conceiving of a broad range of follow-up
questions. To the extent that the additional feasible investigations identi-
fied through legal advice appear to have positive expected values, an
advised party will conduct some investigations she would not otherwise
have conducted.
Second, legal advice should also increase the amount of investiga-
tion performed by helping parties make more effective use of the fruits of
investigation. When an investigation is completed, an advised party will
make better decisions about how to categorize the information obtained,
whether to present, withhold or suppress it, and how to use that informa-
tion to pursue further investigation. For example, a party delegating the
conduct of a file search to her lawyer can expect to identify more relevant
documents and to categorize them more accurately. In consequence, a
party who expects to have advice about using the proceeds of investiga-
tion will anticipate more benefits and fewer adverse consequences from
each feasible investigation of which she is aware. The prospect of advice
therefore strengthens the incentive to investigate.73 Some investigations
that would have had positive expected values will look even better, and
others that would have had negative expected values will now appear
positive. A party with advice about using the fruits of investigation
72. To the extent that sanctions for failure to produce information in response to a formal or
informal demand are sufficiently high to induce production, such demands permit investigation in
the discoverable set. We defer until Section II(C) discussion of how advice influences the level of
expected sanctions for failing to respond to those demands for production that are made.
73. The incentive effect will be strongest when confidentiality is absolute or when the party can
obtain advice without sharing information with her lawyer, because in those cases the decision to
obtain legal advice, standing alone, cannot increase the risk that unfavorable information will reach
the tribunal.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
344 CALIFORNIA LAW REVIEW [Vol. 79:313
should thus conduct more investigations than one who does not expect to
have such advice.
Finally, legal advice improves a party’s accuracy in calculating the
expected value of a feasible investigation and thus improves her selection
of investigations. Advice reduces the probability that a party will
mistakenly forego investigations that have positive expected value or mis-
takenly conduct investigations that have negative expected value.74 For
example, a cross-examining lawyer improves the selection of investiga-
tions by abandoning a line of inquiry that would require her to divulge
unfavorable information. This final effect alters the mix of investigations
conducted, but, unlike the first two effects, may increase or decrease the
total number of investigations conducted.
The effects of advice on investigation uniformly tend to increase the
number and proportion of investigations conducted that have positive
expected value: a party will conceive of a greater number of feasible
investigations, more investigations in that expanded universe will have
positive value, and the party is more likely correctly to choose to conduct
those positive value investigations. Whether advice will increase the
number of investigations conducted that have negative expected value is
ambiguous. Legal advice about investigation selection eliminates some
investigations with negative expected value that an unadvised party
would mistakenly have conducted. But advice also causes a party to con-
ceive of additional feasible investigations with negative expected value
that an unadvised party would not have considered. Legal advice, not
being perfect, will not wholly eliminate error in deciding whether to con-
duct those additional negative expected value investigations, and there-
fore some will be conducted in error. Thus, while legal advice will tend
to reduce the proportion of negative expected value investigations con-
ducted, it may increase or decrease the total number conducted.
In general, advice that influences investigation should provide a
party with a larger pool of information to present. The additional posi-
tive expected value investigations conducted because of legal advice
should lead to the identification of more favorable information, more
unfavorable information that the other party has also obtained, and more
irrelevant information that provides helpful investigative leads. The
effects of advice on the identification of useless irrelevant information
and on the amount of additional unfavorable information made available
to the opponent because of investigation are less straightforward. As
argued above, some of the additional investigations conducted on
account of legal advice will actually have negative expected value and are
therefore likely to produce such untoward consequences. Moreover,
74. See supra notes 41-48 and accompanying text (distinguishing negative and positive value
searches).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 345
some additional investigations with positive expected values will involve
an unavoidable risk of adverse consequences.” For example, it may be
worth conducting discovery to identify critical documents even though it
is likely that the discovery will alert the opponent to unfavorable infor-
mation that he would not otherwise obtain. On the other hand, advice
that improves a party’s selection of investigations tends to reduce the
occasions on which a party identifies irrelevant information or causes
additional unfavorable information to reach the tribunal. The general
tendency of advice, then, is to increase the amount of favorable informa-
tion identified without any predictable increase or decrease in the identifi-
cation, gathering, or divulging of unfavorable information that the
opponent would not otherwise obtain.76
2. The Combined Effects of Unilateral Advice on Categorization and
Investigation
We are now in a position to describe the composite effects of unilat-
eral advice on the investigation and categorization of information, on the
temporary assumption that legal advice does not influence the perceived
level of evidentiary sanctions for either side.77
a. Information Whose Identification Is Influenced by Advice
Information whose identification is influenced by advice will either
be suppressible or nonsuppressible. For nonsuppressible information,
advice influencing investigation should strongly increase the presentation
of favorable information. Advice influencing investigation causes a party
to identify additional favorable information, and advice about categoriza-
tion increases the probability that the party will correctly present that
75. See supra notes 46-48 and accompanying text.
76. These effects of advice may be diminished, but not eliminated, by a feedback effect.
Distinct investigations with different positive or negative expected values can sometimes produce
some of the same favorable or unfavorable information. Accordingly, the expected value of one
investigation may depend on whether another investigation that duplicates some of its positive or
negative consequences is conducted or eliminated as a consequence of advice. An investigation
whose positive consequences are duplicated by a new investigation is less likely to be conducted; an
investigation whose negative consequences are duplicated by a new investigation is more likely to be
conducted. Because the additional investigations caused by advice have more positive than negative
consequences, the feedback effect of those investigations will be a reduction of investigative activity.
Similarly, an investigation whose positive consequences duplicate a search eliminated because of
advice is more likely to be conducted, while an investigation whose negative consequences duplicate
those of an eliminated investigation is less likely to be conducted. Since the investigations eliminated
by advice are likely to have more negative than positive consequences, the feedback effect of
investigations eliminated by legal advice will also tend to reduce the level of investigation. Because
these feedback effects are derived from the principal effects of advice, however, they should only
offset rather than eliminate those effects.
77. We relax this temporary assumption in Section II(C).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
346 CALIFORNIA LAW REVIEW [Vol. 79:313
additional information.78
The effect of advice on the presentation of unfavorable nonsup-
pressible information, in contrast, depends on the net result of conflicting
influences. If the negative consequences of additional investigations con-
ducted with legal advice outweigh the negative consequences of the
investigations eliminated by improved selection of investigations, then
the party may divulge, identify, or gather more unfavorable information
that the opponent would not otherwise obtain. This should increase the
amount of unfavorable information reaching the tribunal.79 If the
decrease in negative consequences from improved selection of investiga-
tions outweighs the increase in negative consequences due to increased
investigations, the nonsuppressible unfavorable information reaching the
tribunal will decrease. Any increase or decrease in unfavorable informa-
tion reaching the tribunal will, however, tend to be smaller than the
increase in presentation of favorable information. This is because advice
tends uniformly to increase the presentation of favorable information but
has offsetting effects on the presentation of unfavorable information. For
similar reasons, advice should reduce the proportion of irrelevant infor-
mation presented, though it may not decrease the amount of irrelevant
information presented.80 Advice thus increases the amount and propor-
tion of relevant nonsuppressible information reaching the tribunal.
For suppressible information, the composite effect of advice about
investigation and categorization is ambiguous: an increase in favorable
78. To the extent attorneys streamline the presentation of cases involving an abundance of
relevant information by selecting the most forceful evidence to present, legal advice might not,
strictly speaking, increase the quantity of favorable information presented. Even if streamlining does
decrease the quantity of information, however, the replacement of marginally relevant information
with information of stronger relevance should still improve the quality of information presented with
effects on adjudication similar to those of an increase in quantity. Thus, our use of the terms
“increase in information” or “more information” should be understood to encompass increases in
the quality as well as quantity of information presented.
79. Negative consequences of investigation can lead to an increase in presentation of
unfavorable information in several ways. First, during investigation the party may divulge
unfavorable information that the opponent would not otherwise have obtained. Second, the party
may identify or gather additional unfavorable information that the opponent would not obtain. The
party may then divulge that information during further investigation, produce it in response to a
discovery demand, or mistakenly present it to the tribunal. Third, the party may identify or gather
unfavorable information that the opponent has obtained but will mistakenly fail to present. If the
party then mistakenly presents it, the unfavorable information reaching the tribunal will increase.
80. The additional investigations conducted with legal advice will tend to increase
identification of irrelevant information, while advice about selection of investigations will tend to
reduce it. The result may be an increase or decrease in the amount of irrelevant information
identified. If additional irrelevant information is identified, some of it will, even with legal advice, be
mistakenly presented. The overall proportion of irrelevant information presented will certainly
decrease, but the absolute amount of irrelevant information presented as a consequence of advice
about investigation may increase. On the other hand, if advice reduces the amount of irrelevant
information identified as a consequence of investigation, then both the proportion and the amount of
irrelevant information presented will decrease.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 347
information reaching the tribunal and an equivalent decrease in unfavor-
able information. Party incentives to identify favorable information in
order to prevent its suppression are comparable to party incentives to
identify unfavorable information in order to suppress it. Legal advice
about investigation should therefore effect a comparable increase in the
party’s identification of both classes of information. An advised party
will thus: (1) rescue from suppression and present more favorable sup-
pressible information, and (2) suppress more unfavorable suppressible
information. Because we assume equal distributions of miscategoriza-
tions and of favorable and unfavorable suppressible information, the
increase in favorable suppressible information reaching the tribunal
should equal the decrease in unfavorable information.
b. Information a Party Would Identify With or Without Advice
A party will identify some information whether or not she has
received advice: presentation of this information will be unaffected by
advice that influences investigation. For such information, the effects of
advice are generally those described in Section II(A): a reduction in the
amount and proportion of irrelevant information, and an increase in
favorable information exceeding the decrease in unfavorable information.
However, the analysis in this Section suggests that the tendency of the
increase to exceed the decrease may even be stronger than described in
Section II(A). That Section proceeded on the assumption that nothing
could be said about the mix of separately identified information.”8 The
set of information a party would have identified with or without advice,
however, in part consists of information the party would have identified
through investigation whether or not she had legal advice. Because these
investigations will tend to have positive expected values, that information
should consist predominantly of favorable information. In general, the
information that would have been commonly identified with or without
legal advice is just as likely to be favorable as unfavorable. Accordingly,
the mix of information identified by only one party with or without legal
advice should be more favorable than unfavorable, and advice about
presentation should to that extent tend to increase the presentation of
favorable information more than it reduces the presentation of unfavora-
ble information.
c. Summary
Out of the set of information that would have been identified with or
without legal advice, the increase in favorable information exceeds the
decrease in unfavorable information. Out of the additional nonsup-
81. See supra note 66 and accompanying text.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
348 CALIFORNIA LAW REVIEW [Vol. 79:313
pressible information identified because of legal advice, favorable infor-
mation increases strongly and unfavorable information may weakly
increase or decrease. Out of the additional suppressible information
identified, the increase in favorable information equals the decrease in
unfavorable information. The combined result is that unilateral advice
about investigation and categorization increases the total favorable infor-
mation reaching the tribunal more than it decreases the total unfavorable
information.
3. Effects of Bilateral Advice
Providing legal advice to both parties enhances the positive effects of
legal advice on investigation. Each party’s own advice enables it to con-
duct more and better investigations. Thus, each should identify and
present more relevant nonsuppressible information. Conversely, whether
bilateral advice that influences investigation alters the amount of sup-
pressible information presented is wholly uncertain. With advice, each
party is better able to identify and prevent the suppression of information
favorable to her position, and to identify and suppress unfavorable infor-
mation. In the bilateral account, improved efforts to suppress are met by
improved efforts to prevent suppression. We can discern no basis for
predicting whether a battle between two advised parties leads to more or
less suppression than a battle between unadvised parties. The net effect
of bilateral legal advice on investigation and information selection is thus
to increase the favorable and unfavorable information about each party
reaching the tribunal. The proportion (and perhaps amount) of irrele-
vant information reaching the tribunal should also decrease.
C. The Effects of Legal Advice on Effective Evidentiary Sanctions
Successful withholding and suppression of unfavorable information
reduces the information reaching the tribunal. Whether a party or her
opponent will find it worthwhile to withhold or suppress information
that she has categorized as unfavorable will depend on the perceived
sanctions for doing so. Legal advice can affect those evidentiary sanc-
tions in three main ways. First, advice can correct misperceptions of
expected evidentiary sanctions. Second, legal advice can influence the
actual level of those sanctions. Finally, because in giving advice the
party’s lawyer sometimes becomes personally susceptible to evidentiary
sanctions, retaining a lawyer can increase the effective evidentiary sanc-
tions the client faces.
1. Effects of Unilateral Advice on Parties’ Evidentiary Sanctions
Legal advice correcting party misperceptions of expected sanctions
for withholding or suppression can increase or decrease the amount of
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 349
information reaching the tribunal. Unadvised parties may over- or
underestimate the scope or level of expected evidentiary sanctions.82 If a
party underestimates evidentiary sanctions, advice will reduce the
amount of information withheld or suppressed; if the party overestimates
sanctions, advice increases withholding or suppression. Since we assume
in our initial stylized account that parties are just as likely to make one
kind of error as the other, advice that corrects misperceptions of sanc-
tions has no general effect on the amount of unfavorable information
reaching the tribunal.83
Legal advice can also influence the level of actual evidentiary sanc-
tions by improving how a party handles information about instances of
withholding and suppression that may or do become the subject of the
“litigation within a litigation” described in Part I. In a unilateral
account, we expect that advice influencing the level of evidentiary sanc-
tions for the party or her opponent will systematically increase the pres-
entation of information favorable to the advice recipient and decrease the
presentation of unfavorable information.
Providing advice to a party should raise expected evidentiary sanc-
tions for her opponent84 because an advised party can identify and pres-
82. Of course, if expected sanctions were sufficiently high to induce the production of all
information the opponent demanded, legal advice on these topics would always lead a party to
produce any of that information. Obviously, evidentiary sanctions are not in fact that high; nor are
they likely ever to be that high because, in a world where tribunals cannot obtain perfect information
about evidentiary conduct, increasing sanctions for apparently undesirable withholding will have
two adverse effects. First, it will increase expected sanctions for truthful nonproduction. See
Shavell, Optimal Sanctions and the Incentive to Provide Evidence to Legal Tribunals, 9 INT’L REV.
LAW & ECON. 3, 4-5 (1989). For example, increasing sanctions for witnesses who appear to be
withholding information because they stare at the floor or stammer will result in higher evidentiary
sanctions for truthful witnesses who exhibit those behaviors. Second, it will increase sanctions for
desirable withholding and suppression that is mistaken for the undesirable variety. Withholding or
suppression might be regarded as desirable despite its adverse informational effects because it
protects privacy, prevents coercion, or promotes other desirable conduct in or out of court. The
fourth and fifth amendments, and various evidentiary privileges, protect the right to withhold or
suppress information in part for these very reasons. Thus, even if optimally chosen, evidentiary
sanctions can do no more than trade off underdeterrence of undesirable withholding and suppression
against overdeterrence of truthful nonproduction and desirable withholding and suppression. Cf id.
(modeling sanctions that would optimize underdeterrence of withholding and overdeterrence of
truthful nonproduction).
83. Legal advice also affects party estimates of the benefits of withholding and suppression.
We have already discussed the effects on presentation, withholding, or suppression of advice that
corrects party perceptions about which side is favored by information. In theory, advice may also
influence party perceptions of the benefits of withholding or suppression by correcting party
misperceptions of the extent to which information is unfavorable. Without legal advice, a party may
over- or underestimate how damaging unfavorable information will be: such advice accordingly
could increase or decrease the unfavorable information reaching the tribunal. Because this
additional marginal effect has no impact on our basic conclusions, we do not consider it further.
84. This increase in evidentiary sanctions may not be socially desirable, because sanctions will
also increase for socially desirable withholding or suppression and for socially desirable conduct
(such as telling the truth) that sometimes is not readily distinguishable from undesirable withholding
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
350 CALIFORNIA LAW REVIEW [Vol. 79:313
ent more secondary information indicating that her opponent is engaged
in improper withholding or suppression. In part, this advice operates
before the opponent makes a decision to withhold or suppress. An
advised party can better frame demands for information that increase
expected sanctions for failures to produce that information.85 Questions
framed with legal advice are, to illustrate, much more likely to expose an
evasive witness.86 In part, this advice operates after the opponent’s
claimed act of withholding or suppression has occurred. Should a dis-
pute arise about whether withholding or suppression has taken place, for
example, legal advice will enable a party to identify and present to the
tribunal more secondary information indicating that the opponent
engaged in improper withholding and less secondary information indicat-
ing that the opponent did not. Finally, some forms of advice operate
before and after the act. The simplest example is perhaps the most tell-
ing: to the extent an advised party identifies more information about the
disputed primary conduct, the odds of detecting an opponent’s lies
increase. These effects, taken together, suggest that the opponent will
experience an increase in expected sanctions for both withholding and
suppression and will therefore withhold or suppress less information
favorable to the party receiving advice. With legal advice about investi-
gation and presentation, the party is thus able to identify and present
more favorable information that the opponent fails to withhold or sup-
press, increasing the overall amount of favorable information presented
to the tribunal.
Providing a party with advice also decreases her own expected sanc-
tions for withholding and suppression.87 Legal advice given prior to an
act associated with a risk of evidentiary sanctions may enable a party to
commit that act in a way that minimizes the creation or disclosure of
or suppression. See supra note 82. The inquiry in this Article is limited to the informational effects
of advice and does not consider when and whether withholding or suppression might be desirable
despite its adverse informational effects.
85. An advised party should also be able to make more demands for production. One could
regard this as advice that increases the opponent’s sanctions for withholding (from nothing to
something). We have, however, chosen to categorize this as advice that increases investigation, see
supra note 72 and accompanying text, and thus, to avoid double-counting, we do not also treat it as
advice that influences the level of sanctions. The choice of categorization does not change the
substance of our conclusions. In either event, advice that increases the number of demands for
production that are made should increase the presentation of information from the opponent’s
discoverable set.
86. This will not always mean that the witness actually is being evasive or withholding
information. Sometimes expert questioning can make the witness appear to be withholding even
when she is not. Thus, legal advice can increase the likelihood that an opponent will suffer
evidentiary sanctions, whether or not she actually was withholding or suppressing information.
87. As with the increase in the opponent’s expected sanctions, this decrease in the party’s
expected sanctions may be undesirable (where the party engages in undesirable withholding or
suppression) or desirable (where the party truthfully said she did not have the information or
engaged in desirable withholding). See supra note 82.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 351
unfavorable secondary information about the act and maximizes the cre-
ation of favorable secondary information. For example, an advised party
who lies about a relevant fact may be less likely to include in her story
details that can be proven false or that are true but sound “too good to be
true.” A party faced with a question in discovery that calls for unfavora-
ble information but that can be evaded may give an incomplete or
nonresponsive answer. Or a party may purposefully overlook a memo-
randum within the scope of a demand for production but take pains to
document other steps suggesting that the party’s production was com-
plete and diligent. Advice thus reduces unfavorable secondary informa-
tion about the evidentiary act available to the opponent, increases the
favorable secondary information available for presentation, and therefore
reduces the likelihood that the act will be sanctioned. In addition, after
an evidentiary act, legal advice in any proceeding where the opponent
claims that improper withholding or suppression has occurred enables
the party, under the foregoing analysis, to present more favorable and
less unfavorable secondary information about its evidentiary conduct-
with the increase in favorable secondary information exceeding the
decrease in unfavorable secondary information.
Both advice about how to commit an act of withholding or suppres-
sion and advice about how to defend such an act reduce the unfavorable
secondary information about the evidentiary act reaching the tribunal
while increasing the favorable secondary information. Expected eviden-
tiary sanctions for the party’s withholding and suppression should there-
fore decrease. A party with advice is thus more likely to find it
worthwhile to withhold or suppress unfavorable information about pri-
mary conduct than a party without advice. The additional unfavorable
primary information withheld or suppressed will reduce the opponent’s
presentation of unfavorable primary information and decrease the unfa-
vorable primary information reaching the tribunal.
The cumulative effect of unilateral advice influencing the level of
evidentiary sanctions is thus ambiguous. It increases the favorable pri-
mary information presented and reduces the unfavorable primary infor-
mation presented. Depending on the relative magnitudes of this increase
and decrease, it may either increase or decrease the total amount of rele-
vant information reaching the tribunal.”88
88. This overall result does not necessarily change with differences in opponent competence.
As opponent competence increases, both the increase in opponent expected evidentiary sanctions
and the decrease in the party’s own expected evidentiary sanctions should be smaller, with offsetting
effects on the total amount of information that the parties find it in their interest to withhold or
suppress.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
352 CALIFORNIA LAW REVIEW [Vol. 79:313
2. Lawyers’ Personal Susceptibility to Evidentiary Sanctions
Sometimes legal advice will decrease the amount of information sup-
pressed or withheld because of evidentiary sanctions applicable to the
lawyer giving advice. Lawyers are sometimes subject to sanctions for
providing advice about withholding or suppressing information or for
failing to disclose unfavorable information to the tribunal. In addition to
criminal sanctions for subornation of perjury and obstruction of justice,
disciplinary rules in almost all American jurisdictions forbid lawyers
from knowingly assisting or counseling the criminal or fraudulent with-
holding or suppression of unfavorable information and require lawyers to
withdraw when they know that their clients are using advice for those
purposes.89 A lawyer is also obligated, in circumstances that vary from
jurisdiction to jurisdiction, to disclose physical evidence of crimes90 and
to disclose unfavorable information to the tribunal in order to rectify
withholding or suppression that results in the presentation of evidence
the lawyer knows is false.91 Withdrawal and disclosure, the two acts
required of a lawyer in order to avoid sanctions, can both severely harm
the party. In addition, public prosecutors and lawyers appearing in ex
parte proceedings must disclose exculpatory information to opponents in
circumstances where parties to civil litigation would ordinarily have no
obligation to do so.”
A lawyer’s exposure to sanctions typically depends in substantial
part on whether she is aware of information unfavorable to her client. A
lawyer may become aware of unfavorable information because a client
divulges it to her or because the client delegates to her the tasks of inves-
tigation and presentation. Not every party will allow her lawyer to
obtain unfavorable information. Sophisticated clients can sometimes
make fully informed decisions without sharing information.93 Others
may decline to share information because the advantages of better
informed decisions do not seem worth the increased risk of their lawyer’s
disclosure or withdrawal.94 But some parties will find that the expected
89. See, e.g., MODEL RULES, supra note 1, Rules 3.3, 3.4; MODEL CODE, supra note 2, DR 7-
102(A)(3)-(6).
90. In some states, a lawyer who obtains possession of physical evidence of a crime committed
by his client comes under an obligation to turn that information over to the prosecutor. See, e.g.,
People v. Meredith, 29 Cal. 3d 682, 686-87, 631 P.2d 46, 48-49, 175 Cal. Rptr. 612, 614-15 (1981);
Morrell v. State, 575 P.2d 1200, 1206-11 (Alaska 1978).
91. MODEL CODE, supra note 2, DR 7-102(B)(2); MODEL RULES, supra note 1, Rule 3.3(a)(4).
92. MODEL RULES, supra note 1, Rule 3.8(d) (prosecutor’s obligation); id. Rule 3.3(d)
(requiring the presentation of “all material facts” in an ex parte proceeding).
93. For example, a sophisticated client’s lawyer may, without learning the client’s information,
be able to describe in hypothetical terms what kinds of information are unfavorable, confident that
the client will be able to apply that knowledge to withhold or suppress information matching that
description without implicating her lawyer. See K. MANN, supra note 30, at 109-11.
94. Prudent sanction-optimizing lawyers may advise clients when to share information or
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 353
benefits of better informed decisions outweigh the expected costs of shar-
ing unfavorable information with their lawyer.95 Others will discover
that as a practical matter they cannot obtain advice about investigation
and presentation without delegating execution of those activities to a law-
yer, who as a result will become aware of information unfavorable to her
client.
The lawyer’s awareness of unfavorable information can be signifi-
cant when it would be in the party’s interest, given the harmfulness of the
information and the level of expected sanctions applicable to the party, to
withhold or suppress the information. In such cases, the lawyer’s per-
sonal susceptibility to sanctions will sometimes cause the lawyer to take
actions that prevent withholding or suppression. Sometimes the lawyer
will increase the client’s effective expected sanctions for withholding or
suppression.96 If the decision to withhold or suppress is made in consul-
tation with the lawyer, for example, the lawyer may insist that the party
refrain from withholding or suppressing the information. The lawyer
may back that insistence with a threat either to withdraw from represent-
ing the party if she commits evidentiary misconduct or to disclose the
misconduct or the unfavorable information to the tribunal.97 Sometimes
the lawyer may disclose the information on her own. When the party has
delegated the task of compliance to her lawyer, the lawyer’s differing
incentives may cause the lawyer, without consulting with the client, to
comply with the client’s obligations more fully than the client would
have done. In either case, the lawyer’s action will increase the amount of
information unfavorable to the party that reaches the tribunal.
delegate investigation by weighing the advantages of the lawyer knowing the information or
executing the investigation against the disadvantages of restricting the client’s ability to withhold or
suppress the shared or attorney-identified information. Where a high risk exists that the information
to be shared or identified is unfavorable and that any evidentiary misconduct by the attorney will be
detected, and a low probability exists that the opponent would obtain the information if the client
commits evidentiary misconduct, such lawyers may advise their clients against sharing information
or delegating investigation. See infra note 235 (describing an instance of this strategy).
95. Cf supra notes 54-59 and accompanying text (assuming that to some extent parties can
obtain legal knowledge only by hiring lawyers during litigation).
96. Obviously, there will be some lawyers who break the rules by actively assisting in unlawful
evidentiary conduct. See Berentson, Integrity Test: Five of Thirteen Lawyers Fail, AM. LAW., May
1980, at 15. The general effect we predict, however, requires only that some lawyers be deterred
from doing so and that their actions sometimes increase the unfavorable information presented by
their clients.
97. See generally Kraakman, Gatekeepers: The Anatomy of a Third-Party Enforcement
Strategy, 2 J.L. ECON. & ORG. 53, 53-56 & n.3, 58-60 (1986) (discussing and contrasting
enforcement strategies that rely on “whistleblowers” with strategies that impose a duty on
“gatekeepers” to disrupt misconduct by withholding their support to wrongdoers). Lawyers who are
uncomfortable forcing the production of information that their clients would find advantageous to
withhold or suppress may achieve the same results by exaggerating to their clients the adverse
consequences the clients might suffer by failing to comply. Cf Gordon, The Independence of
Lawyers, 68 B.U.L. REV. 1, 35 (1988) (“exercises of independence are as likely to be covert as overt,
secreted in the interstices of prudential counseling”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
354 CALIFORNIA LAW REVIEW [Vol. 79:313
The conclusion that evidentiary sanctions applicable to lawyers will
sometimes cause lawyers to act contrary to their clients’ interests is
highly plausible. Lawyers as a class are repeat players with a substantial
investment of human capital in their professions.”98 They also face
evidentiary sanctions that are different from and often proportionately
more stringent than those faced by their clients. In most instances, a
single case represents a small part of an attorney’s career and livelihood,
and the benefit the attorney derives from assisting her client in sanction-
able withholding or suppression will seem correspondingly low.99 The
expected costs of misconduct, in contrast, may loom very large even if
the risk of detection for any given instance of misconduct is low. A law-
yer who becomes involved in suborning perjury, destroying evidence, or
deliberately disobeying discovery requests risks her reputation, discipli-
nary proceedings, and in some cases her career.'” For a client, a win or
loss in a given case is likely to assume much greater importance, while
the effect of sanctions is more likely to be limited to the case at hand.
On some occasions, then, attorneys will conclude that their expected
evidentiary sanctions exceed their personal benefit from withholding or
suppressing information unfavorable to their client and will act contrary
to their clients’ interests.’01 We therefore expect that when a party
98. As Professor Kraakman notes, a wrongdoer has more difficulty corrupting a gatekeeper
when either (1) entry into the gatekeeping market requires a significant investment (which leaves the
gatekeeper more vulnerable to legal sanctions), or (2) the gatekeeper has a diversified client base
(which leaves the gatekeeper less vulnerable to client pressure). Kraakman, supra note 97, at 70-71.
99. To be sure, sometimes lawyers who become aware of information unfavorable to their
client may not experience or act upon the divergent sensitivity to sanctions that we describe. Some
lawyers work “in-house” as full-time employees. Many attorneys in larger firms do most of their
work for, and are beholden to, only one or two clients. R. NELSON, PARTNERS WITH POWER 250-
51 (1988). To the extent that the lawyers in these law firms collectively suffer sanctions for
evidentiary misconduct that benefits a firm client, the firm’s expanded client base should give the
firm incentives to discourage its attorneys from engaging in evidentiary misconduct. If, however, the
firm escapes collective sanctions for participation by its lawyers in wrongdoing, see Pavelic &
LeFlore v. Marvel Entertainment, 110 S. Ct. 456 (1989) (holding that sanctions under Fed. R. Civ.
P. 11 cannot be imposed on the firm of a lawyer who violated the rule), does not have a diverse client
base, or has an imperfect monitoring system, large firm lawyers with few clients should, compared to
other lawyers, experience less of a systemic divergence between the benefits and sanctions they
perceive and those perceived by their clients. Even in such cases, though, the lawyers’ benefit from
assisting withholding will be less than the full benefit to the parties, since the lawyers are not fully
identified with the clients’ interests.
100. More subtle sanctions may also operate. For example, lawyers who do not respond
diligently to discovery requests that cannot be evaded honestly may find themselves repaid in kind
when they make discovery requests in this or other cases.
101. This conclusion may seem at variance with recent accounts of cases in which attorneys
actively engaged in discovery abuse. See, e.g., Kiechel, supra note 44; Rhode, supra note 11, at 598-
99. But none of these anecdotal accounts (nor to our knowledge any systematic study) compares
discovery compliance when clients lack legal advice with compliance when clients have such advice.
Some evidence suggests that clients and attorneys often have the different incentive structure we
suggest. See, e.g., K. MANN, supra note 30, at 109-12, 117-22 (discussing elite white-collar criminal
defense lawyers); Landon, Clients, Colleagues, and Community: The Shaping of Zealous Advocacy in
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 355
receives legal advice, sanctions applicable to lawyers will sometimes
cause a party to fail to withhold or suppress information that the party
would have withheld or suppressed without legal advice.2″‘
3. The Cumulative Unilateral Effects on Effective Evidentiary
Sanctions
In our basic account, the cumulative effects of unilateral advice that
influences the withholding and suppression of information identified as
unfavorable are as follows. Given our assumption that parties do not
systematically over- or underestimate the level of sanctions, advice that
corrects misperceptions about the level of evidentiary sanctions will
neither increase nor decrease the amount of unfavorable information
reaching the tribunal.103 Advice that influences the actual level of evi-
dentiary sanctions has no predictable effect on the amount of relevant
information reaching the tribunal. It increases expected evidentiary
sanctions for the opponent of the party receiving advice, which results in
an increase of uncertain magnitude in the favorable information
presented; it also reduces expected evidentiary sanctions for the party
receiving advice, which causes a decrease of uncertain magnitude in the
Country Law Practice, 1985 AM. B. FOUND. RES. J. 81, 105-09 (discussing lawyers in smaller cities).
Such divergent incentives may be more common when there is a continuing relationship between the
lawyer and opposing counsel, or the lawyer and the tribunal, that increases the importance of the
lawyer’s reputation, and hence the significance of potential reputational sanctions for misconduct.
102. As we spell out in greater detail in Section V(B), this increase in the presentation of
information unfavorable to the party receiving advice is not without costs. For the party receiving
advice, the risk that the lawyer will act contrary to her interest will deter some information sharing
or investigatory delegation which would in turn facilitate advice that would increase the amount of
relevant information presented by each side. Thus, while providing advice to one party in a model
that takes account of lawyer sanctions will cause the presentation of unfavorable information that
would not have been presented if the party had no lawyer, the amount of relevant information
reaching the tribunal in a regime with legal advice and lawyer sanctions may or may not be greater
than the amount of information in a regime with legal advice and no lawyer sanctions.
We leave out here the effect of providing advice for one party on the evidentiary sanctions
applicable to the opposing party’s lawyer, if she has one. The opponent’s lawyer should anticipate
an increase in evidentiary sanctions because the party is better able to detect and prosecute
wrongdoing. But that increase in sanctions will also be accompanied by an increase in advice
deterred. Accordingly, increased evidentiary sanctions for the opponent’s lawyer may or may not
result in an increase in information reaching the tribunal.
103. Similarly, in cases where a party has decided to comply with an obligation to produce or
preserve information, legal advice about how to comply with the party’s obligation will have no
predictable effect on the amount of unfavorable information reaching the tribunal. Without advice a
party might mistakenly produce more or less information than called for. For example, unadvised
parties may lack the technical expertise necessary to identify every document covered by a demand
for all documents relevant to an issue. This may lead them to produce only the documents they can
identify as responsive-thus mistakenly withholding some unfavorable information that has been
demanded-or to take an “everything-and-the-kitchen-sink” approach and produce all documents
they have-thus mistakenly presenting some unfavorable information not called for. Legal advice
about technical compliance thus has no systemic tendency to increase or decrease the amount of
unfavorable information reaching the opponent or the tribunal.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
356 CALIFORNIA LAW REVIEW [Vol. 79:313
unfavorable information reaching the tribunal. Finally, lawyers’ diver-
gent susceptibility to sanctions has a positive effect. In some cases, sanc-
tions applicable to lawyers will cause the presentation of unfavorable
information that would have been withheld or suppressed without legal
advice. The net effect of unilateral advice thus appears to be an increase
in favorable information presented that tends to be larger than the
increase in the unfavorable information withheld or suppressed.
4. The Effects of Bilateral Advice
In a bilateral account, evidentiary sanctions applicable to lawyers
should sometimes cause each party to produce or preserve information
she would have withheld or suppressed without legal advice. This
increases both the favorable and unfavorable information about each
party reaching the tribunal.
The effects of bilateral advice on expected evidentiary sanctions
applicable to parties have less clear implications for the amount of infor-
mation about primary conduct that reaches the tribunal. In a unilateral
account, the party receiving advice experiences a decrease in expected
evidentiary sanctions and her opponent experiences an increase. In a
bilateral account, the superior knowledge and skills of the opposing
attorneys will tend to offset. But the outcome should not be a wash. The
prior Sections suggest that, just as for information about primary con-
duct, legal advice should increase the favorable secondary information
presented by each side about any claimed incident of withholding or sup-
pression more than it decreases unfavorable secondary information.
Bilateral advice should accordingly increase the presentation of favorable
and unfavorable secondary information about both parties’ evidentiary
conduct.
This increase in information about evidentiary conduct need not lead
to a uniform increase in expected sanctions for withholding or suppres-
sion by either side, and thus may not increase the information about pri-
mary conduct reaching the tribunal. It would tend to do so if evidentiary
sanctions were set on the assumption that all withholding or suppression
is socially undesirable. But in fact evidentiary sanctions are often set on
the assumption that some withholding and suppression is socially desira-
ble.”4 Increased presentation of secondary information should, how-
ever, generally increase expected sanctions for undesirable withholding
and suppression and decrease them for desirable withholding and sup-
pression, for reasons we spell out in Part IV.
For example, expected sanctions against the willful destruction of
documents should increase, with a concomitant increase in the amount of
104. See supra note 82.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSAR Y SYSTEM? 357
information reaching the tribunal. But the likelihood that the tribunal
will sustain appropriate assertions of doctor-patient privilege should also
increase, reducing the information reaching the tribunal. The overall
effect depends on whether instances of undesirable withholding or sup-
pression (like willful destruction of documents) that would be
underdeterred without bilateral advice are more or less common than
instances of desirable withholding or suppression (like appropriate claims
of privilege) that would be overdeterred without bilateral advice.
Because we cannot tell whether under- or overdeterrence is more com-
mon, we cannot predict whether bilateral advice that influences the
party’s expected sanctions for withholding or suppression will increase or
decrease the total amount of relevant information about primary conduct
reaching the tribunal.105
D. Summary of the Basic Account
Our basic account enables us to draw some initial conclusions about
the general informational effects of legal advice. In the basic unilateral
account, advice reduces the proportion of irrelevant information
presented. It also increases the presentation of information favorable to
the party receiving advice more than it reduces the presentation of infor-
mation unfavorable to that party. This initial conclusion reflects several
positive effects. Advice about how to categorize relevant commonly
identified information or information unadvised parties would miscat-
egorize as irrelevant increases the presentation of favorable information
strongly but has no comparable negative effect on the presentation of
unfavorable information. The same is true of advice that influences the
investigation of nonsuppressible information. In addition, evidentiary
sanctions applicable to lawyers sometimes cause the presentation of unfa-
vorable information that would not have been presented without legal
advice. Our initial conclusion also reflects ambiguous effects: advice
about how to categorize information that can be withheld or suppressed,
advice about investigations conducted to identify suppressible informa-
tion, and advice about how to assess or influence sanctions for withhold-
ing and suppression each tend to increase the favorable information
presented but to cause possibly equivalent decreases in unfavorable infor-
mation presented. On balance, though, advice increases the amount and
proportion of relevant information.
105. Another complication is that increased opponent competence increases the likelihood that
the opponent will recognize and present any unfavorable information a party does not withhold or
suppress. This has two conflicting effects. On the one hand, it means that a greater proportion of
the information that is not withheld or suppressed will reach the tribunal. This will tend to enhance
the informational increase from bilateral advice. On the other hand, it increases the benefits of
withholding and suppressing unfavorable information. This will tend to offset any increase in
expected evidentiary sanctions and reduce the information reaching the tribunal.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
358 CALIFORNIA LAW REVIEW [Vol. 79:313
In the basic bilateral account, the positive effects of advice are pre-
served and enhanced and some ambiguous effects appear muted. For
each party, the net increase in relevant information presented will be
greater than if she had been the only party to receive advice, because the
positive informational effect of advice about categorization is stronger
when the opponent is more competent.” Moreover, in a bilateral
account, there is no reason to predict that advice influencing investiga-
tions that identify suppressible information or advice influencing the level
of expected evidentiary sanctions will systematically increase or decrease
the presentation of favorable or unfavorable information about either
side’s primary conduct.107
This general account clarifies the serious limitations of the account
of legal advice recently offered by Professors Kaplow and Shavell.
Kaplow and Shavell present a model of advice about information selec-
tion and establish that within the model advice has ambiguous informa-
tional effects. They then assert, with some qualifications regarding
advice about investigation, that the resulting informational “analysis
applies more broadly, to virtually all the choices lawyers make on clients’
behalf.'”108 Their model is, however, too limited to sustain this sweeping
claim or to justify any general policy implications about how litigation
advice should be regulated.
First, because Kaplow and Shavell’s model assumes each party has
“perfect control” over its information,'” they ignore the possibility that
an opponent’s ability to present information favorable to his position
might limit the ability of a party with selection advice to prevent the
tribunal from receiving information unfavorable to her side. Second,
because their model assumes that all information is relevant,”0 they
ignore advice about relevance and thus the possibility that such selection
106. See supra notes 67-70 and accompanying text.
107. See supra text following note 80; supra note 105 and accompanying text.
108. Kaplow & Shavell, supra note 17, at 593.
109. See id. at 576. Kaplow and Shavell believe that the assumption of perfect control alters
only the “extent” and not the basic nature or direction of the effects they predict for advice about
how to categorize information as favorable or unfavorable. See id. Our analysis of commonly
identified information demonstrates that this belief is false. See supra Section II(A)(l)(c)(ii). Our
analysis of investigation also demonstrates that, even if one assumes perfect control, Kaplow and
Shavell are incorrect in believing that the nature and direction of the effects of categorization advice
are not affected by their assumption that lawyers have perfect knowledge. See Kaplow & Shavell,
supra note 17, at 578 & n.26. Because the effect of categorization advice on investigation can
increase a party’s identification of unfavorable information, see supra Section II(B)(2)(a), an
attorney’s erroneous presentation of some of this additional unfavorable information could
conceivably increase the unfavorable information reaching the tribunal.
110. See Kaplow & Shavell, supra note 17, at 577 n.25 (assuming all evidence is either
“definitely favorable or definitely unfavorable”). Our analysis demonstrates that accounting for
advice that corrects either mistaken beliefs that information is relevant or mistaken beliefs that
information is irrelevant has beneficial informational effects that Kaplow and Shavell do not predict.
See supra Sections II(A)(l)(a), (b).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 359
advice could improve the quality of the parties’ presentations. Third,
their model takes no account of the influence of advice on investigation
or the possibility that such advice might increase the information from
which each party selects.”1 Fourth, their assumption that parties with
advice exercise perfect control over their information excludes the influ-
ence of legal advice on the perceived and actual costs of withholding
information. That assumption, and the further assumption that lawyers
act with the sole aim of lowering their clients’ expected sanctions,112 also
eliminate any possibility that the lawyer’s personal susceptibility to sanc-
tions for improper litigation conduct might lead to the production of
additional information.
Our account shows that for all four aspects of litigation excluded
from Kaplow and Shavell’s model, advice has a general tendency to
increase the information reaching the tribunal. Kaplow and Shavell have
modeled the vices of competition between skilled partisans without mod-
eling any of its virtues: competitive selection of information, focused
presentation, improved investigation, and compelled production. In the
cases that best fit their model, such as those where parties have no obliga-
tion to produce information or freely conspire with their lawyers in
wrongful withholding, the social value of legal advice has long been
doubted, and properly so.113 Kaplow and Shavell’s demonstration that
within their tightly restricted focus the informational effects of advice are
formally ambiguous is an important advance. But given the incomplete
111. See Kaplow & Shavell, supra note 17, at 568 (assuming that the “sole function” of lawyers
is assisting parties in “selecting which evidence to present to a tribunal”). Discovery and fact-
investigation are the only areas in which Kaplow and Shavell acknowledge that their model is
“incomplete” in any way that might limit its general applicability. Id. at 594 n.67. Our analysis
demonstrates that the effects predicted under a more complete model that includes investigation do
in fact differ from those predicted under the Kaplow and Shavell model. See supra Section II(B).
112. See Kaplow & Shavell, supra note 17, at 583, 570 n.3, 577.
113. Consider, for example, a fictional example of legal advice that has engaged the attention of
many professional responsibility scholars: the “lecture” from the novel The Anatomy of a Murder.
The client, accused of premeditated murder, tells his attorney information about his mental state
during the killing that plainly indicates his guilt. R. TRAVER, THE ANATOMY OF A MURDER 30-33
(1958). The attorney responds by explaining how different testimony could establish a defense of
temporary insanity, and the client then “forgets” the incriminating information previously disclosed.
Id. at 44-49. Even Monroe Freedman, perhaps the most intemperate defender of zealous advocacy,
concedes that the conduct of the attorney in the “lecture” is socially undesirable and ethically
indefensible. M. FREEDMAN, supra note 9, at 73-74. The “lecture” fits Kaplow and Shavell’s model
well: only the defendant has direct access to information about his mental state, the relevance is
clear, little investigation is necessary to obtain the information, defendant’s sanctions for
withholding are low, and the attorney is apparently indifferent to the risk of sanctions for suborning
perjury. We account for such cases under the heading of advice about how to categorize separately
identified information that would have been identified with or without investigation, and that an
unadvised party would correctly categorize as relevant. See supra notes 63-66 and accompanying
text. As our account makes clear, however, these effects form only a small subset of the effects of
advice.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
360 CALIFORNIA LAW REVIEW [Vol. 79:313
and biased character of their model, it provides a shaky basis for analyz-
ing the overall informational effects of litigation advice or deriving policy
implications.
Our bilateral account seems particularly suited for analyzing the
classic account of civil litigation, which posits parties of roughly equal
competence who benefit equally from legal advice and who have rela-
tively equal access to information. An example demonstrates both how
our account might play out in practice and how the partial perspectives
of some critics and defenders of legal advice lead to inaccurate conclu-
sions. Consider a suit for breach of contract in which the plaintiff and
defendant can each present one eyewitness to the making of an alleged
oral contract. Doubters like Professors Kaplow and Shavell might focus
on the information about the disputed conversation that the parties con-
trol and would identify as relevant without advice. From this partial
perspective, the lawsuit appears to be simply a “swearing contest.” Bilat-
eral provision of advice will have no beneficial impact on the tribunal’s
ability to render judgment. Instead it will skew each party’s presentation
of its controlled information, with no change in the total amount of rele-
vant information reaching the tribunal.
A defender would point, however, to other features of the case that
cast legal advice in a more favorable light. Advice should increase both
parties’ presentation of nonsuppressible relevant information about the
transaction that would mistakenly be categorized as irrelevant. It will
also increase the presentation of information that is not party-controlled
or that can be obtained only through investigation. This additional infor-
mation should provide a better picture of the parties’ conversation, the
setting and circumstances in which the conversation occurred, the par-
ties’ relative sophistication and the history of their dealings with others,
and the custom or practice in the relevant market. The tribunal will thus
have a better baseline for evaluating the plausibility of the parties’ com-
peting claims about the alleged agreement.
The cumulative effect of bilateral advice thus emerges as an overlay
of ambiguous and positive consequences. At a contested trial, the dra-
matic prominence of live testimony by the two witnesses to the conversa-
tion may create the impression that ambiguous effects predominate. But
in reality even the witnesses’ testimony about the critical conversation
ought on balance to be more informative than it would have been with-
out advice, because each will present additional party-controlled infor-
mation about their interaction that they would have miscategorized as
irrelevant or missed in investigation. Moreover, because the parties’ joint
account of the background to the transaction will also be more informa-
tive, the tribunal will be in a better position to render judgment than if
the parties had no legal advice.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 361
Other civil litigation may be more amenable to analysis under the
unilateral account. Consider a case where one party can afford advice,
but the other cannot. The unilateral account allows us to consider the
effects of both a decision to deny advice to the party who can afford it
and a decision to subsidize advice for the party who cannot. Denying
advice to the first party will increase her expected sanctions for primary
conduct. But because legal advice increases the amount of favorable
information presented more than it decreases the amount of unfavorable
information presented, denying advice will also reduce the amount of rel-
evant information reaching the tribunal.
Providing advice to the party who cannot afford it will, on the other
hand, strongly increase the amount of relevant information reaching the
tribunal. An example is the claim brought by a “private attorney gen-
eral,” such as an antitrust, securities, or discrimination class action, or an
environmental enforcement suit. In such cases the retention of counsel is
sometimes made possible by the promise of an award of attorneys’ fees to
successful claimants.114 Providing legal advice to the plaintiffs will
increase the defendants’ expected sanctions for withholding and suppres-
sion, thus increasing the amount of information that is available to both
sides and not suppressed. It will also increase the amount of information
that the plaintiff will identify, gather, and present from the information
available to both sides.
III
THE GENERALITY OF THE BASIC ACCOUNT CONCLUSIONS
In this Part, we introduce various complications into our basic
account and examine the extent to which these complications alter our
initial conclusions. These complications fall into four general categories:
(1) factors relating to the costs of litigation and the possibility of settle-
ment; (2) parties and lawyers who are law-abiding or truth-telling; (3)
information that influences the tribunal in a socially undesirable way;
and (4) unequal distributions of information, access, or errors. All these
complications may modify our initial conclusions. Indeed, some help
identify special conditions under which the general effects of advice pre-
dicted in Part II will not occur. We conclude, however, that many of
these complications provide no affirmative reason to alter our initial
conclusions.
114. The private attorney general may otherwise be unable to raise the funds for legal advice,
not only because those she represents may be too poor or because their expected financial gain from
litigation may not merit the cost, see supra note 57, but also because large groups with low per capita
stakes face greater free rider problems in colledting funds, see M. OLSON, THE LOGIC OF
COLLECTIVE ACTION 33-36 (1965).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
362 CALIFORNIA LAW REVIEW [Vol. 79:313
A. Cost-Related Factors
In our basic account, we abstracted from the costs of legal advice by
setting them at zero. We also abstracted from settlement and other stra-
tegic aspects of litigation in assuming that parties sought only to optimize
sanctions and their own litigation costs. We reintroduce those complica-
tions here.
1. The Costs of Legal Advice
Lawyers charge fees. This may alter the behavior of parties or law-
yers. Taking party behavior first, a party’s incentive to engage in litiga-
tion conduct may decrease if legal fees make such conduct more
expensive. This would, for example, undermine our conclusion that legal
advice generally increases incentives to investigate. But no a priori rea-
son exists to believe that eliminating legal advice would lower litigation
expenses since, without lawyers, parties may simply substitute their own
more inefficient and costly efforts. 15 Furthermore, given this ambiguity
about whether legal advice increases litigation costs, abstracting from the
costs of legal advice does not bias our account in favor of the social desir-
ability of litigation advice. It does, however, require us to limit our con-
clusions, as we do, to the social desirability of the informational effects of
legal advice.
Legal fees may also alter attorney behavior. Specifically, fees may
create “agency problems” by giving attorneys financial incentives to con-
duct litigation in ways that do not maximize their clients’ interests.”16
An attorney paid by the hour, for example, may “leave no stone
unturned” in discovery even if some stones are not worth the cost of
turning them.17 An attorney with a contingent fee, on the other hand,
may have incentives to forego some litigation conduct whose benefits
exceed costs because she bears all the costs but must share the benefits
with the client.”‘ Given the conflicting incentives under the different fee
arrangements, it is ambiguous whether in theory legal fees generally
encourage or discourage litigation conduct. The implications of the
available empirical data are likewise ambiguous.119
115. Accord Kaplow & Shavell, supra note 17, at 600-01.
116. See D. ROSENTHAL, LAWYERS AND CLIENTS: WHO’S IN CHARGE? 96-99 (1974); Miller,
Some Agency Problems in Settlement, 16 J. LEGAL STUD. 189 (1987).
117. See Rhode, supra note 11, at 635 (observing that lawyers are happy to leave no stone
unturned “provided, of course, they can charge by the stone”).
118. R. POSNER, supra note 15, at 535. Both these incentives will decrease to the extent such
conduct affects the attorney’s reputation in the market for legal services. But because market
monitoring is inevitably costly and imperfect, some agency costs will always remain. Cf. Jensen &
Meckling, Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure, 3 J. FIN.
EcON. 305, 309 (1976) (discussing general theory of agency costs).
119. See Kritzer, Felstiner, Sarat & Trubek, The Impact of Fee Arrangement on Lawyer Effort,
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 363
2. Settlement and Opponent Litigation Costs
If settling (or abandoning) a lawsuit is possible, litigants maximizing
their self-interest will take into account the effect of their actions not only
on their litigation costs and expected primary and evidentiary sanctions
but also on their opponent’s litigation costs and the expected terms of
settlement.120 This does not, however, greatly change our analysis
because the terms of settlement are largely influenced by the parties’
expectations about how a trial would come out,”’21 which in turn are
based on the information identified (or expected to be identified) as rele-
vant and presentable. Parties thus have incentives to investigate, catego-
rize, withhold, and suppress information to “win” the settlement that are
similar to their incentives to engage in such conduct to win at trial.’22
The effect of legal advice on marshaling information for settlement (or
predicting the information likely to be identified)’23 should be similar to
19 LAW & Soc’Y REV. 251, 266-72 (1985) (showing that contingent fee lawyers devote at least as
many hours as hourly fee lawyers in some classes of cases).
120. Even where settlement is not possible, a party’s trial effort may still, because it affects the
likely outcome, affect the opponent’s effort. Parties looking only to optimize the trial outcome may
accordingly take into account the effect of their litigation conduct on the opponent’s level of effort.
See Cooter & Rubinfeld, supra note 14, at 1072-73. The effect of one party’s litigation efforts on the
other’s is, however, ambiguous. See id. at 1073. It thus provides no affirmative reason to deviate
from our initial conclusions.
121. See id. at 1076; cf. Kaplow & Shavell, supra note 17, at 594 n.70 (noting that the
informational effects of advice may affect the terms of settlement). At least in securities class
actions, however, there is evidence that factors such as plaintiff lawyer incentives, defendant risk-
aversion and agency costs, asymmetric litigation costs, and the availability of insurance can lead to
settlements that bear little relation to the merits. See Alexander, Do the Merits Matter? A Study of
Settlements in Securities Class Actions, 43 STAN. L. REV. 497 (1991). A critical factor in such cases
is that neither party can make a credible threat to proceed to adjudication on the merits, so that the
expected outcome at trial is not a relevant factor in determining the parties’ threat values. See id. at
524. In classes of litigation where similar conditions hold, the informational effects generated by
advice are hard to predict but it is difficult to see how they could be harmful (or helpful).
122. Indeed, the prospect of settlement increases incentives to investigate for unfavorable
information that the opponent may obtain, because such information may help a party decide
whether to enter settlement negotiations and what settlement offers to make or accept. This should
somewhat increase the likelihood that a party’s investigation will increase the unfavorable
information eventually presented.
The incentives to identify and present information in settlement are not, however, identical to
the incentives to identify and present information at trial. By presenting information to “win” the
settlement negotiation, the party forfeits the element of surprise if a trial occurs. Depending on the
likelihood of settlement and the relative value of the information in negotiation versus at trial, a
party may withhold some information in settlement negotiations in order to preserve the element of
surprise at trial. See Brazil, supra note 30, at 1317-18. To the extent such information is not party-
controlled or suppressible, however, bilateral legal advice should reduce the difference between the
information presented to win the settlement and the information presented at trial because it
increases the likelihood that the opponent will have access to that information already. See supra
Part II.
123. Because settlement is largely motivated by the desire to avoid the costs of identifying
information, settlements will often precede the identification of information. Nonetheless, in those
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
364 CALIFORNIA LAW REVIEW [Vol. 79:313
the effect of legal advice on marshaling information for trial.’24
To be sure, expectations about trial outcomes are not the only factor
influencing the settlement terms. The bargaining range for settlement
will also be influenced by the expected costs to each party of proceeding
to trial versus the costs of settling. Parties may, for example, request or
produce massive amounts of useless information to run up the oppo-
nent’s litigation costs and coerce a more favorable settlement.125 The
terms of settlement will also reflect how successful each party is in bar-
gaining for a larger share of the joint surplus that settlement creates.’26
Unilateral legal advice may well influence both the ability to inflict
costs that affect the limits of the bargaining range and the division of the
surplus from settlement in ways that favor the party receiving advice.
Because an advised party should be better able to disguise or defend the
sanctionable infliction of costs on her opponent, her expected sanctions
for such conduct will be lower. Her lawyer will also recognize more
strategies for inflicting costs on the opponent. At the same time, unilat-
eral advice will reduce the opponent’s ability to inflict costs on the party
receiving advice because the opponent’s expected sanctions for such con-
duct should increase. When the parties to litigation are prepared to
engage in the strategic infliction of costs, unilateral advice will thus
increase the expected value of continued litigation for the party receiving
advice and reduce it for the opponent, shifting the probable terms of set-
tlement in favor of the party receiving advice. Unilateral advice may also
influence the party’s ability to bargain for a larger share of the settlement
surplus, in part because lawyers have more experience negotiating settle-
ments.127 It thus appears likely that unilateral legal advice will result in
better settlement terms for the advised party than the effects of advice on
cases the terms of settlements will still be influenced by the information the litigants expect to
identify, and those expectations should vary depending on whether the party has advice.
124. Whether this has a desirable impact on actual or expected sanctions will depend in part on
whether the expected outcome that forms the basis of settlement reflects the social desirability of the
conduct giving rise to the dispute. If, as we argue in Part IV, the informational effects of advice
generally tend to increase the tribunal’s capacity to determine whether primary conduct was
desirable, settlements based on the expected outcome of litigation with legal advice ought to be
preferable to settlements based on the outcome of litigation without advice.
125. See Brazil, supra note 30, at 1313, 1319-22. Such efforts will not always be feasible since
they will also run up the party’s own litigation expenses. But where the increase in litigation costs is
asymmetric, or where a symmetric increase in litigation costs has asymmetric effects, a strategy of
inflicting litigation costs can be successful. Cf Hurwitz, Abuse of Governmental Processes, the First
Amendment, and the Boundaries of Noerr, 74 GEO. L.J. 65, 71 (1985) (noting that predatory
litigation can be successful if it imposes disproportionate costs on a rival or if the costs imposed have
a disproportionate effect because the smaller litigant’s cost per output is increased more).
126. See Cooter & Rubinfeld, supra note 14, at 1076, 1078-80.
127. The advised party’s skill in bargaining for a bigger share of the settlement surplus should
not, however, leave the opponent any worse off than his expected outcome at trial (determined by the
predicted informational effects) minus the costs of proceeding to trial. Otherwise he would not
settle.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 365
the presentation of information about primary conduct, standing alone,
would suggest.
Where the advice is bilateral, there is no reason to think that the
bargaining advantage will fall to one side or the other. Nor do we see
any a priori grounds for concluding that bilaterally advised parties will
be more likely than unadvised parties to engage in, threaten, or succumb
to such strategic efforts to raise the opponents’ litigation expenses. To
the extent such conduct is sanctionable, our analysis of evidentiary sanc-
tions in Section II(C) should apply. To the extent such conduct is not
sanctioned and both parties persist in it, it would seem to represent the
breakdown of what would be mutually advantageous bargaining to avoid
litigation expenses. As we will soon see, however, there is no reason to
believe that litigation involving legally advised parties is more likely to
result in such bargaining breakdowns than litigation involving unadvised
parties.128
A related question is what effect legal advice has on the likelihood
(as opposed to terms) of settlement. Under modern economic theories of
settlement, the likelihood of settlement is determined by (1) the likeli-
hood of a recognizable gain from settlement, and (2) the likelihood of
avoiding bargaining breakdowns over the distribution of that gain.129
Given these factors, we conclude that bilateral advice should increase the
likelihood of settlement.
A recognizable gain from settlement will exist whenever the parties’
estimates of the expected trial outcome are sufficiently close that the dif-
ference between them is exceeded by the net costs of litigation.’30 If, for
example, the parties expect the same outcome at trial, the gain from set-
tlement is the extra costs the parties would have to incur from litigating
instead of simply settling on that outcome. Bringing parties’ estimates of
trial outcomes closer together will thus increase the likelihood of a settle-
ment gain and the likelihood of settlement.131
128. See infra notes 133-34 and accompanying text.
129. Cooter & Rubinfeld, supra note 14, at 1075-80 (reviewing literature).
130. Id. at 1076. We define the net costs of litigation as the costs to both parties of litigation
minus the costs to both parties of settlement. Because the costs of litigation generally exceed the
costs of settlement, we assume the net expected costs of litigation are positive.
131. See R. POSNER, supra note 15, at 525. The statement in the text simplifies a rather
complex analysis. More precisely, bringing the estimates closer together will increase the likelihood
of a settlement gain where the plaintiff’s initial estimate of her expected winnings from trial exceeds
the defendant’s initial estimate of his expected losses. It will have an ambiguous effect when the
defendant’s initial estimate exceeds the plaintiff’s. This is because a settlement gain exists when (but
only when) the plaintiff’s estimate minus the defendant’s is less than the net cost of litigation. Id. at
524. Where the plaintiff’s estimate exceeds the defendant’s, converging estimates increase the
likelihood that their difference will be smaller than the net costs of litigation. This increases the
likelihood (and size) of a settlement gain. Where, however, the defendant’s estimate exceeds the
plaintiff’s, a settlement gain will always exist, even if the net costs of litigation equal zero. Thus,
bringing parties’ trial expectations closer cannot increase the likelihood of a settlement gain. It
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
366 CALIFORNIA LAW REVIEW [Vol. 79:313
Bilateral advice should bring trial outcome estimates closer together
for two reasons. First, legal advice should increase the discoverable or
commonly available information that both parties identify. This should
cause the parties’ perceptions of the disputed facts (and expected trial
outcomes) to converge.132 Second, with legal advice each party will more
accurately assess the information she expects to be identified and the
expected trial outcomes that will follow from the presentation of such
information. This will bring the parties’ estimates closer together even if
legal advice does not affect the information identified.
Assessing the likelihood of bargaining breakdowns is far more com-
plex. It turns on factors that are difficult to gauge, such as the parties’
knowledge of each other’s preferences and psychological factors relating
to the bargainers’ ability and willingness to make credible threats to
forego any gain unless their share is increased. Nonetheless, to the extent
lawyers are loyal agents and have a psychological makeup that is no
more prone to lead to bargaining breakdowns than that of the litigating
parties, then the fact that lawyers are repeat players should reduce bar-
gaining breakdowns. 33 Moreover, some bargaining models suggest that
increasing the information both sides identify will reduce the chances of
bargaining impasses.13″ Even if lawyers do not change the likelihood of
bargaining breakdowns, the more convergent estimates of trial outcome
should increase the likelihood of settlement by increasing the likelihood
of a settlement gain and thus the number of bargaining opportunities.
B. Different Motivations
Our basic account assumed that parties were interested solely in
optimizing their expected costs and sanctions.’35 We now consider how
our results would change if parties (and lawyers) were instead law-
should decrease the size of the settlement gain, but this has ambiguous effects on the likelihood of
settlement. Id. at 523.
132. Strategic problems will often prevent parties from voluntarily sharing their private
information with their opponents prior to settlement. Id. at 525; Sobel, supra note 29, at 153-54.
133. See Cooter, Marks & Mnookin, Bargaining in the Shadow of the Law: A Testable Model of
Strategic Behavior, 11 J. LEGAL STUD. 225, 232-33, 241 (1982). Attorney disloyalty may increase or
reduce the likelihood of settlement depending on how attorney compensation is determined. See
supra notes 116-19 and accompanying text.
134. See Sobel, supra note 29, at 133, 146.
135. It also excluded, for expositional convenience, the possibility that parties might be risk-
averse. Risk-aversion ordinarily increases the likelihood of settlement. See Cooter & Rubinfeld,
supra note 14, at 1076. It should also affect litigation conduct by increasing the deterrent effect of
any given schedule of evidentiary sanctions. See Polinsky & Shavell, The Optimal Tradeoff Between
the Probability and Magnitude ofFines, 69 AM. ECON. REV. 880, 884-85 (1979). This will, however,
alter the overall effect of legal advice on settlement or litigation conduct only if unadvised parties
would systematically under- or overestimate the harm from losing or suffering evidentiary sanctions.
On our assumption that unadvised parties are not systemically biased in their litigation errors, the
overall effect of litigation advice is not altered by risk-aversion.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 367
abiding or truth-telling in the sense defined in Section I(A).136
1. Uniform Party Motivations
If parties are uniformly truth-telling, the effects of legal advice will
be even more positive than described in Part II. Parties will not inten-
tionally suppress or withhold relevant information. Instead, each party
will present all the information she has identified as relevant, whether
favorable or unfavorable to herself, and withhold all the information she
has identified as irrelevant. Litigation advice will help parties identify
information through investigation and determine whether that informa-
tion is relevant. It will therefore increase the amount of favorable and
unfavorable information presented by each side and decrease the propor-
tion of irrelevant information presented.
If parties are uniformly law-abiding, then the effects predicted in
Part II still largely hold. They differ principally in the informational
effects of advice that influences perceived evidentiary sanctions. Such
advice may correct misperceptions about when withholding or sup-
pressing information is lawful. If law-abiding parties are advised that the
evidentiary conduct is unlawful, they will present more unfavorable
information regardless of the level of expected evidentiary sanctions. If
they are informed the conduct is lawful, they will present less unfavora-
ble information. Assuming again that they are not systemically biased
toward one sort of legal error-unadvised parties are as likely to underes-
timate the scope of unlawful conduct as overestimate it-the overall
effect is a wash. These results parallel those predicted in Section II(C)(1)
for sanction-optimizing parties.
Advice may also correct misperceptions about or alter the level of
expected sanctions. This will not affect the behavior of law-abiding per-
sons when they learn that expected sanctions for conduct they correctly
believed was or was not lawful are lower than they thought. (Even law-
ful conduct might carry expected sanctions because of the risk of trial
error.) Nor will it affect law-abiding persons who correctly believed the
conduct was unlawful and learn that expected sanctions are higher than
they thought. However, persons who correctly believed the conduct was
lawful, but who learn that expected sanctions are higher than they
thought, will produce more unfavorable information. The overall effect,
assuming no bias in legal error, is a mild increase in the amount of unfa-
vorable information reaching the tribunal. This increase was not pre-
dicted in Section II(C)(1).
136. Law-abiding and truth-telling persons will often differ in their willingness to provide
information to the tribunal. A law-abiding litigant, for example, will not produce information that
has not been requested and will take the fifth amendment when it is available. A truth-teller will
volunteer information that has not been requested and will never take the fifth amendment.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
368 CALIFORNIA LAW REVIEW [Vol. 79:313
The divergent susceptibility of lawyers to sanctions has substantially
less impact in cases involving truth-telling or law-abiding parties. This is
in part because truth-telling or law-abiding parties will view less of the
information available to them as party-controlled or suppressible. In
addition, lawyer sanctions for assisting in withholding or suppressing
information typically apply only where the client’s own withholding is
independently sanctionable. But this diminished impact is a direct conse-
quence of the client’s certain production of information that the lawyer’s
differential sensitivity to sanctions might or might not have produced.
Accordingly, eliminating this effect cannot reduce information presented
below the level that would occur if the parties were sanction-optimizing.
Moreover, a law-abiding party has much less reason than a sanc-
tion-optimizing party not to share information with her lawyer. This is
because sharing information can have effects a law-abiding party would
regard as adverse only in the rare case where withholding is lawful, the
lawyer would nonetheless face a positive risk of sanctions for assisting
withholding, and the lawyer’s expected sanctions are sufficiently high to
overcome her loyalty to the party. A sanction-optimizing party, by con-
trast, faces possible adverse consequences whenever the lawyer’s
expected sanctions for assisting withholding (lawful or not) are suffi-
ciently high to induce disloyalty. Because she has more incentive to
share information, it is likely that the law-abiding party will receive more
legal advice. This should enhance the positive effect of advice on investi-
gation and categorization.
The net result is that the nature of the informational effects pre-
dicted in Part II should not differ if parties are law-abiding instead of
sanction-optimizing. The extent of some effects may, however, differ in
ways that are ambiguous or positive.
2. Nonuniform Party Motivations
Legal advice can have far more disturbing effects if the motivations
of opposing parties differ. Take, for example, cases where a sanction-
optimizing party faces a truth-telling opponent. If both parties receive
advice, or if it is unilaterally provided to the sanction-optimizing party,
the decreased presentation of information unfavorable to the sanction-
optimizing party may well outweigh any increased presentation of infor-
mation favorable to that party. A truth-telling opponent will, despite his
ability to withhold or suppress some information, present all relevant
information he identifies, including all such information favorable to the
sanction-optimizing party. Providing legal advice to the sanction-opti-
mizing party will thus increase the presentation only of favorable infor-
mation the truth-telling opponent would not have identified and
categorized as relevant. The amount of information falling in this cate-
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 369
gory may not be trivial, particularly if the truth-teller’s investigations are
not very thorough or if the sanction-optimizer controls a large share of
the relevant information. Nonetheless, it will be smaller than the amount
of favorable information that would have been left unpresented if the
opponent were also sanction-optimizing. This will reduce the positive
effects of advice on the presentation of information favorable to the sanc-
tion-optimizing party.
The decrease in unfavorable information presented, on the other
hand, will be unaffected because truth-telling opponents will make an
effort to block the sanction-optimizing party from withholding or sup-
pressing unfavorable information similar to the effect a sanction-optimiz-
ing opponent would have made.’37 There is thus cause for concern that
providing advice to a sanction-optimizing party facing a truth-telling
opponent will decrease the presentation of unfavorable information more
than it increases the presentation of favorable information. That concern
will increase in proportion to the thoroughness of the truth-telling
opponent.
The results are similarly disturbing if a sanction-optimizing party
receives unilateral advice against a law-abiding opponent. The resulting
decrease in expected evidentiary sanctions for the sanction-optimizing
party will decrease the unfavorable information reaching the tribunal.
But the increase in expected evidentiary sanctions caused by advice
should have little effect on the law-abiding opponent. As we will see in
Section V(A), concerns about advice where one side is law-abiding or
truth-telling may underlie a number of rules regulating access to advice.
3. Lawyer Motivations
Our basic account also assumed that lawyers will optimize their cli-
ents’ sanctions unless the lawyers’ own exposure to evidentiary sanctions
becomes too great. If lawyers are instead truth-telling, they will disclose
to the tribunal any unfavorable information that they learn, moving the
effects of advice in the direction described for truth-telling parties. If
lawyers are law-abiding, then the differential susceptibility to sanctions
described in Section II(C)(2) will be greater, increasing the likelihood
that lawyers will disclose or force the disclosure of unfavorable informa-
tion. But sanction-optimizing parties represented by truth-telling or law-
abiding lawyers may share less information with their lawyers to reduce
the risk of lawyer disclosure. This may moderate or eliminate some of
the positive informational effects of advice that would occur if lawyers
137. Some truth-telling parties may be unwilling to withhold or suppress secondary information
about the opponent’s evidentiary conduct even when it would be lawful to do so and when doing so
would increase the evidentiary sanctions applicable to the opponent. Thus, truth-telling may
paradoxically increase the opponent’s withholding and suppression of information.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
370 CALIFORNIA LAW REVIEW [Vol. 79:313
were sanction-optimizing. The net effects will resemble those that flow
from subjecting lawyers to increased disclosure requirements, which is
discussed in Section V(B). Thus the substitution in our account of
law-abiding or truth-telling attorneys does not change the nature of the
informational effects described in Part II, but has an ambiguous impact
on the extent of those effects.
C. Information That Influences Tribunals in an Undesirable Way
In Section II(A), our basic account assumed that all information
having a favorable or unfavorable impact on the tribunal’s judgment was
actually probative of the disputed facts. Unfortunately, some informa-
tion reaching a tribunal may have a favorable or unfavorable impact that
is misleading. We here address the presentation of irrelevant, false, or
prejudicial information that might have such an effect.
A party might conclude that, because the relevant information likely
to be presented is extremely unfavorable, presenting irrelevant informa-
tion improves her chances by consuming the tribunal’s limited resources
and reducing its ability to evaluate relevant information. This strategic
presentation of irrelevant information might undermine our initial con-
clusion that legal advice reduces the proportion of irrelevant information
reaching the tribunal.
We doubt, however, that such strategic presentations are frequent or
that legal knowledge will increase their likelihood. The instances in
which purposefully presenting irrelevant information will be advanta-
geous seem rare. Such efforts will often be relatively easy to detect
because they take place before the tribunal and seem likely to anger or
frustrate the tribunal enough to provoke (usually inexplicit) evidentiary
sanctions. We suspect this threat of evidentiary sanctions will often out-
weigh the advantage of presenting information that, by definition, lacks
probative value.
There is also good reason to believe that legal advice will decrease
the overall likelihood that presenting irrelevant information will be
advantageous. First, the expected evidentiary sanctions for intentionally
presenting irrelevant information will likely be higher for a legally
advised party because the tribunal will assume that the party “ought to
know better” than to present such information.’38 Second, lawyers
138. In some cases, the tribunal may infer from the presentation of irrelevant information that
the party has no favorable information to present from those sets of information she controls. Legal
advice should reduce the presentation of irrelevant information in these cases. One might initially
regard this effect of advice as socially undesirable because it deprives the tribunal of information that
enables it to draw an inference about the merits of the party’s claim. But where the party presents
irrelevant information in the mistaken belief it was relevant, the tribunal’s inference seems inaccurate
and the presentation of the information socially undesirable.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 371
themselves face sanctions for intentionally presenting irrelevant informa-
tion. Their persuasiveness in future cases, for example, might be under-
mined if they gain a reputation for presenting irrelevant information.
Accordingly, the involvement of lawyers should create a differential sus-
ceptibility to sanctions. Third, because advice increases the presentation
of truthful relevant information, the tribunal’s ability to recognize and
sanction presentation of irrelevant information ought to increase.
Finally, insofar as there exists a separate set of truthful information that
tends to suggest that information is irrelevant, advice on balance should
increase the flow of such information to the tribunal.139
Even if legal advice did not decrease the likelihood that presenting
irrelevant information would be advantageous, our sense that such a
strategy would rarely be advantageous would still lead us to conclude
that legal advice decreases the amount of irrelevant information
presented. The reason is that collecting and presenting irrelevant infor-
mation involve significant resource costs that apply in all cases, including
those where such activities will plainly confer no benefit on the party.
We strongly suspect that the decrease in the proportion of irrelevant
information presented because of legal advice in the more common cases
where presenting irrelevant information is disadvantageous will outweigh
any increase in the strategic presentation of irrelevant information
because of legal advice in the rare cases where it is advantageous.
We next consider two other classes of information whose presenta-
tion influences trial outcomes in a socially undesirable way: information
that is false (whether or not fabricated) and information that appeals to
the prejudice or bias of the tribunal but has little or no probative weight.
To some extent, both parties and lawyers face sanctions for presenting
false or prejudicial information. The differential susceptibility of lawyers
to sanctions should thus operate to decrease the presentation of false or
prejudicial information. Furthermore, because legal advice increases the
presentation of true nonprejudicial information, whether favorable or
139. The particulars of the conclusions in the text differ for the unilateral and bilateral accounts.
In the unilateral account, the effect on the advised party’s expected sanctions for the strategic
presentation of irrelevant information is ambiguous. The party receiving legal advice will face higher
sanctions both because parties with lawyers suffer higher sanctions and because of her lawyers’
differential susceptibility to sanctions. On the other hand, legal advice should help the party present
secondary information that will lower the expected evidentiary sanctions for presenting irrelevant
information. This could increase or decrease the strategic presentation of irrelevant information by
the advised party. Unilateral advice should, however, increase the opponent’s sanctions for
presenting irrelevant information, and thus reduce the opponent’s strategic presentation of irrelevant
information. The net result is an overall reduction in strategic presentation.
In the bilateral account, expected evidentiary sanctions should increase for both sides. Both
will have lawyers and hence are both subject to higher sanctions and both will suffer from lawyers’
greater susceptibility to sanctions. Both will also face more information indicating that strategic
presentation of irrelevant information is taking place. This should reduce the likelihood that either
party can profit from presenting irrelevant information.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
372 CALIFORNIA LAW REVIEW [Vol. 79:313
unfavorable, legal advice should also increase the tribunal’s ability to
detect and sanction the presentation of false or prejudicial information.
But sanction-optimizing attorneys are also likely to identify more preju-
dicial information, think of more useful lies, and identify more persons
willing to lie. The overall effect of advice on the amount of false or preju-
dicial information presented is thus ambiguous in both the unilateral and
bilateral accounts.'” Unilateral advice should, however, ensure that the
mix of false or prejudicial information presented is more favorable and
less unfavorable to the party receiving advice.
D. Skewed Distributions of Information, Access, or Error
In order to reach initial conclusions about the general effects of legal
advice, our basic account assumed equal distributions of information,
access to information, and litigation error. We now reintroduce and
examine the extent and consequences of skewed distributions.
1. Biased Mixes of Information
The amount of information favorable to each party’s case will not
always equal the amount of unfavorable information. Some litigants
have violated the law or unjustly accused their opponent of a violation;
others (their opponents) have not. Given a rational system of evidence,
there should be more relevant information favorable to nonviolators and
just accusers than to violators and unjust accusers.
Ordinarily, we would not expect any general class of litigants receiv-
ing legal advice to systematically enjoy more favorable information than
their opponents.”’41 If, however, one could define a class of cases where
one side systematically has more unfavorable information than favorable
information, then advice to that side would have more dubious social
value. Our basic account’s conclusion about unilateral advice-that it
increases the favorable information presented more than it decreases the
unfavorable information-was based on the assumption that there was
just as much favorable information as unfavorable. If there is more unfa-
vorable information than favorable, then unilateral advice will decrease
140. If parties are truth-telling, they will never knowingly present false, prejudicial, or strategic
irrelevant information. Legal advice should help them avoid doing so by mistake and will thus
decrease the presentation of all three types of information. The same will be true for any advice
given if lawyers are truth-telling. If parties are law-abiding, they will present such information only
if they do not believe it to be unlawful. As Section III(B) suggests, the effect of advice in this
situation will be similar to the effect when parties are sanction-optimizing. If lawyers are law-
abiding, any legal advice given will reduce the unlawful presentation of false, prejudicial, or strategic
irrelevant information. See supra Section III(B)(3).
141. Although their analysis does not extend to cases that settle, Priest and Klein have
concluded, based on empirical data and theoretical modelling, that defendants and plaintiffs have a
50% probability of winning cases that go to trial. See Priest & Klein, The Selection of Disputes for
Litigation, 13 J. LEGAL STUD. 1, 5-6 (1984).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 373
presentation of unfavorable information more and increase favorable
information less.’42 The overall result may be that the decrease in unfa-
vorable information exceeds the increase in favorable information. In
contrast, advice to the opponent (who has more favorable information
than unfavorable) should be even more positive than described in the
basic account. 43
The possibility that unilateral advice to wrongdoers may reduce the
information reaching the tribunal is disturbing. But note that we can
reach these conclusions only by positing a definable class of cases where
we have good grounds for deciding-without adjudication-which side
should win.44 In those circumstances, it is not surprising that we get
bad results from providing advice to the party we know should lose. The
point of adjudication, however, is usually to establish which party is right
where that is not already clear at the outset. In such cases we do not
usually expect a systematic tendency for one side to have more favorable
information than unfavorable.145 Moreover, so long as the parties’ access
to information is equal, bilateral advice should still increase the favorable
and unfavorable information presented about each party even when the
mix of information available to the parties is skewed against one of them.
2. Inequalities of Access to Information
In many cases, one side may be able to control or suppress more
142. One can describe this in greater detail. All the unilateral effects described as ambiguous in
Part II will tend to reduce the information presented because (assuming constant error rates)
unadvised parties will make more erroneous decisions to present or not to suppress unfavorable
information than to withhold or suppress favorable information. By the same token, although the
differential effect of lawyer sanctions will remain, the other positive informational effects will tend to
be weaker or negative.
Advice about relevance will increase information, but less than described in Part II because the
amount of favorable information in the additional information classed as relevant will be lower.
Advice about relevant commonly identified information will have less of a tendency to increase the
information presented because, with the commonly identified information weighted to unfavorable
information, it takes a more competent opponent to create a mix of unpresented information that is
richer in favorable than in unfavorable information. Finally, advice about investigation will increase
information to a lesser extent because more possible investigations will be negative value searches
that advised parties will avoid, and because more of the positive value investigations will search for
unfavorable information that the opponent will obtain rather than for favorable information.
143. The tendency of the increase in favorable information to exceed the decrease in unfavorable
information will be even stronger because the ambiguous effects described at supra note 142 will tend
to lead to the presentation of more relevant information, and the positive effects will tend to be
stronger.
144. Those who believe that such grounds exist-for example those who believe that 90% of
criminal defendants are guilty-may oppose advice to the class of parties they believe should lose.
But they must realize that if that class of parties were unadvised their opponents might prosecute
defendants who are more likely to be innocent.
145. What we do expect is that the nonviolators and just accusers should generally experience a
larger improvement in expected sanctions from unilateral advice than should violators and unjust
accusers. This is discussed further in Part IV.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
374 CALIFORNIA LAW REVIEW [Vol. 79:313
information than the other. If we adjust the controlled sets that we
assumed had equal size in our basic account by increasing the size of one
party’s controlled set and decreasing the size of her opponent’s, the uni-
lateral effects of providing advice to the first party will be affected in an
ambiguous manner. With more information discoverable from the oppo-
nent, the increase in favorable information should be heightened; with
less information discoverable from the party, the decrease in unfavorable
information should be heightened as well. The unilateral effects of pro-
viding advice to the opponent, and the bilateral effects of providing
advice to both parties, will likewise be affected in an ambiguous manner
by the shift in control.’46 Naturally, the party with more control will
benefit more from the mix of information presented. But this makes the
general effects undesirable only if parties with extra control systemati-
cally tend to be violators or unjust accusers.
Asymmetries in the parties’ abilities to suppress information are
more disturbing. Suppose that we increase the information suppressible
by one party and decrease the information suppressible by the opponent.
This has a negative impact on unilateral advice to the first party because
it magnifies the power of that advice to decrease the presentation of unfa-
vorable suppressible information and lessens the significance of advice
that prevents the opponent from suppressing favorable information.
Because this negative effect may outweigh the other positive effects of
unilateral advice, we cannot predict with any confidence that unilateral
advice to parties who can suppress more information than their oppo-
nents will generally increase the presentation of relevant information.
We see no grounds for concluding that one side should generally be
able to control or suppress more information than the other. That ability
may, however, exist in special classes of litigation. In criminal cases, for
example, the fifth amendment and the special obligations of prosecutors
to disclose exculpatory information differentiate the disclosure obliga-
tions of prosecutors and criminal defendants. These rules may increase
the informational control of criminal defendants and decrease that of
prosecutors.
3. Skewed Distributions of Litigation Error
Skewed distributions of unadvised litigation error can alter the
results of our basic account.’47 We consider skewed estimates of eviden-
146. The shift in control will dampen both the increase in information favorable to the opponent
and the decrease in unfavorable information. Whether the increase exceeds the decrease, or vice
versa, depends largely upon whether the increase in one party’s controlled set exceeds the decrease in
the other’s. Unless the overall control by both parties combined is affected, the bilateral effects
should not change. An increase in overall control should decrease, and a decrease in overall control
should increase, the extent of the bilateral effects.
147. This Section considers skews in the types of litigation errors, not differences between
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 375
tiary sanctions first, and then consider skewed categorizations of infor-
mation as favorable or unfavorable.
a. Skewed Estimates of Expected Evidentiary Sanctions
Suppose that unadvised parties are more likely to overestimate
expected sanctions for withholding, suppressing, or falsifying evidence
than to underestimate them. Advice about the true level of expected
sanctions will then tend to decrease the presentation of probative infor-
mation and increase the presentation of falsehoods. Counsel may, for
example, cause criminal defendants not to give unfavorable testimony by
advising them that the fifth amendment will protect them from unfavora-
ble inferences the tribunal might draw about their failure to testify. If,
alternatively, unadvised parties are more likely to underestimate eviden-
tiary sanctions, advice about sanctions should increase the probative
information presented. Parties may, for example, be advised that their
efforts to mask unfavorable information will be regarded as suspicious
and will antagonize the tribunal enough to provoke a higher sentence or
other implicit evidentiary sanctions.
A systematic tendency to underestimate expected evidentiary sanc-
tions will thus heighten, and a tendency to overestimate will undermine,
the positive effects predicted in our basic account.148 This is because
advice about the level of expected sanctions will no longer have a neutral
effect. Positive effects should, however, still flow from lawyers’ differen-
tial susceptibility to sanction and from advice about relevance, investiga-
tion, and categorization of commonly identified information. Thus, even
when parties systematically tend to overestimate expected evidentiary
sanctions, the overall informational effect of litigation advice might still
be positive.
b. Skewed Categorization of Information
The distribution of litigation error about whether information is
favorable or unfavorable might also be skewed. Unadvised parties might
be more likely to mistake favorable information for unfavorable than vice
versa. We call such parties “unadvised pessimists” because they perceive
their chances of winning to be less than the facts merit. Alternatively,
parties’ propensity to make litigation errors. It thus does not consider classes of litigation where one
side is, if unadvised, more likely to make litigation errors than the other. As we explained in Section
I(D), the differential effects of litigation advice in such classes of litigation are those predicted by our
unilateral account.
148. Similar results follow, even if parties’ perceptions of expected sanctions are accurate, if risk
averse parties systematically overestimate (or underestimate) the severity of evidentiary sanctions or
if law-abiding parties systematically overestimate (or underestimate) the scope of sanctionable
conduct.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
376 CALIFORNIA LAW REVIEW [Vol. 79:313
unadvised parties might be more likely to miscategorize information as
favorable. Such parties are “unadvised optimists.”
When parties are unadvised pessimists, the positive informational
effects of legal advice will be heightened. This is because advice about
information categorization will be more likely to correct mistaken beliefs
that information is unfavorable than to correct mistaken beliefs that
information is favorable. This increases the tendency of unilateral advice
to cause an increase in favorable information presented that exceeds any
decrease in unfavorable information. Similarly, unadvised party opti-
mism tends to undermine the positive effects of advice about information
categorization by causing a relatively greater decrease in unfavorable
information and lesser increase in favorable information. The overall
effects, however, may continue to be positive because of the positive
effects flowing from advice about relevance, investigation, and categori-
zation of commonly identified information,149 and from the divergent
susceptibility of lawyers to sanctions.
Is there any reason to believe unadvised parties are systematically
more optimistic than pessimistic? One possible reason might be derived
from economic models of settlement, which predict that parties who are
“excessively optimistic” are less likely to settle and more likely to go to
trial. These models observe that parties have an incentive to settle when-
ever the expected value of judgment perceived by the plaintiff minus the
magnitude of the expected disvalue of judgment perceived by the defend-
ant is less than the joint surplus that can be created by avoiding the costs
and risks of litigation.150 Where parties are “excessively optimistic,” in
the sense that one side believes her chance of winning is greater than the
other side believes is his chance of losing, then the likelihood of settle-
ment is decreased.’51
Because these models address the choice (with or without advice)
149. If party optimism is strong enough, the overall effects of advice regarding commonly
identified information might be negative, but common identification does tend to offset the negative
effects of party optimism.
150. Cooter & Rubinfeld, supra note 14, at 1076.
151. See Bundy, Commentary on “Understanding Pennzoil v. Texaco'” Rational Bargaining
and Agency Problems, 75 VA. L. REV. 335, 337-38 (1989). By definition, each excessively optimistic
party faces another excessively optimistic party. This is because excessive optimism is defined as
believing one’s case is better than the opponent believes it to be. For similar reasons, each
excessively pessimistic party faces an excessively pessimistic opponent.
Note that because “excessive optimism” is defined in relation to the opponent’s expectations, it
does not necessarily correspond to optimism in the sense that we use it, which is defined in relation
to the actual facts. Say, for example, the actual expected trial award is $1000, an unadvised plaintiff
perceives the expected trial award to be $2000, and an unadvised defendant perceives the expected
trial award to be $3000. Such a plaintiff would be an unadvised optimist in the sense we mean if her
overestimation of the expected award results from thinking the mix of information is more favorable
than it actually is. She would, however, be excessively pessimistic under settlement models because
her opponent thinks the plaintiff’s chances are better than the plaintiff thinks they are.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 377
between settlement and trial, they are not directly applicable to the issue
whether unadvised parties entering litigation (whether or not they settle)
tend to believe their chances are better than their attorneys do.152 None-
theless, if the same analysis extended to failures to settle disputes without
incurring the costs of consulting an attorney, then one might predict that
those who consult attorneys will also tend to be excessively optimistic.’53
The extent to which the litigation settlement analysis can be
extended to preconsultation settlement is, however, far from certain. To
begin with, the degree of optimism necessary to make preconsultation
settlement unfeasible will be relatively small because the costs of merely
consulting an attorney will be lower than the costs of litigation and
because there is no risk in consulting an attorney comparable to the
uncertainty of judgment. More importantly, the benefits from consulting
an attorney do not involve solely improving the chances of winning a
judgment: they also include improving the party’s assessment of those
chances. Parties who appreciate their lack of legal knowledge should
realize that they are either being optimistic, in which case consulting a
lawyer may help them avoid litigation expenses, or that they are being
pessimistic, in which case consulting a lawyer will often improve the set-
tlement they exact. Parties pessimistic about the outcome thus have
incentives to consult an attorney that are not relevant to the decision to
settle after a lawyer has been hired.
Even if the analysis of litigation settlement were fully applicable to
preconsultation settlement, it would not compel the conclusion that par-
ties seeking litigation advice are likely to be systematically optimistic
152. The prediction that the parties going to trial will tend to be optimistic does not alter our
conclusions about the effects of litigation advice for two reasons. First, this prediction applies
whether or not the parties are legally advised. Unless parties are systematically more likely to be
optimistic (or pessimistic) with advice than without, then advice should be just as likely to make
parties less optimistic as less pessimistic. (As explained at supra note 131, however, decreasing the
degree of optimism increases the likelihood and size of a settlement gain; decreasing the degree of
pessimism does not decrease the likelihood of a settlement gain although it does decrease its size.)
Second, and more importantly, our conclusions do not require that cases go to trial. It is sufficient
for our purposes if the terms of settlement are influenced by the information collected (or expected to
be collected) with the aid of legal counsel.
153. The issue here is somewhat different from that addressed by various other economic
models that examine when plaintiffs assert legal claims. Those models do not take into account the
possibility that parties may settle before asserting a claim (let alone settle before consulting an
attorney). See Cooter & Rubinfeld, supra note 14, at 1082-84 (reviewing literature). Those models
do conclude that parties are more likely to bring suits when they are “optimistic” about their
chances at trial or settlement. Id. at 1082. But these models mean optimistic only in the sense that
the parties (correctly or incorrectly) believe the likely outcome will be good. They thus do not speak
to the issue whether parties consulting an attorney to bring suit would tend to be “unadvised
optimists” in the sense that they believe their chances are better than the facts merit. Nor do they
address whether such parties will be “excessively optimistic” in the sense that they are more
optimistic about their chances than their opponents, or the optimism (in any sense) of the party
against whom the legal claim is asserted.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
378 CALIFORNIA LAW REVIEW [Vol. 79:313
about whether their information is favorable. Such a conclusion would
both exaggerate the importance of excessive optimism as a cause of fail-
ures to settle and oversimplify the causes of excessive optimism.
Excessive optimism, standing alone, is neither a necessary nor a suf-
ficient condition for failures to settle. Cases where the parties agree on
the outcome or are excessively pessimistic might still be taken to lawyers
because one party attaches a higher value than her opponent to the out-
come at judgment, because settlement costs are high, or because bargain-
ing breaks down due to rational strategic behavior.”54 Conversely, cases
involving excessive optimism will often be resolved without consulting
lawyers when the costs of consultation are high, when one party puts a
premium on quick settlement, or when one party wants to avoid public
litigation or a potential judgment for reputational or precedential rea-
sons. Cases involving excessive optimism therefore may not be dispro-
portionately represented among those brought to lawyers.
In addition, when excessive optimism is present, it need not be the
result of a tendency to misclassify unfavorable information as favorable.
The causes of excessive optimism also include having different expecta-
tions about how much information (that a party correctly views as
favorable or unfavorable) will reach the tribunal or about the amount or
mix of favorable and unfavorable information required to cause judgment
in one’s favor. Moreover, a party who mistakenly believes that the mix
of information is less favorable than the facts merit (and is thus an unad-
vised pessimist) may be excessively optimistic as that term is used in set-
tlement models if her opponent mistakenly believes the mix of
information is even less favorable to the first party than she does.’55
Accordingly, it seems an unwarranted stretch to assume that parties who
seek litigation advice have a systemic tendency to misclassify unfavorable
information as favorable more frequently than they misclassify favorable
information as unfavorable.
Finally, even if the parties who actually consult attorneys do system-
atically tend to be unadvised optimists about whether their information is
favorable, it would not alter the deterrent effects of advice unless parties
knew, before they acted, whether they or their opponent will be optimis-
tic or pessimistic. To the extent that parties’ behavior is affected by the
prospect of litigation advice and that they are sufficiently sophisticated to
pay attention to this distinct subissue, it seems more probable to us that
parties will assess the likely effects of advice on the assumption that they
and their opponents are just as likely to make one kind of error as
another.
154. See Cooter, Marks & Mnookin, supra note 133, at 238-39, 246 (discussing strategic
behavior during settlement).
155. See supra note 151.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 379
E. Conclusion
The general prevalence of the conditions assumed in our basic
account of evidentiary advice ultimately presents a series of empirical
questions. On many of those questions there is no data and little pros-
pect of obtaining any.156 In resolving those issues, we may often have
little more to go on than intuitive judgments. Our own intuition is that
many assumptions in our basic account-such as equal distribution of
favorable and unfavorable information, equal control, and non-skewed
distributions of litigation error-match up fairly well with the conditions
in many classes of routine civil (and some criminal) litigation. However,
we cannot rule out the possibility that these assumptions are incorrect.
We also know that in some respects (including its abstraction from litiga-
tion costs, from agency problems, and from the presentation of false or
prejudicial information) our account abstracts from conditions that are
important in a very substantial number of cases.
Against this background, we think it significant that relaxing our
basic assumptions often does not change our basic account’s conclusions
about the positive informational effects of advice. Nonetheless, relaxing
our assumptions does point to potentially significant classes of cases
where advice can distort trial and settlement outcomes due to losses of
truthful information, increased presentation of false or prejudicial infor-
mation, or the strategic imposition of costs. The likelihood of these
adverse effects is greatest when advice is unilateral, and when the party
receiving advice is a sanction-optimizing violator or unjust accuser, can
control or suppress a large proportion of the relevant information, and
opposes a truth-teller.
As a matter of theory, then, our account confirms the sense of
defenders and critics of the adversary system that the beneficial conse-
quences of legal advice are generally more robust in cases where advice is
provided bilaterally rather than unilaterally. It also predicts why, in dif-
ferent contexts, advice can have either beneficial or detrimental effects.
The differences between defenders and critics of advice can, within this
framework, be traced to different assumptions or a focus on different
classes of cases. As a matter of practice, our account helps sort out the
cases where legal advice is likely to be undesirable and thus may aid in
formulating legal rules concerning access to advice. But before we get to
those issues, we first need to develop some theory of what informational
effects are socially desirable.
156. For example, we are not aware of any study that systematically examines the extent to
which parties to litigation (1) control relevant information, (2) over- or underestimate evidentiary
sanctions, or (3) are unduly optimistic or pessimistic about whether the information available for
presentation is favorable or unfavorable.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
380 CALIFORNIA LAW REVIEW [Vol. 79:313
IV
THE SOCIAL DESIRABILITY OF THE INFORMATIONAL
EFFECTS OF LITIGATION ADVICE
Whether the informational effects of litigation advice should be
regarded as socially desirable depends in part on the purpose of adjudica-
tion. One prominent account of civil and criminal adjudication empha-
sizes its role in encouraging actors to engage in desirable conduct and in
deterring them from engaging in undesirable conduct. Another account
emphasizes directly preventing harm, whether by incapacitating the dan-
gerous criminal or enjoining dangerous acts. A third focuses on the dis-
tribution of rewards or punishment in accord with need or desert:
compensatory theories of civil justice and retributive theories of criminal
justice have this character. Finally, some accounts of adjudication give
weight to the values of procedural fairness, party autonomy, dispute res-
olution, or party satisfaction with outcomes, independent of the substan-
tive policies that guide the imposition of sanctions.
A. Deterrence Theories
For deterrence theories, adjudicated outcomes are socially desirable
to the extent that expected sanctions encourage informed actors to act
desirably. Accordingly, the informational effects of advice are socially
desirable only if they improve the schedule of expected sanctions and
those improvements are communicated to actors who are deciding
whether to engage in potentially sanctionable conduct.’57
As we have seen, advice influences two basic features of the relevant
information reaching the tribunal: (1) the total amount of relevant infor-
mation presented; and (2) the extent to which that information is skewed
in favor of one party or the other. We here consider the social desirabil-
ity of the informational effects of litigation advice in three scenarios.
First, we consider advice that increases the amount of relevant informa-
tion presented without increasing skew.'” This corresponds to the initial
157. Kaplow and Shavell, supra note 17, at 588-90.
158. By “skew” we mean the extent to which the mix of information presented differs from the
mix of all relevant information about a case. The mix of information that would be presented by
unadvised parties may itself be skewed. Legal advice thus increases skew to the extent it pushes the
final mix of information presented further away from the mix of all relevant information.
To the extent violators and nonviolators are equally represented in a given class of litigation,
and unadvised parties do not themselves skew the overall information presented, advice has a general
effect on skew if it changes the presentation of favorable information systematically more or less than
the presentation of unfavorable information. This would not necessarily follow in a class of litigation
where one side is usually the wrongdoer. Say, for example, that the mix of information is
determined by percentages and that the relevant information is 75% unfavorable to defendants.
Assume also that unadvised parties do not skew the information presented. In those circumstances,
bilateral advice would not increase skew if it increases the presentation of favorable information
three times as much as it increases the presentation of unfavorable information. Indeed, increasing
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 381
conclusions reached about bilateral advice in our basic account. “9 It
also corresponds to the general effects of bilateral advice when various
complications are considered to the extent one agrees, as Part III sug-
gests, that most provide no reason to change the conclusions of the basic
account. Second, we consider advice that increases the information
presented, but also increases skew. This corresponds to the initial con-
clusions reached about unilateral advice in our basic account, and per-
haps to bilateral advice where some features of the litigation (such as
informational access) are skewed. Third, we consider advice that both
increases skew and decreases the amount of information presented. This
may correspond to advice for some special classes of litigation discussed
in Part III.”6o
1. Advice That Increases Information Without Increasing Skew
Advice that increases the information presented without increasing
skew improves the schedule of expected sanctions. In our basic account,
for example, bilateral advice reduces the proportion of irrelevant infor-
mation presented and increases the presentation of relevant information
favorable and unfavorable to each party.161 Under a theory of adjudica-
tion that emphasizes deterrence, this unskewed increase in favorable and
unfavorable information presented ought to improve the tribunal’s ability
to distinguish desirable from undesirable conduct. This in turn increases
expected sanctions for those who act undesirably at the same time that it
decreases sanctions for those who act desirably.
The behavioral benefits of increased accuracy will be realized to the
extent that those about to act (1) have some knowledge of which acts the
law deems sanctionable, and (2) perceive that tribunals will be more
accurate when parties have advice than when they do not. Actors who
consult lawyers before acting or who have had repeated contact with the
legal system should normally satisfy these conditions. We think these
conditions often will also be satisfied by persons who act on the basis of
an intuitive or commonsense understanding of the law and the legal sys-
tem.’62 True, popular perceptions of the accuracy of adjudication with
the favorable and unfavorable information equally would increase skew in such a case because it
would move the mix of information presented further away from the 75/25 mix of all relevant
information.
159. See supra Section II(D).
160. Logically, a fourth category might exist in which advice decreases information without
increasing skew. This might occur, for example, where there is bilateral advice, strong party control,
and habitual overestimation of evidentiary sanctions. In such cases, advice should decrease expected
sanctions for undesirable acts and increase them for desirable acts. Such cases, however, seem
sufficiently rare that we forego a more extended treatment.
161. Where unadvised parties skew the mix of information, bilateral advice may even decrease
skew.
162. Louis Kaplow has pointed out to us that to the extent parties know the substantive rules
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
382 CALIFORNIA LAW REVIEW [Vol. 79:313
legal advice are often harshly critical. But popular perceptions would
probably be even more negative in the less accurate system that would
result if legal advice were barred.
Thus, where actors correctly perceive, however vaguely, the con-
tours of sanctionability and would also perceive the greater inaccuracy
that would result without advice, their behavior should be improved by
advice that increases information without increasing skew. Moreover,
the greater accuracy made possible by such litigation advice should
encourage actors to obtain legal advice before they engage in potentially
sanctionable primary conduct because more accurate adjudication will
increase the expected costs of choosing sanctionable conduct.’63 The
social desirability of advice received by those who consult lawyers before
they engage in primary conduct should also improve, since litigation
advice will normally result in expected sanctions more closely matched
to the desirability or undesirability of the party’s contemplated
conduct.’64
The relationship between litigation advice and primary conduct
advice is thus symbiotic: by enhancing the accuracy of adjudication, liti-
gation advice increases incentives to seek primary conduct advice and
increases its social value; by communicating the accuracy of adjudication
resulting from litigation advice to those about to engage in primary con-
duct, primary conduct advice enhances the social value of litigation
advice. Cases will remain where the beneficial behavioral effects of litiga-
tion advice will not be realized because actors are poorly informed.
Those cases, however, do not demonstrate that litigation advice is not
socially desirable, since the behavior of those poorly informed actors will
be no worse than if there had been no litigation advice, while the behav-
ior of better informed actors will improve.
applicable to their primary conduct before they act, litigation advice will have somewhat less effect
on the information they select for presentation. Although this may limit the extent of the effect on
behavior, it will not alter its nature or direction. Parties who know what conduct is sanctionable
would still perceive that in a regime without advice, the evidence presented is less likely to reveal
what they actually did. The increased accuracy in an advised system stems from many sources other
than advice about what conduct is sanctionable. It also stems from advice to the party’s opponent,
from advice about what inferences the tribunal will draw from information, from advice about
relevance, from better investigation, and because a necessary cost of getting the advice needed to
succeed in the adversary system will be the revelation of more unfavorable information because of
the lawyer’s differential susceptibility to sanctions.
163. Cf Shavell, Legal Advice About Contemplated Acts, 17 J. LEGAL STUD. 123, 127-28 (1988)
(noting that increasing the benefits of avoiding unlawful conduct will increase a person’s incentive to
obtain advice).
164. Kaplow and Shavell appear to agree that more accurate adjudication will increase the
social value of advice about primary conduct. See Kaplow & Shavell, supra note 17, at 588-89 &
n.58. But their model of litigation advice leads them to conclude that it does not improve the
accuracy of adjudication. Id. at 589 & n.59.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 383
2. Advice That Increases Information and Skew
Advice that increases the amount of information presented but also
increases skew will improve expected sanctions for parties who benefit
from the increased skew. Unilateral advice in our basic account, for
example, increases skew by increasing the presentation of information
favorable to the advised party and decreasing the presentation of unfa-
vorable information.165 This improves expected sanctions for the party
receiving unilateral legal advice. Sometimes this will be desirable:
expected sanctions should decrease for those parties benefiting from the
skew who did not violate the law and expected outcomes should improve
for those who justly accuse others of violating the law. Sometimes this
will be undesirable: expected outcomes should improve for unjust accus-
ers and violators. For example, allowing surgeons to hire legal counsel in
defending against malpractice claims will desirably reduce expected sanc-
tions for competent surgery and undesirably reduce expected sanctions
for incompetent surgery. Allowing malpractice plaintiffs to hire legal
counsel will desirably increase expected sanctions for incompetent sur-
gery and undesirably increase expected sanctions for competent surgery.
Because the advice also increases the total amount of relevant infor-
mation presented, however, the increase in information favorable to the
parties benefited by the skew will generally tend to exceed any decrease
in unfavorable information.’66 For example, in the basic unilateral
account, the increase in information favorable to the advised party
exceeds the decrease in unfavorable information. Because favorable
information is relevant only if it makes the factual conclusion favorable
to the presenting party more probable than it would be without the evi-
dence,167 nonviolators or just accusers are by definition more likely to
have potentially presentable favorable information than violators or
unjust accusers. Competent surgeons are, for example, likely to have
more favorable information to present, and less unfavorable information
165. See supra Section II(D).
166. See id. Sometimes skew will be increased even though favorable and unfavorable
information are increased. For example, in a general class of cases where the information is 50%
favorable and 50% unfavorable, advice that increases the items of favorable information by 80 and
increases the items of unfavorable information by 30 increases skew. Within our basic account, the
effects of such advice could be described as partly bilateral and partly unilateral. (This advice would
correspond to cases where both parties get counsel but one side gets better counsel than the other.
See supra Section I(D).) The bilateral part might increase the items of both favorable and
unfavorable information by 50 items; the unilateral part might increase favorable by 30 items and
decrease unfavorable by 20 items. Such advice is usually even more likely to have the differential
effect on expected sanctions described in the text than advice that simply increases favorable
information more than it decreases unfavorable.
167. See FED. R. EVID. 401 (defining “relevant evidence” as “evidence having any tendency to
make the existence of any fact that is of consequence to the determination of the action more
probable or less probable than it would be without the evidence”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
384 CALIFORNIA LAW REVIEW [Vol. 79:313
to withhold or suppress, than incompetent surgeons. Plaintiffs suing
incompetent surgeons are likely to have more favorable information and
less unfavorable information than plaintiffs suing competent surgeons.
Litigation advice that increases information and skew should there-
fore generally benefit nonviolators and just accusers more than violators
and unjust accusers.168 This will tend to cause a differential change in
expected sanctions with socially desirable effects on behavior.’69
The differential effect will be clearest in classes of litigation where,
without advice, the level of expected sanctions would be roughly the
same for violators and nonviolators (or for accusers and just accusers).
In such classes of litigation, advice benefiting defendants, such as unilat-
eral advice in the basic account, will decrease expected sanctions for
nonviolators more than for violators.170 Competent surgeons will, for
example, experience a larger decrease in expected sanctions than incom-
petent ones. Advice benefiting accusers (plaintiffs or prosecutors) will
improve expected outcomes more for just accusers than for unjust accus-
ers, thus harming violators more than nonviolators. Incompetent sur-
geons will, for example, experience a larger increase in expected
sanctions than competent ones.
Because the improvement in expected outcomes for nonviolators (or
just accusers) will be greater than the improvement in expected outcomes
for violators (or unjust accusers), advice that causes a differential
increase or decrease in expected sanctions increases the probability that
actors choosing between desirable and undesirable acts that are both
associated with a risk of sanctions will choose the desirable acts. Sur-
geons should, for example, be induced to take greater care because the
difference between the expected sanctions for competent and incompe-
tent surgery will be greater if there is a differential increase or decrease in
expected sanctions that favors competent surgery.’77
In classes of litigation where, without advice, the level of expected
168. In contrast, because Kaplow and Shavell find it completely ambiguous whether unilateral
advice increases favorable information more or less than it decreases unfavorable information, they
conclude that it is a matter of “happenstance” whether the decrease in expected sanctions for
desirable acts outweighs the decrease in expected sanctions for undesirable acts. Kaplow & Shavell,
supra note 17, at 614.
169. The conclusion of social desirability assumes of course that the law defining violations
bears some positive correlation to the social desirability of the primary conduct. Where there is no
such correlation, any system of adjudication will produce undesirable results.
170. The size of the differential change in expected sanctions will, however, also vary with the
competence of the opponent of the party receiving advice. See supra Section II(A)(1)(c)(ii).
171. Say, for example, that the expected sanctions without advice are $2000 for both competent
and incompetent surgery. This schedule of expected sanctions means that the effective sanction for
acting incompetently instead of competently is $2000 minus $2000, or $0. If advice to surgeons
reduces expected sanctions to $1000 for competent surgery and $1500 for incompetent surgery, then
the effective sanction increases to $500. If advice to plaintiffs increases expected sanctions to $3000
for competent surgeons and $4000 for incompetent surgeons, then the effective sanction increases to
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 385
sanctions is higher for violators than for nonviolators, or lower for unjust
accusers than for just accusers, the tendency of advice to improve out-
comes more for nonviolators or just accusers may be offset if marginal
returns to legal advice diminish more sharply for nonviolators and just
accusers. Such diminishing marginal returns might exist, for example,
because the expected sanctions without advice for incompetent surgeons
exceed those of competent surgeons, meaning that the unadvised situa-
tion of incompetent surgeons has greater room for improvement. If there
is a sufficient difference between the marginal returns to advice for viola-
tors and nonviolators or for unjust and just accusers, advice may cause a
differential change in expected sanctions that benefits violators and
unjust accusers more than it benefits their counterparts.172
In the extreme case, where nonviolations are associated with a zero
risk of sanctions, advice to defendants will clearly have diminishing mar-
ginal returns. Such advice cannot lower expected sanctions for nonviola-
tors at all,’73 but somewhat lowers expected sanctions for violators. This
undesirably makes actors more likely to commit violations. In such an
extreme case, however, advice to accusers will be especially desirable.
The advice will increase the expected recovery of just accusers. But
unjust accusers will have a zero chance of recovery (since nonviolations
have a zero risk of sanctions) and thus cannot be helped by advice. This
means that advice to accusers will make actors less likely to commit vio-
lations.’74 Moreover, because just accusers should have more favorable
information than violators, the increase in expected sanctions violators
face because of advice to accusers should generally exceed the decrease in
expected sanctions violators enjoy because of advice to defendants. Thus
where unilateral advice may benefit either accusers or defendants, the
overall effect over the whole class of litigation will still be a differential
change in expected sanctions.175
$1000. The increase in effective sanctions resulting from either differential change in expected
sanctions should increase the likelihood surgeons will act competently instead of incompetently.
172. Suppose the damages imposed for malpractice are $100,000. Some surgical conduct may
be so clearly competent that it is subject to a very low risk of sanction (say 5%) even if the surgeon
lacks litigation advice. Assume also, however, that even with advice, the surgeon faces an
irreducible risk of sanctions (say 2.5%). Advice thus reduces expected sanctions for competent
surgery by $2500. On the other hand, an incompetent surgeon is subject to a 45% risk of sanctions
without advice and an irreducible 15% risk of sanctions with advice. Thus legal advice reduces the
expected sanction for incompetent surgery by $30,000.
173. Even if unadvised nonviolators run no risk of losing a suit, they should, if there is any
chance a suit will be brought, still receive some advantage from litigation advice in a system where
meritless suits can be sanctioned. In such systems, providing advice to the nonviolator should help
deter unjust accusations and thus lower the nonviolators’ exposure to litigation costs.
174. It will also make just accusers more likely to sue.
175. A similar analysis applies in the less likely case where nonviolations are associated with a
risk of sanctions but violations are not. Advice to accusers will be undesirable in such a class of
litigation because it will make actors more likely to commit violations. It also makes it more likely
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
386 CALIFORNIA LAW REVIEW [Vol. 79:313
It will sometimes be possible to eliminate the harmful effects of
advice that increases information and skew and still achieve levels of
expected sanctions that are superior to those that would have prevailed
in the absence of legal advice. If, over the relevant class of litigation,
such advice benefits nonviolators or just accusers more than it benefits
violators or unjust accusers, that can be exploited to improve deterrence
as long as the regime can appropriately adjust either the standard of
proof for finding liability or the actual sanction imposed when liability is
found. The standard of proof can be decreased, or the actual sanctions
increased, to offset the bad effects of advice on the expected sanctions of
violators, without completely offsetting the good effects of advice on the
expected sanctions of nonviolators. Similarly, the standard of proof can
be increased, or the actual sanctions decreased, to offset the bad effects of
advice on the expected outcomes of unjust accusers without completely
offsetting the good effects of advice on the expected outcomes of just
accusers.176
Adjusting the standard of proof might occur in either criminal or
civil cases. For example, if legal advice alters the mix of favorable and
unfavorable information criminal defendants present on the issue of guilt,
the tribunal might in theory effectively reduce the standard of proof for
finding advised defendants guilty by requiring less evidence to overcome
reasonable doubt. 77 The “lawyer discount” can be used to offset the
improved showing made by guilty defendants. If advice increases the
odds of acquitting innocent defendants more than it increases the odds of
that actors will be diverted from desirable activity that is associated with a risk of sanctions into
other less desirable activity that is not associated with a risk of sanctions. An increase in expected
sanctions for nonnegligent obstetrics, for example, may cause doctors to shift to other medical
specialties that are less needed but carry less risk of erroneous liability.
By the same token, advice to defendants in such a class of litigation is especially desirable. By
decreasing expected sanctions for nonviolators, without affecting sanctions for violators, it makes
actors less likely to commit violations. It also increases the likelihood that a party choosing between
a nonviolation associated with a risk of sanctions and a nonviolation that is not so associated will
choose the nonviolation that is more personally advantageous. A decrease in expected sanctions for
nonnegligent obstetrics, for example, will induce more people to shift from other lawful careers into
obstetrics.
Moreover, because nonviolators should have more favorable information than unjust accusers,
the decrease in expected sanctions nonviolators enjoy because of advice to defendants should
generally exceed the increase in expected sanctions nonviolators face because of advice to accusers.
Advice to defendants should decrease the expected sanctions for nonnegligent obstetrics more than
advice to plaintiffs increases those expected sanctions.
176. Kaplow and Shavell discuss the theoretical possibility of adjusting sanctions to offset the
informational effects of advice. Kaplow & Shavell, supra note 17, at 590-93. It is not clear, however,
why such adjustments would be worthwhile in their model given the ambiguous informational effects
of advice they predict.
177. In practice, however, there may be significant practical and moral objections to ad hoc
adjustments of standards of proof to take account of legal advice. See infra notes 183-84 and
accompanying text. Nonetheless, it is clear that such adjustments sometimes occur. See infra note
184.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 387
acquitting guilty defendants, more innocent defendants will be acquitted
under the revised standard, with no decrease in the number of guilty
defendants convicted. Alternatively, the discount could be used to offset
the larger benefits that accrue to innocent defendants, increasing the
number of guilty defendants punished without increasing the number of
innocent defendants punished. In civil cases, where the preponderance
of the evidence standard fixes the probability required to establish liabil-
ity at just over fifty percent, the tribunal can achieve a similar adjustment
by discounting the weight given to evidence produced by the parties ben-
efiting from the advice-caused skew.
Such adjustments may be feasible even where diminishing marginal
returns cause a greater reduction in the magnitude of expected sanctions
for violators than nonviolators, as long as the proportional decrease in
expected sanctions is greater for the nonviolators. Suppose the damage
award imposed after a finding of medical malpractice is fixed at
$100,000. Suppose further that unadvised competent surgeons have a
10% chance of liability, which can be reduced to 2.5% with advice.
Unadvised incompetent surgeons have a 20% chance of liability, which
can be reduced to 10% with advice. The expected sanctions for compe-
tent surgeons have thus decreased from $10,000 to $2500, or by $7500.
The expected sanctions for incompetent surgeons have decreased from
$20,000 to $10,000, or by $10,000, a greater amount. Advice to defend-
ants could, however, be coupled with an adjustment that makes the stan-
dard of proof twice as easy for plaintiffs to meet. In that case, the
resulting expected sanctions would be $5000 for competent surgeons
(2 times 2.5% times $100,000) and $20,000 for incompetent surgeons
(2 times 10% times $100,000). The net result would be a decrease in
expected sanctions for the competent without a decrease in expected
sanctions for the incompetent. Making the standard of proof four times
as easy for plaintiffs would increase expected sanctions for the incompe-
tent without increasing them for the competent.178
The second strategy for benefiting from the differential change in
expected sanctions involves adjusting actual sanctions. Where the advice
causes a differential decrease in the expected sanctions of defendants,
actual sanctions can be increased just enough to offset the decrease in
expected sanctions for violators without offsetting the decrease for nonvi-
olators. This combination of advice and sanction adjustment leaves the
world better off by decreasing deterrence of desirable activity without
178. However, if advice benefits only defendants and, because of diminishing marginal returns,
decreases the magnitude and proportion of expected sanctions for violators more than for the
nonviolators, then no adjustment can eliminate the harmful effect of advice on deterring violations
without increasing the expected sanctions of nonviolators. A similar problem results if advice has
these sorts of diminishing marginal returns and benefits only accusers.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
388 CALIFORNIA LAW REVIEW [Vol. 79:313
lessening deterrence of undesirable activity. Alternatively, actual sanc-
tions might be increased enough to offset fully the decrease in expected
sanctions for nonviolators, leaving the world better off by increasing
deterrence of undesirable activity at no cost to desirable activity.
Similarly, where the advice aids just accusers, actual sanctions could be
decreased just enough to offset the improvement in outcomes for unjust
accusers (increasing deterrence of undesirable activity at no cost to desir-
able activity) or decreased enough to offset fully the improvement in
expected outcomes for just accusers (decreasing deterrence of desirable
activity without lessening deterrence of undesirable activity). Or, for
either form of advice, sanctions could be adjusted to achieve any result in
between: increasing deterrence somewhat for undesirable acts and
decreasing it somewhat for desirable acts. Although different strategies
might be optimal in different circumstances, all result in a social state
that is preferable to the social states possible without the differential
change in expected sanctions made possible by litigation advice that
increases information and skew.”‘ As with adjusting the standard of
proof, a strategy of adjusting actual sanctions may be feasible even with
diminishing marginal returns as long as the proportional decrease in the
expected sanctions is greater for the nonviolators than the violators.
A strategy of adjusting liability standards or actual sanctions to off-
set the adverse skewing effects of litigation advice does, however, face
other obstacles. Risk-aversion may be a problem. If, for example, those
who commit desirable acts associated with a risk of sanctions are system-
atically more risk-averse (with legal advice) than those who commit
undesirable acts associated with a risk of sanctions, then increasing
actual sanctions will increase deterrence more for desirable actors than
for undesirable actors.1s8 This will tend to offset the differential change
in deterrence levels resulting from advice that reduces expected sanctions
more for desirable acts than for undesirable acts. Society can, however,
still benefit by coupling a differential change in expected sanctions with
179. Efforts to adjust sanctions might be complicated by two factors considered in Part III.
First, advice can improve effective expected sanctions not only by improving the information
presented, but also by improving the party’s bargaining for the distribution of the settlement surplus.
See supra notes 126-27 and accompanying text. But since this affects all advised parties, it does not
alter the tendency of advice to aid nonviolators and just accusers more than violators and unjust
accusers. Second, advice can improve expected sanctions by improving the mix of false or
prejudicial information presented. See supra Section III(D). This is more troubling because one
might well expect violators and unjust accusers to be more likely to present false or prejudicial
information. But we doubt that this will have an impact substantial enough to alter the overall
differential change in expected sanctions that we predict.
180. Cf Polinsky & Shavell, supra note 135, at 880 (discussing the effects of risk-aversion).
Similarly, if desirable actors are systematically less risk-averse than undesirable actors, then
decreasing actual sanctions will decrease deterrence more for undesirable actors than for desirable
ones.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 389
an increase in actual sanctions, even if undesirable actors are systemically
less risk-averse than desirable actors, as long as the systemic difference in
risk-bearing costs is not substantial enough to outweigh the differential
change in sanctions. 181
Another problem is that adjustments in actual sanctions or stan-
dards of proof may have a feedback effect on the level of effort parties
exert to identify and present information. Decreasing actual sanctions
may, for example, decrease the trial efforts of defendants by making them
less valuable.182 Because the trial efforts of nonviolators should be more
effective than those of violators, this reduction in effort should dampen
the improvement in the tribunal’s ability to distinguish violators from
nonviolators and thus reduce the effective differential change in expected
sanctions. We doubt, however, that this feedback effect from adjusting
sanctions will be sufficient to offset the differential change in expected
sanctions that provoked the sanction adjustment. Moreover, where the
adjustments call for increased sanctions, trial effort should increase and
thus enhance the differential change in expected sanctions. This will tend
to offset the adverse effect increasing sanctions has on risk-bearing costs
where nonviolators are systematically more risk-averse than violators.
Special problems may attend strategies that depend on ad hoc
adjustments of sanctions to take account of differences in the quality of
the parties’ legal advice. Tribunals may not be able to recognize the
appropriate level of adjustment in every case.183 In addition, ad hoc
adjustments may conflict with the ideal of a neutral tribunal.184 But
181. In those circumstances, actual sanctions can simply be increased to achieve both (1) the
same deterrence level for violators (taking into account both changes in expected sanctions and the
increased risk-bearing costs caused by increasing actual sanctions) as would exist if litigation advice
did not cause the differential reduction in expected sanctions, and (2) a lower deterrence level for
nonviolators. Cf id. (noting that the “possibility of risk aversion does not imply that individuals
cannot be induced to make the same decision about engaging in the activity as they would” under a
risk neutral model of expected sanctions). Adjusting sanctions to compensate for unilateral advice
aiding defendants will be even more feasible if undesirable actors are more risk-averse than desirable
actors or do not have systemically lower levels of risk-aversion. But where the risk-aversion of
desirable actors is systematically higher than that of undesirable actors, and the difference in risk-
bearing costs is sufficiently large, then increasing actual sanctions will not be a feasible strategy for
offsetting the effect of litigation advice on expected sanctions.
182. See Rubinfeld & Sappington, Efficient awards and standards of proof in judicial
proceedings, 18 RAND J. ECON. 308, 313 (1987).
183. Tribunals may have considerable difficulty in determining how much of a discount to apply
to evidence presented by a party who has the only lawyer in the case, or, more commonly, who has
the “better” lawyer. Jurors, who typically have limited experience in observing lawyers and
litigation, may find it particularly difficult to apply such a discount, or indeed, to determine which of
two lawyers is “better.” An example of two successful lawyers each attempting to appear less
effective than the other appears in a recent account of the famous trial in which Pennzoil obtained a
$10 billion jury verdict against Texaco. See S. COLL, THE TAKING OF GETTY OIL 422-23 (1987).
Who won the “game within a game” is indicated not only by the final outcome but by one juror’s
strong praise for Texaco’s trial attorney as a “terrific lawyer.” Id. at 469.
184. Sometimes the process of discounting the claims of the party with better advice or
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
390 CALIFORNIA LAW REVIEW [Vol. 79:313
adjustment of standards of liability and sanctions prompted by the effects
of advice need be no more ad hoc in character than any restrictions on
advice that might be prompted by those effects. In general, restrictions
would be desirable only if advice systematically benefited one side in a
definable class of litigation. Where that is so the standard of liability or
actual sanctions can, as the examples above illustrate, be adjusted across
the class of litigation without making case-by-case adjustments.
The problem of individual variation would remain.’85 Parties vary
in, among other things, the degree to which they benefit from litigation
advice, their perceptions of expected sanctions, and their aversion to risk.
These variations make it impossible to offset precisely the changes in the
probability of liability or the level of deterrence each person experiences
because of legal advice. Adjustments to the rule of decision or actual
sanctions will instead have to generalize about these varying characteris-
tics, missing fine detail. Generalization and imprecision, however, are
necessary features of every legal regime. Unless there is some reason to
believe that the pattern of imprecision that would exist without litigation
advice has some special virtues that the pattern of variation in an
adjusted regime with advice would not share, the risk of imprecision can-
not support an argument against combining legal advice and adjusted
sanctions to achieve a baseline level of deterrence that is generally higher
for undesirable acts and lower for desirable acts.
A final problem where advice benefits defendants is that sometimes
actual sanctions cannot be increased sufficiently to offset the effect of
advice without other untoward consequences. There may be physical,
constitutional, moral, or practical limits on increasing actual sanctions to
offset the effects of advice. Of these, moral and practical constraints
seem most likely to be significant.186 (None of these constraints operates
enhancing the claims of the party with worse advice might be seen as covert advocacy against the
stronger party and in favor of the weaker party. In effect, the tribunal must supply from its
experience and imagination the information and inferences that the party with inferior advice would
have presented had she been more able. Arguably, this process undermines the neutral passivity that
some accounts stress is critical to the functioning of adversary adjudication, cf. Fuller, The Forms
and Limits of Adjudication, 92 HARV. L. REV. 353, 385 (1979) (a judge should not “interfere
voluntarily in . . . cases which are presented to him”), and others argue is important to the
appearance of fairness, see Resnik, supra note 23, at 427 (a judge’s personal involvement renders
subsequent judgment unfair). These objections may seem less severe, however, when the adjustment
consists of, say, the judge’s decision to cross-examine a witness after a lawyer has done an
exceptionally poor job at that task or to comment unfavorably on the evidence presented by the
party with the better lawyer.
185. This is the obstacle stressed by Kaplow & Shavell, supra note 17, at 590-93.
186. We doubt that the fact that physical punishments are ultimately limited to death often
poses a significant barrier to increasing sanctions. Nor do we think that the constitutional
limitations on cruel and unusual (or disproportionate) punishment, as interpreted by the Supreme
Court, will impose significant constraints on increasing actual sanctions within any range that a
contemporary legislature would regard as morally permissible. Cf id. at 591 & n.63 (concluding
that recent Court decisions cast doubt on the effectiveness of that clause as a constraint).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 391
when the needed adjustment calls for decreasing actual sanctions to offset
the effect of advice benefiting accusers.) Society may be unwilling on
moral grounds to countenance even an informal adjustment in the stan-
dard of liability like that described above or to impose actual sanctions at
a level sufficient to deter the conduct in issue. For other offenses, society
may be unwilling to pay the high cost of extended incarceration.
It is important, however, not to overstate the need for or difficulty of
adjustments to offset the harmful or ambiguous effects of litigation
advice. In some classes of litigation where sanctions cannot be adjusted,
the net behavioral effects of reducing expected sanctions more for desira-
ble acts than for undesirable acts may still be socially desirable. In other
classes of litigation, sanctions may already have been set to take account
of the effects of litigation advice. In most civil litigation, for example,
sanctions may be set on the assumption that the parties have roughly
comparable litigation advice of a specified quality. If a unilateral
increase in one party’s advice simply brings the advice available to the
parties into that rough balance, no adjustment in sanctions is required.
Similarly, those who believe that sanctions in criminal cases are already
sufficiently severe and should not be further increased to offset the effects
of advice should note that those sanctions operate in a system in which
all felony defendants already receive some advice. Hence criminal sanc-
tions may already be set on the assumption that both parties will have
comparable levels of litigation advice. The question is thus not so much
whether present sanctions should go higher, but whether they should be
lowered if defendants’ access to advice is limited. As a practical matter,
then, the problems of adjusting sanctions may exist primarily where con-
ditions of relative equality are not met: criminal cases where the defend-
ant has an exceptionally good lawyer187 or civil cases where there is an
imbalance of representation.
3. Advice That Decreases Information and Increases Skew
Our analysis in Part III suggests that sometimes advice may
decrease information and increase skew. For example, in a criminal case
where a sanction-optimizing defendant faces a truth-telling prosecutor,
the defendant has acted wrongly, and there is a substantial amount of
suppressible information unfavorable to the defendant, unilateral legal
advice for the defendant might well have such effects. Such advice
reduces the tribunal’s ability to assess the desirability of defendants’ con-
duct and has undesirable behavioral consequences that adjusting sanc-
187. To the extent that this category includes those cases where defendants can retain
exceptionally talented counsel with the alleged proceeds of their crimes, the rules permitting the use
of forfeiture statutes to prevent that result may therefore have socially desirable effects on trial
outcomes that could not be achieved by adjusting sanctions. See infra text accompanying note 203.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
392 CALIFORNIA LAW REVIEW [Vol. 79:313
tions cannot fully offset.s88 This has implications if, for example, one
believes that an identifiable class of parties (such as social welfare agen-
cies or prosecutors) are generally thorough truth-tellers and that their
opponents are generally sanction-optimizers who can withhold or sup-
press a disproportionate share of the relevant information. One who
holds that belief should find some possibility of improving overall out-
comes through a combination of reducing legal advice to all opponents
and adjusting rules of decision or sanctions to make them more favorable
to opponents. We return to this disturbing class of advice in Section
V(A).
B. Nondeterrence Theories
1. Advice That Increases Information Without Increasing Skew
The informational effects of advice that, like bilateral advice in our
basic account, increases information without increasing skew should also
be desirable from perspectives other than pure deterrence. In theories of
civil or criminal justice that emphasize preventing harm, accurate judg-
ments about the character of the act sought to be enjoined or the defend-
ant who faces incapacitation are obviously important.’89 Accuracy is
also important if the goal of adjudication is to ensure fair compensation
for harm suffered or just retribution for harm done. In all these regimes,
the improved accuracy made possible by bilateral advice appears socially
desirable.
Such advice also seems likely to have a positive impact on the dis-
pute resolution function of the tribunal. First, bilateral advice increases
the ability of both parties to participate effectively in determining the
outcome of adjudication by reducing effort wasted on gathering and
presenting irrelevant information and by increasing the amount of rele-
vant information that each party is able to present. Second, many parties
undoubtedly believe that outcomes are fairer the more closely they corre-
spond to what actually happened or will happen. To the extent that
more effective participation and improved accuracy enhance the likeli-
hood that the parties will accept and abide by the tribunal’s resolution of
188. Where such advice aids defendants, it decreases expected sanctions more for violators than
for nonviolators. Increasing sanctions to offset the decrease in expected sanctions for violators will
result in a net increase in expected sanctions for nonviolators as well. Similarly, where this type of
advice aids accusers, sanctions cannot be decreased enough to offset the increase in expected
recovery for unjust accusers (and thus an increase in expected sanctions for nonviolators) without
resulting in a net reduction in expected recovery for just accusers (and thus a decrease in expected
sanctions for violators).
189. Theories of adjudication based on incapacitation differ in one important way from theories
that emphasize deterrence: for incapacitation theories, the increased accuracy of adjudication
resulting from advice need not be communicated to prospective actors in order to have socially
desirable effects.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 393
the dispute, the bilateral informational effects of legal advice should also
increase party satisfaction with outcomes.
2. Advice That Increases Information and Skew
The effects of advice that increases information and skew, such as
unilateral advice in our basic account, should also generally be desirable
in accounts of adjudication that stress harm prevention, fair compensa-
tion, just punishment, or dispute resolution. For example, unilateral
advice to defendants will, leaving aside diminishing marginal returns,
benefit harmless or deserving actors more than it will benefit dangerous
and undeserving actors. Unless it is not possible to adjust the tribunal’s
rule of decision or actual sanctions to take advantage of that difference,
the combination of advice and appropriate adjustments ought to produce
outcomes more closely matched to the character of the parties and their
conduct than those outcomes resulting without advice.-90 Even when it
is not possible to make such adjustments, it may be sufficient to justify
the advice in that it prevents inaccurate conclusions that prejudice
deserving actors more often than it causes inaccurate conclusions that
benefit the undeserving. This is especially true for advice to criminal
defendants, because of the traditional judgment that “it is far worse to
convict an innocent man than to let a guilty man go free.””‘9
3. Advice That Decreases Information and Increases Skew
Finally, we come to advice that decreases information and increases
skew. Such advice seems clearly undesirable under theories of justice
that stress enjoining wrongdoing, incapacitating criminals, compensating
injuries, or exacting retribution. By making adjudication less accurate,
such advice makes it harder to prevent harm or distribute rewards and
punishments justly.
Some might argue, however, that fairness requires providing such
advice notwithstanding its potential ill effects if the party’s opponent has
or can obtain comparable advice. Consider, for example, the stated obli-
gation to tell “the truth, the whole truth, and nothing but the truth.” On
190. For example, suppose the purpose of adjudication is to compensate those who suffer a
specified harm, whom we will call “deserving claimants.” To offset the inflationary effect of
evidentiary advice upon claims of harm, the threshold for proof of harm can be raised. If the rise in
the threshold offsets the improved showing made by undeserving claimants, then more deserving
claimants will receive compensation (because their presentations improve more with evidentiary
advice than do the presentations of the undeserving) without increasing the number of the
undeserving claimants who receive compensation. If the threshold is raised sufficiently to offset the
improved showing made by the deserving, fewer undeserving claimants will receive compensation
(because their presentations improve less with evidentiary advice than those of the deserving)
without reducing the number of deserving claimants who receive compensation. Either outcome is
preferable to that which would prevail in the absence of legal advice.
191. In re Winship, 397 U.S. 358, 372 (1970) (Harlan, J., concurring).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
394 CALIFORNIA LAW REVIEW [Vol. 79:313
its face, this oath creates an enforceable legal obligation to volunteer
information. In fact, failure to volunteer information not asked for,
although a plain violation of the oath, is not sanctionable. A party who
acts on the assumption that failure to comply with the oath by volunteer-
ing information would be detected and heavily sanctioned, while her
opponent understands and acts upon the real rules, will often be victim-
ized by the opponent.’92 One might think that if society cannot or will
not deter the wrongful conduct of those who are already knowledgeable,
the fairest way to achieve a rough balance of advantage is to allow both
parties to learn and act upon the real rules of the game.
Often these claims extend beyond fairness. Conduct that is nor-
mally undesirable in itself may be a desirable response to conduct by the
other side. Say, for example, that two sellers get advice that lowers their
expected antitrust sanctions for an anticompetitive merger that will give
them monopoly power and allow them to raise prices. A group of buyers
might then assert that “fairness” entitles them to advice that lowers their
expected antitrust sanctions for forming an illegal anticompetitive buying
cartel that will give them countervailing market power. This claim
would extend beyond fairness if they claim that forming the buying asso-
ciation is allegedly a “second-best” solution that will lead to prices and
outputs that are more efficient (and socially desirable) than those that
would result if only the sellers had monopoly power.193 All this estab-
lishes, however, is that if an opponent is known to be engaging in unde-
sirable conduct, advice that lowers sanctions for conduct that would
ordinarily be considered undesirable (such as cartels) may become desira-
ble. In these circumstances, the argument for advice that lowers sanc-
tions for the illegal conduct in question is not an individual fairness claim
in tension with social welfare: the advice is desirable to the extent that
the illegal (and ostensibly undesirable) conduct is, in this special setting,
socially desirable.194
Much more troubling is the claim that fairness or respect for party
autonomy requires that advice be provided even in cases where it has
undesirable effects that admittedly impair the tribunal’s ability to distin-
guish the desirable acts from the undesirable. Some argue that furthering
party autonomy or dignity is an intrinsic good that may justify even
192. The case for advice will be even stronger if the tribunal (1) has already adjusted sanctions
to reflect the general understanding among litigants that information should not be volunteered, but
(2) has not adjusted sanctions to compensate for the fact that one of the litigants does not have legal
advice. In such cases, the sanction adjustment effectively assumes that the witness is not
volunteering information, even if she in fact is being forthcoming.
193. See generally P. AREEDA & L. KAPLOW, ANTITRUST ANALYSIS 40-42, 195 & n.24, 223
n.46 (1988) (discussing countervailing power and second-best theory).
194. Cf Simon, supra note 12, at 1090 (arguing that professional norms should authorize
deceptive conduct in litigation when necessary to prevent unjust results).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSAR Y SYSTEM? 395
advice that makes litigation less accurate.195 Others argue that advice
which decreases the information reaching the tribunal may serve to legit-
imate litigation outcomes. In particular, the argument runs, the appar-
ent fairness of the proceeding to the losing party may depend on the
party’s sense that she has been an informed participant and that the best
possible case has been made on her behalf. 196
If, however, litigation advice decreases the accuracy of adjudication,
furthering fairness, autonomy, or party satisfaction for the advised party
has a necessary cost: denying fairness, autonomy, and party satisfaction
to those injured by inaccurate adjudication. Some might be injured
directly because they are improperly deprived of a remedy or improperly
sanctioned. Some might be injured indirectly because the resulting fail-
ure to deter undesirable activity or overdeterrence of desirable activity
infringes their freedom. For example, the autonomy of women is
infringed if they do not feel free to walk alone at night because rape is
underdeterred. And erroneous findings of malpractice might overdeter
obstetrics and make the costs of malpractice liability prohibitive, restrict-
ing the autonomy of doctors to become obstetricians and of individuals
to further their reproductive goals. Whether injured directly or indi-
rectly, those injured by inaccurate adjudication would seem to have
claims to fairness, autonomy, and party satisfaction that are far stronger
than those claiming an entitlement to advice that decreases the accuracy
of adjudication.
C. Conclusion
The informational effects of advice that, like bilateral advice in our
basic account, increases information without increasing skew are socially
desirable whether one measures its social value by its contribution to
optimal deterrence, to other goals of adjudication, or to some combina-
tion of the two. Under the analysis in Part III, this provides strong sup-
port for bilateral advice unless certain special conditions are present.
The overall informational effects of advice that increases information but
also increases skew, such as unilateral advice and bilateral advice where
informational access is skewed, also seem generally desirable. The desir-
ability of this latter form of advice may, however, depend in part on
195. See M. FREEDMAN, supra note 9, at 4. The general proposition that client autonomy
justifies lawyer conduct that would otherwise be viewed as socially undesirable is developed in Fried,
The Lawyer as Friend: The Moral Foundations of the Lawyer-Client Relation, 85 YALE L.J. 1060,
1073 (1976).
196. See Professional Responsibility: Report of the Joint Conference, 44 A.B.A. J. 1159, 1216
(1958) (legal advice legitimates outcomes by easing the “fear… that perhaps more might have been
said for the losing side”); Curtis, supra note 50, at 12 (Defending legal advice on the ground that
“the law… must give the losing party… as much satisfaction as any loser can expect. At least the
most has been said for him.”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
396 CALIFORNIA LAW REVIEW [Vol. 79:313
whether advice has diminishing marginal returns and on the legal
regime’s ability to adjust the tribunal’s rule of decision and actual sanc-
tions. Finally, where advice decreases information and increases skew, it
seems undesirable. In the classes of litigation identified in Part III where
advice is likely to give rise to such harmful effects, restrictions appear
warranted.
V
REGULATING THE PROVISION OF LITIGATION ADVICE
Many rules regulate access to litigation advice. In this Part, we first
discuss the informational effects of measures designed to increase or
decrease access to counsel. We then apply our analysis to more narrowly
targeted proposals that would limit attorney-client confidentiality or
restrict the advice a party receives about responding to investigation or
questioning. Our account clarifies the compromises inherent in those
rules and the possible justifications for the contours of present doctrine.
A. Measures That Encourage or Discourage the Use of Lawyers
To our knowledge, no legal doctrine establishes an outright ban on
consulting with a lawyer in connection with a legal dispute. A handful of
rules, however, impose relatively broad restrictions on advice. Some
states, for example, forbid lawyers from participating in small claims
court hearings.197 If it is likely that both parties would obtain advice
without this ban, then we can evaluate the ban under our bilateral
account of advice. On the classic assumption that “small” claims are
simple enough for laypersons to handle (so that lawyer participation
would normally produce no informational benefit), such a ban might be
justified as a form of paternalism to protect parties from wasting their
money.’98 Our account suggests, however, that if a substantial number
of “small” claims are in fact complex,199 such bilateral restrictions could
significantly reduce the information reaching the tribunal.
Small claims restrictions also illustrate the difficulty of determining
whether nominally bilateral restrictions on advice have bilateral or uni-
lateral effects. If only one party would retain legal advice without a ban,
then a nominally bilateral ban may in fact limit only that party’s advice.
Such a restriction might be an important safeguard against presentations
skewed to the advantage of a party willing and able to pay for legal coun-
sel. On the other hand, suppose instead that one of the parties is a repeat
197. See, e.g., CAL. CIV. PROC. CODE ? 117.4 (West Supp. 1990). See generally Yngvesson &
Hennessey, Small Claims, Complex Disputes: A Review of the Small Claims Literature, 9 LAW &
Soc’Y REV. 219, 223 n.4 (1975) (listing states barring attorneys from small claims court).
198. See Yngvesson & Hennessey, supra note 197, at 223, 225.
199. Id. at 258-59.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 397
litigant, like a bank or finance company, who has become familiar with
the small claims process through repeat exposures while the other is a
one-shot player, like a consumer debtor. In such cases, the ban will pre-
vent the one-shot player from obtaining advice that would enable her to
match the presentation of the repeat player, thus locking in place the
disadvantage imposed by the party’s inexperience.
More common than outright restrictions are legal rules that
encourage or discourage advice, typically through the use of financial
subsidies or disincentives. Often in these cases, the relevant rules clearly
have unilateral effects because they burden or benefit identifiable classes
of litigants. We have already shown how rules awarding attorney’s fees
to prevailing private attorneys general will tend to increase the informa-
tion reaching the tribunal, particularly when the party benefited by the
rule opposes a sanction-optimizing party who is also represented by
counsel.200
Whether sixth amendment subsidies for criminal defense counsel
will increase the information reaching tribunals depends in part on the
extent to which police and prosecutors are thorough and truth-telling.201
Lack of thoroughness may result from incompetence, laziness, or scar-
city of resources. Such lack of thoroughness will reduce the govern-
ment’s identification and presentation of information favorable to the
defendant. The government’s desire to optimize sanctions (rather than
simply tell the truth) will reduce its willingness to identify and present
such information. Surprisingly, judicial opinions upholding the right to
counsel seldom focus directly on these issues. Distrust of the govern-
ment’s thoroughness and motives is, however, clearly an important
empirical predicate of both the sixth amendment and the Bill of
Rights.202
200. See supra Section II(D). In some private attorney general cases the benefits flowing from
advice may be attenuated due to agency problems resulting from how attorney fees are calculated
and the plaintiff’s inability to monitor her lawyer. These difficulties may be especially acute in class
actions. See generally Coffee, The Unfaithful Champion: The Plaintiff as Monitor in Shareholder
Litigation, 48 LAW & CONTEMP. PROBs. 5 (Summer 1985).
201. Prosecutors (and some other government litigants) may differ from the parties assumed in
our basic account because they have a duty to pursue the truth over victory and do not have the
same consistent personal interest in optimizing expected sanctions that most private litigants do. To
the extent this public interest duty is enforced by evidentiary rules, such as the rule requiring that a
conviction be vacated or reversed when prosecutors have withheld material exculpatory information
from the defendant, see, e.g., United States v. Bagley, 473 U.S. 667, 674-75 (1985), our sanction-
optimizing account addresses it when we discuss evidentiary sanctions. To the extent, however, that
public interest motivations are internalized by governmental litigants, such litigants are less likely to
seek solely to maximize the sanctions imposed on their opponents. A prosecutor may, for example,
turn over exculpatory evidence even though, given the low probability that withholding would be
detected, withholding that evidence would on balance increase the sanctions imposed on the
defendant.
202. Nonetheless, some undoubtedly believe that prosecutors are inherently truth-telling or law-
abiding. To the extent this belief is correct, advice for criminal defendants is less desirable. Those
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
398 CALIFORNIA LAW REVIEW [Vol. 79:313
Conversely, the positive informational effects of subsidized advice
in criminal cases will be weakened to the extent that defendants are
sanction-optimizers who control or can suppress most of the information
unfavorable to their cause. Undeniably, many accused wrongdoers face
horrific sanctions that any reasonable person would strongly wish to
minimize. In addition, the fourth and fifth amendments afford criminal
defendants opportunities not available to civil litigants to withhold or
suppress unfavorable information. Both factors create uncertainty about
whether the net informational effects of subsidized advice for criminal
defendants are as positive as those in civil cases.
If one can identify a definable class of advice to criminal defendants
that seems likely to have negative effects, it may even seem desirable to
implement laws creating financial obstacles to acquiring that advice. For
example, in upholding a drug forfeiture statute, the Supreme Court
treated as legitimate the statutory purpose of preventing organized crime
and drug enterprises from using drug proceeds to hire high-priced legal
counsel, despite recognizing that this would adversely affect the quality
of legal representation.203 Within our account, the statute’s informa-
tional effects flow from the unilateral decrease in legal advice resulting
from the difference between the high-priced counsel that defendants
could afford without forfeiture and the appointed legal counsel that
defendants would get with forfeiture. Our model suggests these informa-
tional effects are likely to be desirable for known drug dealers. The more
difficult question, however, is whether these effects are desirable for
the whole class of cases covered by the forfeiture provisions of the
Continuing Criminal Enterprise (CCE) and Racketeer Influenced and
Corrupt Organizations (RICO) statutes. Given the notorious vagueness
of the latter in particular,204 one might well have doubts that the statutes
reliably define a class of undesirable advice meriting restriction.
Just as distrust of the government supports providing advice to
criminal defendants, a showing that the government is thorough and
benignly motivated may sometimes support rules that restrict or discour-
age legal advice to the government’s opponents in civil cases. In Walters
who believe prosecutors are law-abiding or truth-telling would, however, do well to ask themselves
to what extent their understanding of prosecutorial behavior is based on their observation of
prosecutors who anticipate that the defendant will have legal advice (and who calculate the expected
value of withholding, suppression, and prosecution accordingly) or who have strong moral and
prudential constraints against picking the wrong defendant. The documented performance of
prosecutors who imagine that their opponents will not have effective advice or who do not labor
under such moral and prudential constraints is often disturbing. See, e.g., D. CARTER,
SCOTTSBORO: A TRAGEDY OF THE AMERICAN SOUTH (2d ed. 1979); R. FROST, THE MOONEY
CASE (1968); E. Morris, The Thin Blue Line (Miramax Films 1988) (film).
203. See Caplin & Drysdale v. United States, 109 S. Ct. 2646, 2654-55 (1989).
204. See generally H.J., Inc. v. Northwestern Bell Tel. Co., 109 S. Ct. 2893, 2897-909 (1989)
(struggling with the problem of defining a “pattern of racketeering activity”).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 399
v. National Association of Radiation Survivors, 205 the Supreme Court held
that the due process clause was not violated by a statute that prohibited
claimants from paying more than $10 in legal fees when seeking benefits
from the Veterans Administration. The Court agreed that the statute
restricted claimants’ access to advice, but found the restriction justified.
In part, the Court relied on statistical evidence that claimants who used
lawyers obtained results similar to those achieved by claimants who used
lay agents and on the Court’s own judgment that most claims were rela-
tively simple and did not require a lawyer.206 Had the Court limited
itself to those arguments, the case could be seen as reflecting a rationale
similar to that used in denying advice to litigants in small claims court:
where legal advice makes no difference, no harm is done by restricting its
availability.
Other portions of the Walters opinion, however, suggest that the
critical factor was the Court’s view of the thoroughness and motivation
of the Veterans Administration. The Court repeatedly stressed that
Veterans Administration proceedings were not adversarial.207 The
agency was required to ” ‘assist the claimant in developing the facts per-
tinent to his claim’ “208 and to operate with “a high degree of. . . solici-
tude for the claimant.”209 The Court emphasized that the agency
decisionmakers possessed legal, medical, and occupational expertise, as
well as the resources to conduct their own investigations.210 Further, the
agency’s rule of decision strongly favored the claimant.211 Finally,
although the Court minimized the differences in relevant knowledge and
skill between attorneys and other agents, it concluded that the attorney’s
distinctive ethical orientation might sometimes require her to prevent
unfavorable information from reaching the tribunal.212
Our account of advice provides a framework for understanding the
Court’s opinion. The Court apparently believed that the Veterans
205. 473 U.S. 305 (1985).
206. Id. at 330-31.
207. Id. at 309, 323, 324; see id. at 324 n.l 1 (rejecting as unfounded the district court’s finding
that the process was adversarial).
208. Id. at 310 (citing 38 C.F.R. ? 3.103(a) (1984)).
209. Id. at 311; cf id. at 314 n.6 (rejecting the district court’s suggestion that the Veterans
Administration might have a conflict of interest between its solicitude for veterans and its desire to
protect the public fisc).
210. Id. at 309-10.
211. Id. at 310 (citing 38 C.F.R. ? 3.102 (1984)) (Administration must resolve all reasonable
doubts in favor of the claimant); id. at 311 (Administration must read any submission in the light
most favorable to the claimant).
212. The Court cited Judge Henry Friendly for the proposition that “‘under our adversary
system the role of counsel is not to make sure the truth is ascertained but to advance his client’s
cause by any ethical means. Within the limits of professional propriety, causing delay and sowing
confusion not only are his right but may be his duty.’ ” Id. at 325 (quoting Friendly, Some Kind of
Hearing, 123 U. PA. L. REV. 1267, 1287-90 (1975)).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
400 CALIFORNIA LAW REVIEW [Vol. 79:313
Administration was a thorough truth-teller, both capable of bringing out
all favorable and unfavorable information about the claimant’s alleged
injury or illness and motivated to do so. Given the Court’s further
assumption about the professional obligations of lawyers, allowing
claimants access to legal advice would reduce the information reaching
the tribunal. It would enhance the claimant’s ability to withhold and
suppress unfavorable information that the agency would otherwise
obtain and present, and skew the mix of information in favor of the
claimant.213
As shown in Section IV(A)(3), this effect would generally benefit
undeserving claimants more than it would benefit deserving claimants,
because undeserving claimants normally have more unfavorable informa-
tion. Moreover, the two responses available to the Veterans Administra-
tion could well be harmful to deserving claimants. The Administration
might adjust its own behavior to become less truth-telling.214 In that
case, more claimants with valid claims would need lawyers to prevail.
Alternatively, the Administration could adjust its claimant-favoring rules
of decision to reduce the success rate of undeserving claimants. But
because the change in information presented on account of advice would
generally benefit the undeserving more than the deserving, the results
with the adjustment would tend to deny deserving claimants benefits
more often than undeserving claimants.
Although the Court’s optimistic view of the Veterans Administra-
tion may have been incorrect,215 the case illustrates the possibility of a
general class of cases where the government’s thoroughness and benign
motivation might make restricting advice to the government’s opponents
desirable.216 As our analysis also suggests, such restrictions on advice
213. This general phenomenon is discussed in supra Section III(B)(2).
214. The Court hinted at this fear by citing a case where a potential change in the government’s
presumptively benign attitude toward its adversary justified denying the right to counsel. Walters,
473 U.S. at 324 (citing Gagnon v. Scarpelli, 411 U.S. 778 (1973)). In Gagnon, the issue was whether
a parolee was entitled to counsel in a revocation proceeding. The Court stressed that parole was
rehabilitative rather than punitive, and that introducing lawyers into the proceeding might cause the
parole board to become “less attuned to the rehabilitative needs of the individual.” See Gagnon, 411
U.S. at 787-88.
215. Without citing any contrary evidence, the Court reversed several district court findings
suggesting that the agency was neither competent nor benign. See supra notes 207 & 209.
Moreover, after the Court’s decision, the agency was found to have destroyed or hidden documents
relevant to claims against it. See National Ass’n of Radiation Survivors v. Turnage, 115 F.R.D. 543,
546 (N.D. Cal. 1987).
216. See supra Section IV(A)(3). Professor William Simon’s analysis of the role of counsel in
welfare hearings reaches a similar conclusion. See Simon, Legal Informality and Redistributive
Politics, 19 CLEARINGHOUSE REV. 384, 386 (1985) (stressing the benign orientation of the welfare
bureaucracy and the likelihood that the presence of lawyers may lead to rule adjustments that
disadvantage deserving claimants); see also id. at 390-91 (stating that formal welfare hearings
reflected elite professional distrust of the motives and competence of working-class lower-level
bureaucrats).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSAR Y SYSTEM? 401
will often be coupled with rules of decision favorable to the party denied
advice. The relevance of government motivation may explain why
restrictions on legal advice are more tolerable as part of a program of
government benefits. As the proponents of the “new property” have long
noted, the denial of government benefits can have consequences as severe
or more severe than the denial of other forms of property.217 It may,
however, seem particularly likely that the government will be benignly
motivated in benefits cases. This could make advice to beneficiaries
undesirable despite the severity of denying benefits. Indeed, the severity
of denials would only heighten the undesirability of advice because such
advice makes it harder to channel scarce benefits to deserving
beneficiaries.
B. Attorney-Client Confidentiality
A variety of rules protect the confidentiality of the attorney-client
relationship. The attorney-client privilege generally bars compelled dis-
closure of the contents of communications between lawyer and client
made in confidence for the purpose of getting or giving legal advice.218
The ethical obligation of confidentiality generally bars attorneys from
voluntarily disclosing information harmful to their clients, no matter
what its source, unless the client insists on committing or has committed
fraud on the tribunal in order to withhold the information.219 Work
product immunity provides qualified protection from discovery of docu-
ments and tangible things reflecting the results of investigation (“ordi-
nary work product”) and absolute or near absolute protection for the
lawyer’s professional judgments (“opinion work product”), even if those
judgments are not embodied in advice communicated in confidence to
the client.220
The privilege is intended to encourage parties to communicate
harmful information to lawyers by reducing the risk that lawyers will
subsequently disclose that information to the tribunal.221 The general
obligations of confidentiality and work product immunity similarly
217. See Goldberg v. Kelly, 397 U.S. 254, 264 (1970); Reich, The New Property, 73 YALE L.J.
733, 769 (1964).
218. UNIF. R. EVID. ? 502, 13A U.L.A. 256 (1986); McCORMICK ON EVIDENCE ? 87 (E.
Cleary 3d ed. 1984).
219. MODEL RULES, supra note 1, Rule 3.3; ABA Standing Comm. on Ethics and Professional
Responsibility, Formal Op. 87-353 (1987).
220. See Special Project, The Work Product Doctrine, 68 CORNELL L. REV. 760, 773-80 (1983)
(authored by Jeff A. Anderson, Gena E. Cadieux, George E. Hays, Michael B. Hingerty & Richard
J. Kaplan); Note, The Work Product Doctrine in Subsequent Litigation, 83 COLUM. L. REV. 412,
413-20 (1983) (authored by Caroline T. Mitchell). We discuss the work product doctrine more fully
at infra notes 247-59 and accompanying text.
221. E.g., 8 J. WIGMORE, EVIDENCE IN TRIALS AT COMMON LAW ? 2291, at 545 (J.
McNaughton rev. ed. 1961).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
402 CALIFORNIA LAW REVIEW [Vol. 79:313
encourage parties to involve lawyers in conducting investigations and
evaluating the fruits of those investigations. We will describe both deci-
sions to communicate information directly to a lawyer and decisions to
allow the attorney to learn information from other sources as decisions to
“share” information.
It is often suggested that narrowing confidentiality rules, with a con-
comitant expansion of the attorney’s obligation to disclose information to
the tribunal or the opponent, would increase the amount of information
reaching the tribunal.222 Our account assists in understanding what is at
stake in this claim, because it illuminates the informational effects of the
information sharing encouraged by confidentiality, the informational
costs of confidentiality where it does not affect information sharing, and
the factors influencing the causal connection between confidentiality
rules and information sharing.223
For clarity, we contrast two simplified regimes. The regime of abso-
lute confidentiality neither requires nor permits attorneys to disclose
information harmful to their clients’ interests and requires attorneys to
provide litigation advice no matter what their clients intend to do with it.
The regime of absolute disclosure requires attorneys to disclose to the
tribunal all relevant information and all information that might lead to
the identification and presentation of relevant information.224 We ini-
tially assume that the attorney’s obligations under either regime are per-
fectly enforced. Thus, under the regime of absolute confidentiality, a
client’s decision to share information with a lawyer never increases the
likelihood that unfavorable information will reach the opponent or the
tribunal. Under the regime of absolute disclosure, harmful information
shared with an attorney is always disclosed. We also initially assume
that decisions to share information reflect a full understanding of the
applicable rule of confidentiality or disclosure. Later, we discuss the
implications of imperfect enforcement of attorneys’ obligations and
imperfect client information concerning the risk that the attorney will
disclose shared information.
A regime of attorney disclosure will deter a party from sharing
information with her lawyer only225 where the party fears such sharing
222. See Frankel, The Search for Truth: An Umpireal View, 123 U. PA. L. REV. 1031, 1055-57
(1975) (evaluating this approach).
223. We limit our discussion in this Section to the informational effects of advice. To the extent
that confidentiality in litigation would also promote socially desirable withholding of information,
foster advice about settlement or related primary conduct, or allow the sharing of information about
party preferences that would permit more satisfactory outcomes, the case for the social desirability
of confidentiality would evidently be stronger.
224. This is, in essence, Judge Frankel’s proposal. See Frankel, supra note 222, at 1057-58.
225. A party may also be deterred from communicating information to her lawyer by the risk
that favorable information will be disclosed, if disclosure would be harmful or embarrassing
independent of its effect on the merits. To the extent the communication of favorable information
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 403
will cause unfavorable information to reach the tribunal that otherwise
would not have.226 Confidentiality thus encourages a party to share
information that she fears may be unfavorable or may lead to unfavora-
ble information.227 In addition, confidentiality encourages the attorney
to investigate for information that the party would dismiss as irrelevant,
and thus not affirmatively share, despite the risk that such investigation
may produce unfavorable information.
The information-sharing caused by confidentiality improves the
attorney’s ability to give advice.228 When the party is correct that dis-
closing the shared information would be harmful, a lawyer operating
under a regime of confidentiality will often advise her client to withhold
the shared information. But this consequence cannot be considered a
cost of confidentiality nor of the advice that it facilitates, because neither
the attorney nor the tribunal would have received that information in the
absence of confidentiality. Without confidentiality, the lawyer would not
have learned the information and would thus have had nothing to dis-
close to the tribunal. And if a client would not allow her lawyer to learn
information because she feared the lawyer would disclose it, she certainly
would not herself disclose it directly to the opponent or the tribunal.229
The lawyer may instead advise that the information communicated
is favorable, rather than unfavorable or irrelevant, and should be
presented. Or she may advise that, although the shared information is
unfavorable or may lead to unfavorable information, the party should
produce it because either the evidentiary sanctions are higher or the
degree of unfavorability is lower than the party believed. These types
of advice will increase the information reaching the tribunal.230 An
would be deterred, it strengthens the case for confidentiality, but to clarify the exposition we do not
discuss it further.
226. Although our account assumes the fear of unfavorable information reaching the tribunal is
necessary to deter information-sharing, we do not assume it is sufficient, even when the fear is
correct. See infra note 231 and accompanying text.
227. If parties are truth-tellers rather than sanction-optimizers, they will present all information
even if it is unfavorable. Thus, confidentiality does not result in more information being shared by
truth-tellers. For law-abiders, however, confidentiality is important because, without it, sharing
information with an attorney may result in the attorney disclosing information that the party was
not herself legally obliged to produce.
228. It may sometimes be possible for a party to receive advice about information without
actually sharing it. The party may, for example, ask her attorney hypothetical questions without
revealing the actual facts. Our analysis assumes only that the attorney’s ability to use or give advice
about the information is at least somewhat impaired by the lack of actual sharing.
229. Indeed, unless the information is the type about which clients can testify, clients are
unlikely even to have any opportunity to present the information directly to the tribunal.
230. A recent article justifies the attorney-client privilege based on one specific form of beneficial
advice: advice that apprises clients of a “contingent claim” that may make apparently unfavorable
or irrelevant information favorable. See Allen, Grady, Polsby & Yashko, A Positive Theory of the
Attorney-Client Privilege and the Work Product Doctrine, 19 J. LEGAL STUD. 359, 363-69 (1990). It
seems plain to us, however, that advice about contingent claims is but one subset of advice that
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
404 CALIFORNIA LAW REVIEW [Vol. 79:313
information increase should also result where confidentiality facilitates
attorney investigation by causing the party either to communicate infor-
mation that aids the attorney’s investigation or to delegate the execution
of investigations to that attorney. To be sure, sharing information may
also allow the identification of additional suppressible information. But
some of this suppressible information will be unfavorable (and thus sup-
pressed) and some favorable (and thus saved from suppression by the
opponent), with ambiguous effects on the amount of information reach-
ing the tribunal. On balance, then, advice resulting from the sharing of
information caused by confidentiality should increase the information
reaching the tribunal.
The information lost to the tribunal because of confidentiality will
be a portion of the unfavorable information the party would share with
her lawyer whether or not she was guaranteed confidentiality. Even
under a regime of absolute disclosure, there are several reasons why a
party might share information with her attorney notwithstanding the risk
that doing so will cause unfavorable information to reach the tribunal.
She may believe the information being shared is more likely to be
favorable than unfavorable, or that, given perceived evidentiary sanc-
tions, she will have to produce the shared information anyway. Or she
may believe that the benefits of litigation advice obtainable only by shar-
ing information (such as the skilled execution of investigations) outweigh
the risk that sharing will cause unfavorable information to reach the tri-
bunal.231 Once the party shares the information, advice may correct the
party’s misperceptions that the information is favorable, overestimations
of evidentiary sanctions, or underestimations of the degree of
unfavorability.232 In any of these circumstances, a regime of confidenti-
ality will, compared to a regime of disclosure, result in less information
reaching the tribunal.233
The net informational effects of confidentiality thus depend on
might correct party misperceptions that information is unfavorable or irrelevant. Such advice is in
turn, as our textual analysis reveals, but one subset of the types of advice that can increase the
information reaching the tribunal. We thus do not share the conclusion of Allen, Grady, Polsby,
and Yashko that where “legal claims are not contingent . . . there would be no reason to have the
privilege.” Id. at 367.
231. Cf. supra Section I(C)(1) (explaining how identifying even unfavorable information the
opponent will not obtain can be helpful, and how even investigations with perceived or actual
positive expected value can collect harmful unfavorable information).
232. One important determinant of the degree of unfavorability is the likelihood that the
opponent will independently obtain and present the functional equivalent of the information. Where
the attorney advises the party that this likelihood is low, the party should be more likely to withhold
the information despite the threat of evidentiary sanctions.
233. Where the opponent would present the information independently, or where the unadvised
party would correctly determine that information is favorable or should be produced despite its
unfavorable nature, the information will reach the tribunal whether or not there is a confidentiality
rule and does not contribute to the social value of compelled disclosure.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSAR Y SYSTEM? 405
whether the increase in information resulting from confidentiality-
induced information sharing is greater than the unfavorable information
lost because it would have been shared (and thus disclosed) with or with-
out confidentiality. That tradeoff in turn depends on the extent to which
confidentiality causes information sharing. If confidentiality has no effect
on information sharing, then it causes the loss of much information with-
out offsetting informational benefits. If relatively little information shar-
ing would occur without confidentiality, then relatively little unfavorable
information is lost (because attorneys would not have much to disclose
without it) but more favorable (and perhaps some unfavorable) informa-
tion is gained as a result of greater information sharing.
Imperfect enforcement of attorneys’ obligations under a regime of
disclosure or confidentiality may alter the extent of our conclusions but
not their qualitative nature. The attractiveness of imposing disclosure
obligations on attorneys is not that those obligations can be perfectly
enforced, but rather that attorneys are often more susceptible to sanc-
tions than their clients.234 As long as attorneys respond to sanctions by
disclosing information more often under a regime of disclosure than a
regime of confidentiality, and in some cases disclose information that the
party would not, then the risk that sharing information will cause unfa-
vorable information to reach the tribunal (and thus the disincentive to
share information) will be greater under a regime of disclosure, even if
enforcement is imperfect. To be sure, imperfect enforcement will
decrease the extent to which disclosure obligations discourage informa-
tion sharing. But it will also decrease the extent to which disclosure obli-
gations cause disclosures. It is thus ambiguous, without empirical
evidence, whether imperfect enforcement will make a rule of disclosure
more or less attractive.
Imperfect knowledge about the applicable rules of disclosure or con-
fidentiality (and thus about the extent of the risk that sharing informa-
tion will cause unfavorable information to reach the tribunal) makes the
informational effects of confidentiality less desirable because it weakens
the causal connection between the existence of confidentiality and the
level of information sharing. But though the magnitude of the effects will
be influenced by imperfect knowledge, the nature of the effects we
describe will still hold so long as there is not complete ignorance, that is,
so long as there is some positive correlation between the level of informa-
tion sharing and the degree of confidentiality.
Insofar as the causal connection depends on this awareness of the
applicable rule of confidentiality, there is good reason to believe it will be
stronger for attorney-client confidentiality in litigation than for attorney-
234. See supra Section II(C)(2).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
406 CALIFORNIA LAW REVIEW [Vol. 79:313
client confidentiality in nonlitigation settings or for confidentiality rules
involving other professions. A litigation lawyer under an enforceable
obligation to disclose will likely recognize a larger and more immediate
risk to the client from information sharing, will be able to calculate that
risk more accurately (and where necessary communicate it more clearly
to the client), and will be more concerned with the professional and per-
sonal consequences of having to make disclosures harmful to the
client.235
Knowledge about the risks of disclosure is, however, only one deter-
minant of the causal connection between confidentiality and sharing of
information. Another is the private value of sharing information despite
the absence of confidentiality. Thus, even if litigants’ decisions to share
information are more likely to reflect awareness of the applicable rules of
confidentiality than the decisions of clients in other relations, the causal
connection may still be weaker if the private value to litigants of sharing
information is unusually high.
The private value of sharing information may vary for different
kinds of litigation advice. In particular, the multiple contributions of
advice to the execution of investigation or to the in-court presentation of
235. These considerations are not refuted by reports that under present confidentiality rules
many clients are not advised about the details of confidentiality. See, e.g., Zacharias, Rethinking
Confidentiality, 74 IOWA L. REV. 351, 377, 379 (1989). Apart from the fact that those reports do
not focus on advice in litigation, their findings do not support an inference either that present rules
do not affect information sharing or that changes in those rules would not do so. Parties may often
have a general awareness of the applicable rule of confidentiality without having been specifically
advised about it. Indeed, Zacharias’ own study suggests that most clients have a general notion of
lawyer-client confidentiality, although few learned about it from their lawyers and most
overestimated or underestimated its scope. Id. at 383.
A lawyer may also not bother to advise a client about confidentiality when correcting a client’s
misconceptions does not seem likely to help the client, given the benefits and risks of sharing
information. Furthermore, if a client’s misconceptions will cause a harmful failure to share
information, advice that confidentiality is absolute, although technically inaccurate and arguably
paternalistic, may be more efficient, more likely to induce helpful sharing of information, and, under
present rules, substantively correct for most clients most of the time. If a client’s misconceptions will
cause harmful sharing, the attorney may find it more efficient and tactful simply not to inquire into a
particular subject or to halt information sharing about that subject, without explaining the law of
confidentiality. The imprecision of clients’ understanding and of lawyers’ advice about
confidentiality therefore may demonstrate only that, under existing rules, few serious issues arise, or
that when such issues arise lawyers are able to advance their clients’ interest in sharing or not
sharing information without fully explaining those rules.
In fact, the only study of which we are aware that has observed lawyer-client interaction in
litigation suggests that sanctioning attorneys for withholding information has a significant impact on
information sharing. Professor Mann’s study of white-collar criminal defense attorneys showed that
they sometimes avoided inquiry into subjects when they believed the client was likely to share
unfavorable information. See K. MANN, supra note 30, at 103-11. Although these attorneys
operated under the strong confidentiality rules of the New York Code of Professional Responsibility,
their fear that prosecutors might detect and sanction misstatements made in plea bargaining
negotiations sometimes caused them to remain intentionally ignorant of their client’s information so
that they could avoid disclosure of unfavorable information in those negotiations. Id. at 83.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 407
information from nonparties normally cannot fully be realized unless the
party delegates the conduct of those activities to her lawyer and shares
with her the information required to make that delegation effective.
Because this increases the likelihood that such information sharing
would take place without confidentiality, reduced protection may be
appropriate. This may explain why the law provides less protection for
information an attorney learns from persons other than the client, which
is exempt only from voluntary disclosure by the attorney, while informa-
tion an attorney learns from client communications is exempt even from
compelled disclosure.236
This analysis does not pretend to demonstrate the desirability of a
rule of absolute confidentiality, let alone of the present confidentiality
rules. Such a demonstration would, at a minimum, require better empiri-
cal evidence about the causal connection between confidentiality and
information sharing. Our aim is rather to frame the inquiry more accu-
rately. In particular, the analysis demonstrates the error of conflating
the issue of the effects and desirability of litigation advice generally with
the issue of the effects and desirability of attorney-client confidentiality in
litigation. In our account, the informational effects of litigation advice
are, on balance, socially desirable, but the informational effects of confi-
dentiality are uncertain because of empirical questions about the effect of
confidentiality on the willingness of parties to share information with
their lawyers.
But even if it were true that, on balance, litigation advice has ambig-
uous informational effects, it would not follow, as Professors Kaplow and
Shavell have argued, that the case for attorney-client confidentiality in
litigation is a priori the “weakest” of any privilege.237 Even in Kaplow
and Shavell’s simplified model of advice, which considers only the effect
of advice about categorization upon the production of information that is
effectively party-controlled, confidentiality should encourage the party to
share with her lawyer some information the party fears is unfavorable.
Where the party’s fears are correct, the advised party will not present
unfavorable information. But that cannot be deemed a cost of confidenti-
ality since the party would not present information to the tribunal that
she would not disclose to her attorney solely because of the risk the attor-
ney would disclose it to the tribunal. Where, however, the shared infor-
mation is actually favorable, the lawyer will advise that it should be
presented. The sharing of information induced by confidentiality should,
under Kaplow and Shavell’s model, thus increase the presentation
of favorable information without reducing that of unfavorable
236. See supra note 45; supra text accompanying notes 218-20; infra note 249 and
accompanying text.
237. Kaplow & Shavell, supra note 17, at 600 n.84.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
408 CALIFORNIA LAW REVIEW [Vol. 79:313
information.238 Confidentiality should, on the other hand, prevent the
lawyer from disclosing some unfavorable information that the party
would have shared even without confidentiality.
Thus, as in our account, the effects and desirability of confidentiality
in Kaplow and Shavell’s model depend critically on the extent to which
parties would share information with their lawyers in the absence of con-
fidentiality. If the causal connection between confidentiality and infor-
mation sharing is strong, then the privilege will cause a strong increase in
favorable information presented, with little loss of unfavorable informa-
tion. This is normally desirable under any theory of adjudication, at least
if both parties have lawyers.239 While it is true that their model-like
ours-provides no a priori reason to think that the favorable information
gained because of attorney-client confidentiality is worth the unfavorable
information lost, that a priori ambiguity exists for all privileges: we can
238. Kaplow and Shavell reach their conclusion that litigation advice presents the weakest case
for protected confidentiality because they argue that the social value of some advice, like medical
advice, is clear, whereas the social value of litigation advice is ambiguous. See id. at 600.
Accordingly, they seem to reason that, while other privileges at least generate some clear social
benefit to offset against the information lost because of confidentiality, privileging litigation advice
may generate no benefit to weigh against the information lost. See id. at 600 & n.84, 610 & n.116.
This is wrong on two scores. First, it erroneously conflates the effects of advice encouraged by
confidentiality with the effects of advice generally. Even if the Kaplow/Shavell model accurately
captured the real world, and litigation advice ambiguously increased favorable information and
decreased unfavorable information, the analysis in text demonstrates that the effects of the litigation
advice encouraged by confidentiality would be an increase in favorable information without any
decrease in unfavorable information. Such an increase in favorable information standing alone is
ordinarily desirable for the reasons articulated in Part IV. To be sure, this social gain from the
communications encouraged by litigation advice must be weighed against the loss of information
from the communications that would have occurred even without confidentiality. But that is no
different from weighing the social gain of the doctor-patient communications encouraged by
confidentiality against the loss of information from doctor-patient communications that would have
occurred even without confidentiality.
Second, the social desirability of encouraging medical advice is not always so clear-cut. Medical
care is often wasteful or excessively costly and can cause injury. See I. ILLICH, MEDICAL NEMESIS
(1976) (arguing that the medical system creates more illness than it relieves). Medical care can also
restore the health of bad people who go on to commit bad acts. Finally, medical advice can be
abused for purposes that have nothing to do with good health: for example, fraud rings use medical
advice to trump up personal injury claims. Indeed, such dubious uses of medical advice may seem
particularly likely when patients would not confide in doctors without assurances that their
communications would be kept from a tribunal.
239. Perhaps Kaplow and Shavell would go further, and contend that even a bilateral increase
in information presented to the tribunal is socially ambiguous, based on their assumption that parties
are unaware of any increases in the accuracy of adjudication resulting from legal advice at the time
parties engage in primary conduct. See Kaplow & Shavell, supra note 17, at 588-90, 610. We have
already argued that the assumption that parties are, at the time they act, generally ignorant of court
outcomes is implausible, and that, even if the assumption were accurate, increased accuracy would
still be desirable under theories of adjudication that stress corrective justice, prevention of harm, or
fairness. See supra Part IV. More importantly here, if informational increases have ambiguous
social value, then that ambiguity attaches both to the benefits and the costs of confidentiality. Thus,
it offers no grounds for concluding that the attorney-client relation in litigation presents the weakest
case for confidentiality.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 409
never know whether the benefits resulting from facilitated communica-
tions are worth the costs of the information lost without knowing the
extent to which confidentiality facilitates communications.240
Although our account of confidentiality cannot justify the precise
contours of present doctrine, it does generate some rough predictions
about those contours. First, our account predicts that there should be no
confidentiality protection for information that clients will normally dis-
close to lawyers with or without confidentiality. This prediction is
clearly consistent with the requirement that to receive protection under
the privilege, client communications seeking legal advice must be made
“in confidence.”241 It may also help explain why the attorney-client priv-
ilege generally does not protect a client’s identity.242 Presumably most
clients would communicate their identity to the lawyer with or without a
guarantee of confidentiality in order to secure representation: in such
cases, the privilege would deprive the tribunal of information without
increasing the flow of legal advice to the client. It is also consistent with
our analysis that the general rule about client identity has an exception
where the client’s identity is the “last link” in proving a crime.243 The
“last link” exception appears to identify a relatively narrow set of cir-
cumstances in which it is highly likely that lack of confidentiality would
deter a client from seeking advice: when her identity is all the govern-
ment needs to impose criminal sanctions.
Second, our account suggests that confidentiality should be
restricted when the additional advice facilitated by information sharing is
highly likely to lead to the presentation of false evidence or the withhold-
ing or suppression of truthful information. Evidently this is consistent
with the exception to the privilege in cases where the client sought advice
in aid of conduct that he knew or reasonably should have known was
criminal or fraudulent2″ and with the exceptions to some general confi-
dentiality rules requiring lawyer disclosure of unfavorable information to
prevent or remedy the use of the lawyer’s services in offering known false
evidence.245 It may also help explain the otherwise anomalous rule
240. See generally Developments in the Law–Privileged Communications, 98 HARV. L. REV.
1450, 1471-500 (1985) (explaining why the a priori desirability of any privilege cannot be determined
under a variety of utilitarian and nonutilitarian theories).
241. See, e.g., 8 J. WIGMORE, supra note 221, ? 2311.
242. In re Michaelson, 511 F.2d 882, 889 (9th Cir.), cert. denied, 421 U.S. 978 (1975); 8 J.
WIGMORE, supra note 221, ? 2313.
243. United States v. Hodge & Zweig, 548 F.2d 1347, 1353 (9th Cir. 1977).
244. See UNIF. R. EVID. ? 502(d)(1), 13A U.L.A. 257 (1986); 8 J. WIGMORE, supra note 221,
? 2298.
245. See MODEL RULES, supra note 1, Rule 3.3. The disclosure obligations in Rule 3.3 are
express exceptions to the general confidentiality obligations established in Model Rule 1.6. Id. Rule
3.3(d). The Model Code of Professional Responsibility does not permit attorneys to present or assist
in presenting known false evidence. MODEL CODE, supra note 2, DR 7-102(A)(4), (6), (7). But in
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
410 CALIFORNIA LAW REVIEW [Vol. 79:313
requiring criminal defense attorneys in possession of physical evidence of
a crime to turn it over to the prosecutor.246 The apparent effect of that
rule is to reduce the extent to which lawyers examine, move, or analyze
apparently incriminating physical evidence. This effect may be socially
desirable if the attorney advice in such cases would normally result in
suppression of evidence. Because the consequences flowing from the
prosecution’s discovery of incriminating physical evidence will often be
severe, because physical evidence may often be relatively easy to sup-
press, and because the fifth amendment may well reduce evidentiary
sanctions for client suppression, the thesis that such advice is likely to
lead to suppression is relatively plausible. If, however, much physical
information that appears incriminating actually proves on investigation
to be exculpatory, then the rule may also deter advice that would
increase the presentation of truthful information. The apparent wide
acceptance of the rule may reflect “hindsight bias” resulting from the
fact that challenges to the rule occurred in cases where defendants were
found guilty and where the relevant physical information was clearly
inculpatory.
Our account also provides a clearer understanding of the justifica-
tion for work product immunity. A recent rational actor account justi-
fies the work product immunity on the ground that it increases the
production of information to the tribunal by encouraging investigations
that risk uncovering both favorable and unfavorable information.247
Without an immunity, the account argues, litigants may have an insuffi-
cient incentive to investigate in such cases of “joint production.”248 As
applied to fact investigation, there appears to be a serious doctrinal objec-
tion to this theory: if work product immunity is intended to encourage
parties to risk uncovering unfavorable factual information, then it should
protect that information from discovery. But the Supreme Court’s state-
ment of the doctrine and later cases interpreting amended Rule 26 of the
Federal Rules of Civil Procedure both expressly state that the immunity
does not protect information identified in investigation from discovery.249
A more plausible theory of the relation between work product
the name of confidentiality, it sharply restricts the lawyer’s ability to correct known false evidence
that the client has already presented in the course of the representation. Id. DR 7-102(B)(1).
246. See supra note 90 (citing cases following this rule).
247. Allen, Grady, Polsby & Yashko, supra note 230, at 385-87.
248. Id. at 385-86.
249. The Court in Hickman v. Taylor, 329 U.S. 495 (1947), stated that interrogatories drawn
with sufficient breadth can oblige a party to disclose “all pertinent information gleaned by [the
party’s lawyer] through his interviews with the witnesses,” id. at 508-09, and that “searching
interrogatories” serve to reveal the “facts” in the lawyer’s possession, id. at 513. The same result
apparently follows under Federal Rule of Civil Procedure 26(b)(3). See 8 C. WRIGHT & A. MILLER,
FEDERAL PRACTICE AND PROCEDURE ? 2023, at 194 & nn.16 & 17 (1970 & Supp. 1990) (citing
cases); 4 MOORE’S FEDERAL PRACTICE ? 26.64[1], at 26-348 (3d ed. 1989) (citing cases).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 411
immunity and the production of information would begin with a clearer
understanding of what the immunity protects. Clearly the rule protects
“documents and tangible things” prepared in anticipation of litigation by
a party or her agents, including her lawyer.250 It also accords absolute or
near-absolute protection to documents or tangible things that reflect the
“mental impressions, conclusions, opinions and legal theories” of the
party’s lawyer.25 Finally, in the view of some, the rule also protects–or
at least ought to protect-the mental impressions and opinions of the
party’s lawyer even when they are not recorded or reflected in a docu-
ment or tangible thing.252 As it applies to factual investigation, then, the
central thrust of the doctrine is to lower the risks associated with record-
ing, manipulating, and analyzing information, rather than the risks of
learning that information in investigation. Moreover, the absolute or
near-absolute protection afforded to the opinions and mental impressions
of attorneys reflects special concern with encouraging parties to use
attorneys to manipulate and evaluate information in situations where the
privilege would not apply.
The general effect of the conduct encouraged by the immunity is
thus to enhance party preparation, and in particular to enhance the
effects of legal advice across the full range of preparatory activities that
influence the information reaching the tribunal. On balance, our account
predicts that the additional preparation encouraged by the immunity-
whether or not performed by a lawyer253—ought to increase the amount
of information presented, although the overall positive effect should be
most pronounced where the rule causes additional lawyer involvement.
In our account, however, that overall increase in information presented
will reflect a mix of conflicting effects of advice, rather than a simple
increase in investigation. Moreover, to the extent that the immunity
increases investigation, it will do so not by protecting information discov-
ered in investigation from disclosure, but by decreasing the costs of party
efforts to identify and execute appropriate investigative strategies and to
evaluate and manipulate the fruits of those investigations.
Finally, it is important to recall that the social value of work prod-
250. FED. R. Civ. P. 26(b)(3).
251. Id.; Upjohn Corp. v. United States, 449 U.S. 383, 400-02 (1981) (recognizing that such
materials are entitled to “special protection” but declining to decide whether they are absolutely
protected).
252. Shapiro, supra note 45, at 1071.
253. Nonlawyers often have legal knowledge that can improve the presentation of a party’s case.
Even when they do not, their ability to record the results of their investigations without fear that
doing so will ease the opponent’s task should enhance the ability of those who are legally trained to
make effective use of their work. The suggestion by Allen, Grady, Polsby, and Yashko that the
extension of immunity to the product of nonlawyer investigators is difficult to explain except in
terms of their “joint production” theory, is thus inaccurate. Allen, Grady, Polsby & Yashko, supra
note 230, at 391.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
412 CALIFORNIA LAW REVIEW [Vol. 79:313
uct immunity is not established simply by showing that the conduct
caused by granting immunity increases the information reaching the tri-
bunal. One must also ask how much of that preparation would take
place in the absence of the immunity. If the marginal effects are slight,
then they may not be worth the cost of the information lost on account of
the immunity. In this connection it is relevant both that parties have
very strong incentives to prepare their cases even in the absence of an
immunity254 and that an ill-judged statement in an internal memoran-
dum or draft can be capitalized upon by a competent opponent with dev-
astating effects. In Hickman v. Taylor, the Supreme Court appears to
have adopted the view that the latter threat was sufficiently strong that it
would significantly change preparation practices.255 To the extent that
this view is correct, the work product doctrine appears to have substan-
tial social value.
Like our general account of confidentiality, our account of work
product predicts that work product protection will be denied where it
will not significantly increase the net private benefits of recording,
manipulating, and analyzing information. The limitation of the immu-
nity to materials prepared “in anticipation of litigation” is consistent
with this prediction. The deterrent effect of discovery requests from the
opponent is evidently heightened when the recording and analysis of
information take place in the shadow of a ripening controversy with a
specific opponent.256 Our account also predicts that work product pro-
tection will be denied when there is no prospect that the additional prep-
aration facilitated by protection will increase the flow of information to
the tribunal. The rule allowing for contention interrogatories257 and
court decisions allowing for discovery of trial exhibits258 may be consis-
tent with this prediction. A party’s theory of the case and her trial exhib-
254. See Waits, Work Product Protection for Witness Statements: Time for Abolition, 1985 WIs.
L. REV. 305, 331-35 (parties have sufficient incentives to prepare without work product immunity);
Easterbrook, Insider Trading, Secret Agents, Evidentiary Privileges and the Production of
Information, 1981 SUP. CT. REV. 309, 362 (same).
255. See Hickman, 329 U.S. at 511 (Without immunity “much of what is now put down in
writing would remain unwritten” and “[i]nefficiency, unfairness and sharp practices would
inevitably develop in the giving of legal advice and in the preparation of cases for trial.”).
256. Accord Allen, Grady, Polsby & Yashko, supra note 230, at 392. Those authors are,
however, mistaken in suggesting that the limitation to documents and things “prepared in
anticipation of litigation . . . is one of the most important pieces of evidence in support of the joint
production theory.” Id. The same limitation (or something very much like it) should apply in any
account of work product which assumes (1) that the activities encouraged by granting immunity
generally are socially desirable, and (2) that allowing a litigation opponent to discover the results of
those activities significantly reduces their private value.
257. FED. R. CIV. P. 33(b) (allowing interrogatories that call for “an opinion or contention that
relates to fact or the application of law to fact”). Significantly, such interrogatories are normally
disfavored if propounded too early in the trial preparation process. In re Convergent Tech. Sec.
Litigation, 108 F.R.D. 328, 333-38 (N.D. Cal. 1985).
258. Zimmerman v. Superior Court, 402 P.2d 212 (Ariz. 1965).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 413
its both reflect the party’s preparatory efforts and the lawyer’s expertise
in manipulating and evaluating information. When a party has had full
discovery and a fair opportunity to prepare her case, however, prepara-
tory activity will inevitably reach a point of diminishing returns. To pro-
tect the party’s case beyond that point serves only to allow the party to
surprise her opponent. The working assumption of the Federal Rules of
Civil Procedure appears to be that surprise normally reduces the flow of
truthful information to the tribunal and increases the flow of false infor-
mation.259 On that assumption, to extend the protection of a party’s con-
tentions and trial exhibits past the point of diminishing returns to
preparation would reduce rather than increase the information reaching
the tribunal.
C. Restrictions on Advice About Responding to Investigation
Some advice about responding to investigation (or other question-
ing) reduces the information reaching the tribunal. One might therefore
think that it would be desirable to ban or restrict only such advice, while
allowing other litigation advice with more desirable effects.260 It turns
out, however, that it is difficult to prohibit advice about responding to
investigation without encompassing or deterring other advice that has
socially desirable informational or behavioral effects. We employ our
account to explore these issues for both civil and criminal cases.
Advice about responding to investigation (or questioning) can
reduce the unfavorable information reaching the tribunal in three ways.
First, advice about whether information is favorable or unfavorable will
cause the party to withhold more unfavorable information that is unre-
quested or party-controlled. Second, advice can lower the party’s
expected sanctions for withholding information, thus increasing the
amount of information she controls and withholds. Third, advice about
the level of expected evidentiary sanctions can reduce the amount of
unfavorable information the party discloses if parties overestimate those
sanctions.
One might attempt to prevent advice likely to lead to such untoward
259. See, e.g., 8 C. WRIGHT & A. MILLER, supra note 249, ? 2001, at 17 & n.16. The
conclusion that reducing surprise will improve the quality or quantity of information reaching the
tribunal is not obvious. Surprise reduces the amount of time available for an opponent to investigate
the facts underlying the party’s presentation and to prepare a response. It therefore makes it more
likely that the tribunal will receive only the information favorable to the party who benefits from
surprise. For the same reason, surprise reduces the party’s expected sanctions for presenting
favorable false information. But rules permitting surprise may also increase the information
reaching the tribunal by increasing the opponent’s incentive to prepare its own case or by enabling a
party to catch an opponent’s witness in a lie.
260. Kaplow and Shavell, for example, put forth (without advocating) the possibility of
deposing parties before allowing them to consult an attorney. Kaplow & Shavell, supra note 17,
at 613.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
414 CALIFORNIA LAW REVIEW [Vol. 79:313
results by formulating restrictions on the provision of advice that can be
enforced without inquiring into the content of attorney-client communi-
cations.261 Complete restrictions on advice about responding to investi-
gation may, however, also deter other desirable advice. Parties receive a
great deal of advice about whether information is favorable or unfavora-
ble in connection with advice about the conduct of their own investiga-
tions. This advice should increase the amount of information reaching
the tribunal. Parties also receive advice about how to categorize infor-
mation that is integral to desirable advice about settlement and related
primary conduct. To the extent that a ban on harmful advice about
responding to investigation will encompass or deter such desirable
advice, restrictions may have to be limited to special situations where
they can be narrowly targeted and easily enforced. This, as we will show,
is what the law in fact does.
Even if restrictions can be narrowly targeted to advice about
responding to investigation, they may result in several adverse informa-
tional and behavioral effects. First, denying advice to the responding
party will reduce the opponent’s expected sanctions for suppressing
information, and thus increase the opponent’s suppression of information
favorable to the party denied advice. Second, the unadvised party will
recognize, categorize, and present less favorable information divulged in
investigation by the opponent. Third, the lack of advice will increase the
party’s mistaken failure to produce favorable information. Although the
party may later be advised that the omitted information is favorable, sub-
sequent presentation may be infeasible or discounted by the tribunal, par-
ticularly if the party withheld the information by lying. Fourth, if parties
without advice systematically underestimate evidentiary sanctions,
depriving the party of litigation advice will decrease the favorable and
unfavorable information reaching the tribunal.262
The rules regulating the provision of litigation advice reflect an
awareness of these considerations. The easiest place to enforce restric-
tions on advice is at trial. When a witness is actually on the stand, advice
must be provided in public and in the tribunal’s presence. Moreover,
restrictions on advising witnesses can easily be monitored, provided that
they are limited in time. Accordingly, rules governing the examination
of witnesses at trial sharply restrict advice to a testifying witness. Law-
261. Such an approach is ex ante in the sense that it attempts to prevent the legal advice from
ever being provided. Another approach would be ex post in that it would allow legal advice to be
provided but would only punish legal advice on how to reduce the information presented. Such an
ex post approach would require observing attorney-client conferences or relying on reports of what
occurred in them. This would raise issues discussed in supra Section V(B).
262. In addition to these informational effects, the party denied advice will experience an
increase in expected sanctions for desirable withholding and for desirable production that resembles
undesirable withholding.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSAR Y SYSTEM? 415
yers may not ask leading questions of a friendly witness on direct exami-
nation: this effectively forbids advice during direct examination about
whether information is favorable or unfavorable. More important, when
witnesses are responding to cross-examination, lawyers cannot make
“leading” objections that suggest whether particular information is
favorable or unfavorable. In some cases, sequestration orders may bar
lawyers from speaking with a witness on any topic between the time that
the witness takes the oath and the time she is excused.
Restrictions on leading questions avoid many of the adverse infor-
mational effects of more general bans on responding to investigation.
Because the opponent is not conducting the investigation, denying advice
will not increase opponent suppression or reduce recognition of informa-
tion divulged by the opponent. Although the party or witness may mis-
takenly omit favorable information, that effect will be ameliorated by the
attorney’s ability to ask additional nonleading questions. Nor does the
restriction threaten advice about investigation, settlement, or primary
conduct.
Restrictions on advice about responding to cross-examination also
seem likely to generate substantial additional information while avoiding
most adverse effects. Once a witness has testified on direct for the party
who called her, she is likely to have produced almost all the information
favorable to that party. Accordingly, information she has not presented
is likely to be unfavorable. Denying advice about potential testimony on
cross-examination should thus increase the amount of unfavorable infor-
mation reaching the tribunal. The costs of this increase are probably
small. Because cross-examination takes place before the party’s lawyer
and the tribunal, it should be possible to detect and sanction opponent
suppression, to recognize unfavorable information divulged by the oppo-
nent, and to facilitate socially desirable withholding by objecting to ques-
tions that call for privileged information or are unduly coercive. Again,
the reduction will not encompass advice to a party witness about investi-
gation, settlement, or primary conduct.
So long as rules restricting advice to witnesses apply only when the
witness is on the stand, they can be closely tailored to target only advice
with doubtful informational effects. When rules about advice to wit-
nesses extend beyond the time when the witness is on the stand, however,
they pose difficult problems of enforceability and scope, particularly for
party witnesses.
The problem is illustrated by Perry v. Leeke, 263 where the Supreme
Court held that a criminal defendant’s sixth amendment right to counsel
was not violated by a court order that barred the defendant from consult-
263. 488 U.S. 272 (1989).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
416 CALIFORNIA LAW REVIEW [Vol. 79:313
ing with his lawyer during a fifteen-minute break between the end of
his testimony on direct examination and the commencement of cross-
examination. In upholding the order, the Court expressly rejected the
notion that the validity of the order required a showing that the lawyer
was likely to engage in “unethical” conduct.264 Rather it was “simply an
empirical predicate” of the adversary system that cross-examination of
an uncounseled witness is more likely to lead to the discovery of truth.265
It was therefore sufficient to justify the order that legal advice would
enable the party “to regroup and regain a poise and sense of strategy that
the unaided witness would not possess. “266 The Court distinguished its
earlier decision in Geders v. United States, 267 which had invalidated an
order barring all consultation during an eighteen-hour overnight recess,
reasoning that consultation during a longer break would “encompass
matters that go beyond the content of the defendant’s own testimony …
such as the availability of other witnesses, trial tactics, or even the possi-
bility of negotiating a plea bargain.”268 In contrast, during a fifteen-
minute break, “it is appropriate to presume that nothing but the testi-
mony will be discussed.”269
The reasoning in Perry, and its distinction between fifteen-minute
and eighteen-hour recesses, is consistent with the analysis offered here.
Because a party who has completed her direct examination has already
presented her favorable information, the predominant effect of advice
about whether information is favorable or unfavorable or advice about
how to shape her testimony to lower evidentiary sanctions will be a
reduction in unfavorable information reaching the tribunal. This effect
occurs whether or not the advice given to her is technically “unethical”
in the sense that it amounts to “knowing” assistance in perjury. More-
over, on the Court’s assumption that during a fifteen-minute break
between direct and cross-examination the defendant’s testimony would
be the only topic of conversation, the order upheld in Perry posed no risk
of deterring desirable advice about investigation or settlement. In the
case of an overnight eighteen-hour ban on advice, however, the risk of
preventing desirable advice may be increased sufficiently to justify the
holding of unconstitutionality, notwithstanding the risk (which the Court
expressly recognized) that advice given during an overnight consultation
would reduce the amount of information reaching the tribunal just as
effectively as advice given during a fifteen-minute break in testimony.270
264. Id. at 282.
265. Id.
266. Id.
267. 425 U.S. 80 (1976).
268. Perry, 488 U.S. at 284.
269. Id.
270. The Court acknowledged that the socially desirable discussions during an extended recess
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSARY SYSTEM? 417
As Perry suggests, outside the courtroom it is extremely difficult to
craft enforceable restrictions on advice about responding to investiga-
tions without deterring desirable advice. Again the problem is most
acute for party witnesses. Once investigation moves outside the presence
of the tribunal, the opponent’s opportunity and incentive to suppress
favorable information presented by the party increase. The informa-
tional costs of denying advice to the responding party increase accord-
ingly. It also becomes more difficult to prevent parties from obtaining
litigation advice without barring desirable consultations about related
primary conduct, settlement, or the party’s own investigations. This is
particularly true in civil litigation, because parties normally consult law-
yers about whether to abandon, settle, or litigate their claim before an
action is filed. Thus each party will already have received a good deal of
advice before the tribunal is involved in any way. Since many disputes in
which lawyers are consulted settle before filing, the task of monitoring
this advice would be enormous. Once a case has commenced, most par-
ties receive additional advice about settlement and investigation, and
many receive advice about related primary conduct.
Since narrowly targeted restrictions on out-of-court advice would
likely be ineffective and unenforceable and because broader restrictions
would bar too much desirable advice, it makes sense that outside the
courtroom parties are normally free to obtain advice about responding to
investigation-subject only to the normal prohibition against advice that
amounts to knowing assistance in crime or fraud. It is also consistent
with this analysis, however, that the law severely restricts giving such
non-courtroom litigation advice to persons other than the client. It is
often a disciplinary offense and sometimes a crime for a lawyer to pro-
vide accurate advice to a nonparty with the intention of causing the non-
party to withhold or destroy relevant evidence, even where such
withholding or destruction is lawful.271 In such cases, the fact that the
advice about responding to investigation is directed to a nonparty makes
it easier to determine the likelihood that it will reduce the amount of
information reaching the tribunal and also ensures that prohibiting the
advice will not reduce the flow of desirable advice to the lawyer’s client.
Moreover, in most cases the prohibition of advice to the nonparty will
not prevent her from obtaining advice from another, more disinterested
source.
So far as we know, the foregoing examples represent the only cases
in which litigation advice is expressly prohibited due to its informational
“will inevitably include some consideration of the defendant’s ongoing testimony,” but concluded
that this “does not compromise” the defendant’s right to unrestricted access to her lawyer during a
longer recess. Id.
271. See supra note 39.
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
418 CALIFORNIA LAW REVIEW [Vol. 79:313
effects. Our account, however, should also assist in evaluating numerous
other legal rules that encourage or limit advice about responding to
investigation. In civil cases, these include the rule that prohibits a lawyer
from knowingly communicating with a represented party concerning the
subject of the representation without the consent of the party’s lawyer.272
In criminal cases, they include fifth and sixth amendment rules that gov-
ern the existence, invocation, and waiver of the criminal defendant’s
right to counsel before and after indictment.273
Depending on the motivations and knowledge of the parties and the
amount of information that is party-controlled or suppressible, such rules
may have important informational effects. For example, if most criminal
defendants routinely overestimate how favorable their information is274
or overestimate the sanctions for keeping silent in the face of custodial
interrogation,275 rules that make it easy to waive the pre-indictment right
to counsel under the fifth amendment will increase the party-controlled
unfavorable information reaching the tribunal.276 This should generally
hurt the innocent less than the guilty. On the other hand, if police offi-
cials conducting a custodial investigation are willing and able to suppress
information favorable to the defendant and a substantial amount of the
information likely to be uncovered in a custodial investigation is sup-
pressible,277 a relaxed waiver rule may reduce the government’s expected
272. See MODEL RULES, supra note 1, Rule 4.2; MODEL CODE, supra note 2, DR 7-104(A)(1).
273. Under the fifth amendment, a party in custodial interrogation must be advised of her right
to counsel, Miranda v. Arizona, 384 U.S. 436, 467 (1966), and cannot be interrogated if she invokes
that right, Edwards v. Arizona, 451 U.S. 477, 484 (1981). Under the sixth amendment, a defendant
has a right to counsel during any post-indictment interrogation, Massiah v. United States, 377 U.S.
201, 206 (1964), or post-indictment lineup, United States v. Wade, 388 U.S. 218, 237 (1967).
274. Indeed, even if party errors are unbiased, advice given in connection with investigation that
corrects the suspect’s mistakes about whether information is favorable or unfavorable may reduce
the unfavorable information presented. If the party under investigation mistakenly believes that her
information is, on balance, unfavorable, so that she claims the privilege, advice about responding to
investigation would cause her to change her decision. Unless later presentation of the information
will be infeasible or discounted, however, this will affect only the timing of her decision to produce,
since the lawyer appointed for her later in the case will also counsel her to disclose the information.
Ill-advised decisions to present unfavorable information, however, may well be more difficult to
reverse. Accordingly, unless favorable information that is withheld during an initial interview will
be discounted, denying advice about whether information is favorable or unfavorable will increase
the flow of information to the tribunal.
275. Under the line of cases following Griffin v. California, 380 U.S. 609 (1965), and Doyle v.
Ohio, 426 U.S. 610 (1976), the government’s right to make use of silence following Miranda
warnings, even for impeachment purposes, is extremely limited. Thus silence, at least for the period
required to consult a lawyer, is essentially costless. It therefore seems quite likely that
overestimating the costs is more common than underestimation, and that allowing easy waivers of
counsel will have the positive effects described in text.
276. See generally Stuntz, Waiving Rights in Criminal Procedure, 75 VA. L. REV. 761 (1989)
(discussing the effects of waiver).
277. Concern with the government’s opportunity and motive to suppress information favorable
to the defendant has sometimes figured in the Court’s decisions concerning the right to counsel in
pretrial investigation. For example, in United States v. Ash, 413 U.S. 300 (1973), the Supreme
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
1991] DO LAWYERS IMPROVE THE ADVERSAR Y SYSTEM? 419
sanctions for suppressing that information. That effect would reduce the
amount of favorable information reaching the tribunal, harming the
innocent more than the guilty. While the outcome of these calculations
cannot by itself determine the social desirability of the waiver rule cho-
sen, its relevance is clear.
CONCLUSION
In a liberal society, it is widely believed that it is desirable, indeed
essential, for citizens to know the law.278 We have attempted to deter-
mine the conditions under which that belief might be justified for legal
advice about the litigation of factual disputes in an adversary system.
While the matter ultimately turns on difficult empirical questions, our
overall conclusion is positive. Much litigation advice appears to have
desirable informational and behavioral effects, and advice with undesir-
able or ambiguous effects is often difficult to regulate without deterring
desirable advice. Our positive general conclusion is striking because it is
derived despite the adoption of the cynical premise that litigants and law-
yers recognize no duty to obey the law, and indeed suppress or withhold
information whenever it furthers their own interests. If parties are gener-
ally more law-abiding or truthful than our basic account assumes, then
the effects of advice may be even more desirable.
Our conclusions, however, are richer and more refined than this
general statement suggests. Indeed, a principal lesson of our analysis is
that one cannot make responsible judgments about the desirability of
legal advice or its regulation without attention to multiple and mixed
effects. The complexity of our account stands as a rebuke to both sim-
plistic defenders and simplistic critics of advice who reach sweeping
judgments about its general desirability based on partial descriptions of
its effects. In many cases, our account suggests hitherto unrecognized
weaknesses and complications in arguments that have classically played
a central role in the debate about the adversary system.
Defenders of advice, for example, frequently point to the value of
Court held that the sixth amendment did not entitle defense counsel to attend post-indictment
sessions where a witness sought to identify the defendant from an array of photographs. In
distinguishing such sessions from post-indictment lineups (where the right to counsel does apply) the
Court argued that the defense counsel’s opportunity to cross-examine the persons involved in the
photo session at trial could “serve as a substitute for counsel at the pretrial confrontation.” Id. at
316; see also id. at 324 (Stewart, J., concurring) (arguing that improper influence from use of
photographs can be “readily reconstructed at trial”). This arguably reflects the judgment either that
information cannot be effectively suppressed in a photographic lineup, or that expected sanctions for
suppression were already sufficiently high to deter suppression without the presence of counsel. The
majority also stressed that the “ethical responsibility of the prosecutor” would tend to minimize
exploitation of opportunities to suppress. Id. at 320.
278. See L. FULLER, THE MORALITY OF LAW 49-51 (1964).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
420 CALIFORNIA LAW REVIEW
competition in improving the presentation of information to the tribu-
nal.279 Our account suggests that legally advised competition between
the parties will often have that effect, especially in the case of nonsup-
pressible information available to both parties. But competition can
sometimes result in a reduction of information reaching the tribunal.
Whether advice increases or decreases the competitive presentation of
information turns out to depend on empirical questions about whether
information is suppressible, who has access to it, and the misconceptions
entertained by unadvised parties. Defenders of advice would do well to
look more closely at those questions.
Critics of the system, on the other hand, frequently stress the harm-
ful or doubtful effects of advice that is not balanced between the par-
ties.280 Our account confirms that when both parties have comparable
advice, the positive effects of advice are more robust.281 It also confirms
that imbalances may have harmful or doubtful effects, particularly
because they increase the ability of the more knowledgeable party to
present false or prejudicial information and to impose costs on her oppo-
nent. But we also show that even unbalanced advice has a strong general
tendency to benefit those who have acted desirably more than it benefits
those who have acted undesirably. This will most clearly be socially
desirable if the tribunal can make appropriate adjustments to its rules of
decision. In a substantial range of cases, imbalances of representation
may therefore produce results superior to those that would be possible if
both parties were denied advice.
Beyond providing a more sophisticated understanding of general
claims about the effects of advice, our account also explains many of the
problems that arise in regulating it. It points to some significant classes
of cases where broad restrictions on advice may be socially desirable. It
also identifies several narrowly limited classes of advice, including advice
to testifying witnesses and advice to nonclient witnesses, which are so
likely to be harmful and so easily targeted that rules barring such advice
may well be socially desirable. We also provide a clear framework for
assessing rules of attorney-client confidentiality, which accurately identi-
fies the nature of the tradeoffs required by such rules and the empirical
questions that determine whether those tradeoffs will be worthwhile.
279. E.g., R. POSNER, supra note 15, at 491-93.
280. See, e.g., sources cited supra notes 10-13.
281. See supra Section II(D).
This content downloaded from
������������73.211.34.7 on Mon, 14 Aug 2023 01:04:27 +00:00�������������
All use subject to https://about.jstor.org/terms
Contents
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
Issue Table of Contents
California Law Review, Vol. 79, No. 2 (Mar., 1991), pp. 267-590
Front Matter [pp. 497-497]
Freedom of Expression and the Civic Republican Revival in Constitutional Theory: The Ominous Implications [pp. 267-311]
Do Lawyers Improve the Adversary System? A General Theory of Litigation Advice and Its Regulation [pp. 313-420]
Back to the Future and up to the Sky: Legal Implications of “Open Skies” Inspection for Arms Control [pp. 421-496]
Comments
Can a Pregnant Woman Morally Refuse Fetal Surgery? [pp. 499-540]
Rejecting Unreasonable Sexual Expectations: Limits on Using a Rape Victim’s Sexual History to Show the Defendant’s Mistaken Belief in Consent [pp. 541-576]
Book Review
Review: untitled [pp. 577-589]
Back Matter
Defining “Legal Loopholes”
A legal loophole is a provision or ambiguity in a law, regulation, or contract that allows an
individual or organization to exploit a technicality or gap to achieve a result that may be
unintended by the drafters, often without technically breaking the law. Loopholes often arise
due to vagueness, imprecision, outdated language, or conflicts within statutes or regulations,
and they are typically closed when they are detected and seen as undermining the law’s intent.
Examples of Legal Loopholes
1. Tax Avoidance: Companies shifting profits to jurisdictions with lower taxes by using
complex international structures (e.g., the “Double Irish with a Dutch Sandwich” tax
scheme).
2. Campaign Finance: Using “Super PACs” to indirectly support political candidates,
bypassing direct donation limits imposed on individual campaigns.
3. Airbnb Short-Term Rentals: Some landlords rent properties for short stays through
platforms like Airbnb despite zoning laws or lease agreements meant to prevent
commercial subletting.
4. Bankruptcy and Asset Protection: Placing assets in irrevocable trusts or offshoring them
before declaring bankruptcy to protect wealth from creditors.
5. Corporate Veil Piercing Avoidance: Business owners may structure businesses as
separate legal entities to shield themselves personally from liability, even if ethically
questionable, unless specific conditions for “piercing” the corporate veil are met.
6. Patent Law Gaming: Filing a series of narrow patents or engaging in “evergreening” to
maintain monopoly protection over a pharmaceutical product, extending its market
exclusivity.
7. Parking Ticket Loopholes: Loopholes in city ordinances sometimes allow tickets to be
contested successfully based on the wording or gaps in enforcement policy.
8. Mandatory Sentencing Loopholes: Sentencing reductions or special exceptions in
criminal justice statutes may allow some offenders to escape harsher punishments when
technicalities exist in the wording.
9. Anti-Money Laundering (AML): Exploiting gaps in reporting requirements in specific
financial transactions to circumvent AML regulations.
You Try … Hypothetical Questions for Students
1. Tax Loophole Scenario: A company restructures its operations by creating a subsidiary in a
low-tax jurisdiction, funneling much of its global profits through this subsidiary to minimize
overall tax liability. The restructuring adheres to current tax laws but reduces tax revenues
significantly.
o Question: Is this exploitation of a tax loophole ethical, even if it’s legal? What
could be done legislatively to address such tax avoidance schemes?
o Model Answer: The ethical dimension depends on one’s perspective on
corporate social responsibility and tax fairness. Many would argue it’s unethical
as it undermines equitable tax distribution and deprives governments of
essential revenue. Legislative fixes could include closing known loopholes
through international cooperation (e.g., global minimum tax agreements),
enforcing stricter residency requirements, or modifying profit-shifting rules.
2. Zoning Loophole Hypothetical: A homeowner begins renting out their property on a short-
term basis, despite residential zoning prohibiting such use. The homeowner claims that the local
ordinance’s language does not explicitly define short-term rental or apply to online services.
o Question: What arguments might the homeowner raise to defend their position,
and how could the municipality close this loophole?
o Model Answer: The homeowner may argue ambiguity in zoning definitions and
reliance on implied permissions within existing ordinances. The municipality
could close the loophole by updating zoning definitions, explicitly regulating
online platforms, and providing enforcement mechanisms tailored to short-term
rental activity.
3. Patent Evergreening Case: A pharmaceutical company files multiple minor modification
patents for an existing drug, each extending its exclusivity period, despite minimal therapeutic
improvements.
o Question: Discuss the pros and cons of this practice from the perspectives of
innovation policy and public health. Should patent law be revised to prevent
“evergreening”?
o Model Answer: Proponents argue that incremental innovations can benefit
patients and should be protected. Critics highlight that evergreening exploits a
loophole to stifle competition, maintaining high drug prices. Reform might
include limiting secondary patents for minimal improvements or creating stricter
standards for demonstrating genuine innovation.
4. Bankruptcy Asset Loophole: Before declaring bankruptcy, an individual transfers ownership
of their home to a trust and claims it cannot be used to satisfy debts.
o Question: What legal doctrines might be invoked to challenge this transfer, and
how can such loopholes be prevented in bankruptcy law?
o Model Answer: Doctrines like “fraudulent transfer” or “constructive fraud”
might be invoked to invalidate the transfer if done to frustrate creditor claims.
Closing this loophole could involve stricter transfer scrutiny within a defined
period before bankruptcy filings or heightened trustee investigatory powers.
