# Reputation Failure: The Limits of Market Discipline in Consumer Markets

Canonical page: https://works.battleoftheforms.com/papers/ssrn-3239995/

[p. 1]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
REPUTATION FAILURE: THE LIMITS OF MARKET
DISCIPLINE IN CONSUMER MARKETS
Yonathan A. Arbel*
Many believe that consumer-sourced reputational
information about products should increasingly replace topdown regulation. Instead of protecting consumers through
coercive laws, reputational information gleaned from the
wisdom of the crowd would guide consumer decision-making.
There is now a growing pressure to deregulate in diverse
fields such as contracts, products liability, consumer
protection, and occupational licensing.
This Article presents a common failure mode of systems
of reputation: “Reputation Failure.” By spotlighting the
public-good nature of reviews, rankings, and even gossip, this
Article shows the mismatch between the private incentives
consumers have to create reputational information and its
social value. As a result of this divergence, reputational
information is beset by participation, selection, and social
desirability biases that systematically distort it. This Article
argues that these distortions are inherent to most systems of
reputation and that they make reputation far less reliable
than traditionally understood.
The limits of reputation highlight the centrality of the
law to the future of the marketplace. Proper legal institutions
can deal not only with the symptoms of reputation failure—
consumer mistakes—but improve the flow and quality of
reputational information, thus correcting reputation failures
before they arise. This Article offers a general framework and
*. Assistant Professor of Law, University of Alabama School of Law. For
useful comments and conversations, I thank Oren Bar-Gill, Lisa Bernstein,
Alfred Brophy, Shahar Dillbary, Janet Freilich, Brian Galle, John Goldberg,
Michael Heller, Richard Hynes, Louis Kaplow, Daniel Klarmen, Ronald
Krotoszynski, Irina Manta, Murat Mungan, Nicholas Marquiss, Michael Pardo,
Gregg Polsky, Barak Richman, Ken Rosen, Roy Shapira, Steve Shavell, Henry
Smith, Andrew Tuch, Fred Vars, and Rory Van Loo. I am also thankful to
participants at the American, Midwestern, and European Law & Economics
Conferences, Contracts Conference XIII, and workshop participants at the
Universities of Alabama, Bar-Ilan, and Chicago. The editors of the Wake Forest
Law Review provided many thoughtful suggestions. For excellent research
assistance, I thank Hamilton Millwee, Victoria Moffa, Kenton McGilliard, and
Brenton Smith.

[p. 2]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1240 WAKE FOREST LAW REVIEW [Vol. 54
explores a number of strategies. A more robust system of
reputation can preserve consumer autonomy without
sacrificing consumer welfare.
TABLE OF CONTENTS
I. INTRODUCTION ...................................................................... 1240
II. LAW VS. MARKETS AND ASYMMETRIC INFORMATION ............ 1246
A. Law vs. Markets ............................................................. 1246
B. The Supposed Reliability of Reputation in Legal
Thought .......................................................................... 1251
III. REPUTATION FAILURE: MICROFOUNDATIONS, DISTORTIONS,
AND SOCIAL WELFARE ........................................................... 1254
A. The Microfoundations of Reputation ............................ 1256
1. The Costs of Gossip .................................................. 1256
2. Internal Drives: On Spite and Gratitude ................ 1256
3. Social Pressures and Herd Behavior ....................... 1258
4. Material Incentives: Shilling and Cherry-Picking . 1261
B. Reputational Distortions ............................................... 1262
1. Reputational Sluggishness ...................................... 1263
2. Regression to the Extreme ........................................ 1265
3. Reputation Integrity ................................................. 1269
C. Flawed Information, Flawed Decisions ........................ 1270
1. Informational Distortions ........................................ 1271
2. Overcoming Bias ...................................................... 1275
IV. LEGAL IMPLICATIONS OF REPUTATION FAILURE .................. 1286
A. Reputation Failure and Contemporary Debates in
Contracts and Torts ....................................................... 1286
B. Reputation-by-Regulation ............................................. 1287
1. Leveraging Market Players: The Role of
Reputational Platforms ............................................ 1288
a. Regulating Platforms ........................................ 1292
b. Policing Platforms ............................................. 1293
c. Platform Accreditation ...................................... 1293
2. Professional Publications ........................................ 1294
3. Fighting Fake Reviews ............................................. 1295
4. Fostering Positive Incentives ................................... 1297
5. Controlling Costs: First Amendment and
Reputation ................................................................. 1299
V. CONCLUSION ......................................................................... 1303
I. INTRODUCTION
How much trust should we place in consumer-sourced
reputational information? This Article develops the argument that
systems of reputational information are subject to a number of
distortions that limit the reliability of reputational information. As a
result, trusting these systems to replace the law should be done with

[p. 3]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1241
great caution. The source of these distortions and the role of the law
in addressing them are the key themes developed here.
Some of the most important debates in contract law involve a
basic dilemma: to what extent can markets be trusted to regulate
themselves?1 One reason why regulation may be needed is
asymmetric information—if sellers know more, they can exploit
buyers and promise high but deliver low. A mitigating factor, which
counsels against less regulation, is reputation. Reputation
information, once the province of small-knit communities, allows
parties to develop trust based on self-interest.2 If the seller cheats,
her reputation will suffer, costing her opportunities to deal with other
buyers. In the last two decades, reputational information has
permeated almost all aspects of consumer markets, online and offline.
Through the use of rankings, reviews, and stars, reputation facilitates
transactions between complete strangers.
The explosion of reputational information has instilled a sense of
optimism among many that the end of asymmetric information is
nigh.3 Why regulate markets, the argument goes, if consumers can
easily know in advance which seller is honest, which product is best,
and which service provider is most reliable? The increased trust in
1. The debate on law versus markets assisted by reputation is longstanding. For example, Milton Friedman argued that “consumers do not have to
be hemmed in by rules and regulation . . . because they are protected by the
market itself.” Milton Friedman’s Free to Choose: Who Protects the Consumers?
(PBS television broadcast Jan. 11, 1980), https://www.freetochoosenetwork.org
/programs/free_to_choose/index_80.php?id=the_power_of_the_market; see also
ADAM SMITH, LECTURES ON JURISPRUDENCE 327 (R.L. Meek et al. eds., 1978)
(“When a person makes perhaps 20 contracts in a day, he cannot gain so much
by endeavoring to impose on his neighbors, as the very appearance of a cheat
would make him lose.”); Lior Jacob Strahilevitz, Less Regulation, More
Reputation, in THE REPUTATION SOCIETY 71 (Hassan Massum & Mark Tovey eds.,
2012) (arguing that a world with strong reputational information has
“diminished need for regulatory oversight and legal remedies”); Rory Van Loo,
Helping Buyers Beware: The Need for Supervision of Big Retail, 163 U. PA. L. REV.
1311, 1347 (2015) (“One common argument in consumer protection is that
reputational concerns will stamp out many bad practices, thus making some
regulations unnecessary.”). On these notions, see infra Subpart II.A.
2. See e.g., Lisa Bernstein, Contract Governance in Small-World Networks:
The Case of the Maghribi Traders, 113 NW. L. REV. 1009, 1009 (2019) (tracing
reputational flows in complex trade networks); Lisa Bernstein, Opting Out of the
Legal System: Extralegal Contractual Relations in the Diamond Industry, 21 J.
LEGAL STUD. 115, 152 (1992) [hereinafter Bernstein, Opting Out] (analyzing the
behavior of Jewish Orthodox diamond traders in New York); Barak D. Richman,
An Autopsy of Cooperation: Diamond Dealers and the Limits of Trust-Based
Exchange, 9 J. LEGAL ANALYSIS 247, 247–50 (2017) (exploring how market
perturbations lead to the decline of trust-based institutions). Recent trends also
include the scoring of consumers themselves, see Yonathan A. Arbel & Roy
Shapira, Theory of the Nudnik: The Future of Consumer Activism and What We
Can Do to Stop It, VAND. L. REV. (forthcoming 2020).
3. Alex Tabarrok & Tyler Cowen, The End of Asymmetric Information,
CATO UNBOUND (Apr. 6, 2015), https://www.cato-unbound.org/2015/04/06/alextabarrok-tyler-cowen/end-asymmetric-information.

[p. 4]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1242 WAKE FOREST LAW REVIEW [Vol. 54
reputation has galvanized support for deregulatory policies from
conservatives and liberals alike.4 Various scholars have made calls
to abolish consumer protections in contracts, torts, and occupational
licensing.5 The Trump Administration has effectively defanged the
Consumer Financial Protection Bureau and otherwise stalled many
regulatory interventions in markets.6
This trust in reputation-based market ordering overlooks a key
feature of reputation: it is a public good.7 Through gossip, word-of4. See ARUN SUNDARARAJAN, THE SHARING ECONOMY: THE END OF
EMPLOYMENT AND THE RISE OF CROWD-BASED CAPITALISM 138 (2016) (“Eventually,
peer-to-peer platforms may provide a basis upon which society can develop more
rational, ethical, and participatory models of regulation.”); Benjamin G. Edelman
& Damien Geradin, Efficiencies and Regulatory Shortcuts: How Should We
Regulate Companies Like Airbnb and Uber?, 19 STAN. TECH. L. REV. 293, 300
(2016) (“By all indications, reputation systems are serving the intended
purpose.”); Christopher Koopman et al., The Sharing Economy and Consumer
Protection Regulation: The Case for Policy Change, 8 J. BUS. ENTREPRENEURSHIP
& L. 529, 530 (2015); Adam Thierer et al., How the Internet, the Sharing Economy,
and Reputational Feedback Mechanisms Solve the “Lemons Problem,” 70 U.
MIAMI L. REV. 830, 830–31 (2016). But see Sofia Ranchordás, Does Sharing Mean
Caring? Regulating Innovation in the Sharing Economy, 16 MINN. J.L. SCI. &
TECH. 413, 414 (2015) (critiquing aspects of the sharing economy); Abbey Stemler,
Feedback Loop Failure: Implications for the Self-Regulation of the Sharing
Economy, 18 MINN. J.L. SCI. & TECH. 673, 673 (2017) (highlighting the issues
posed to the sharing economy by flawed reputational mechanisms and offering
regulatory solutions); Chris Nosko & Steven Tadelis, The Limits of Reputation in
Platform Markets: An Empirical Analysis and Field Experiment 1 (Nat’l Bureau
of Econ. Research, Working Paper No. 20830, 2015) (considering dynamic biases
resulting from buyers who leave the marketplace without producing reviews).
5. See e.g., Oren Bar-Gill, The Behavioral Economics of Consumer
Contracts, 92 MINN. L. REV. 749, 756 (2008); Albert H. Choi & Kathryn E. Spier,
Should Consumers be Permitted to Waive Products Liability? Product Safety,
Private Contracts, and Adverse Selection, 30 J.L. ECON. ORG. 734, 755 (2014);
Richard A. Epstein, Behavioral Economics: Human Errors and Market
Corrections, 73 U. CHI. L. REV. 111, 120 (2006); A. Mitchell Polinsky & Steven
Shavell, The Uneasy Case for Product Liability, 123 HARV. L. REV. 1437, 1449
(2010); Alan Schwartz & Robert E. Scott, Contract Theory and the Limits of
Contract Law, 113 YALE L.J. 541, 557 (2003) (“[s]tate enforcement
of . . . agreements is unnecessary when the agreements . . . can be enforced with
reputational sanctions.”); Stephen D. Sugarman, Doing Away with Tort Law, 73
CALIF. L. REV. 555, 564 (1985); see also THOMAS SZASZ, OUR RIGHT TO DRUGS: THE
CASE FOR A FREE MARKET (1992) (arguing for deregulation of access to drugs);
Walter Gellhorn, The Abuse of Occupational Licensing, 44 U. CHI. L. REV. 6, 6
(1976).
6. See, e.g., Renae Merle & Tracy Jan, Trump is Systematically Backing Off
Consumer Protections, to the Delight of Corporations, WASH. POST (Mar. 6, 2018),
https://www.washingtonpost.com/business/economy/a-year-of-rolling-backconsumer-protections/2018/03/05/e11713ca-0d05-11e8-95a5-c396801049ef
_story.html?noredirect=on; Tracking Deregulation in the Trump Era, BROOKINGS
(Sept. 23, 2019), https://www.brookings.edu/interactives/tracking-deregulationin-the-trump-era/ (tracking ninety-six areas where there are deregulatory
attempts).
7. Technically, reputation is a public good because it is neither excludable
nor is its consumption rivalrous. See Tyler Cowen, Public Goods, in CONCISE

[p. 5]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1243
mouth, online reviews, and product ranking, consumers create a body
of reputational information covering innumerable products and
services—from restaurants and keychains to doctors and car
mechanics.8 This information is then used by future consumers to
guide their own decision-making, but the original creators of this
information are rarely, if ever, compensated for their efforts.9 That
is, while the costs of creating reputational information are private,
the benefits are public. Observing the divergence of private and
public costs presents a deep puzzle for all systems of reputation: Who
chooses to participate in the creation and dissemination of
reputational information, why, and to what effect?
In spotlighting this puzzle and exploring its consequences, this
Article identifies a common failure mode of reputational information,
called “reputation failure.” Prospective consumers use reputational
information to learn about the experiences of a representative sample
of similarly situated consumers. This kind of “poll” could inform the
consumer about the expected quality of service, the frequency of
errors, and the honesty of the seller. However, this poll is subject to
three confounding factors: sluggishness, regression to the extreme, and
an integrity bias. Reputational information is sluggish, i.e., slow to
develop, because sharing consumers are not sufficiently incented to
share their experiences. The motivations to share are not only weak,
but they are also asymmetric; psychologically, individuals are more
inclined to share information when they had a very positive or
negative experience. Thus, reputational information tends to “regress
to the extremes,” or develop over time in a way that overly emphasizes
extreme experiences at the expense of middling ones. And even those
experiences that are shared are not always authentic: social and
financial motivations to share lead some individuals to misstate their
experiences in ways that put themselves in a better light or otherwise
favor them. This leads to an integrity bias.
In the presence of reputation failure, dishonest sellers can thrive,
not just in the short term but also over longer spans of time.
Reputation failure thus suggests the limits of market discipline
through reputational information.10
ENCYCLOPEDIA OF ECONOMICS (Lauren F. Landsburg et al. eds., 2008); Agnar
Sandmo, Public Goods and Pigouvian Taxes, in PALGRAVE ENCYCLOPEDIA OF
ECONOMICS 10,975 (2018).
8. Daniel B. Klein, Knowledge, Reputation, and Trust, by Voluntary Means,
in REPUTATION: STUDIES IN THE VOLUNTARY ELICITATION OF GOOD CONDUCT 1, 3
(Daniel B. Klein ed., 1997) (“[T]he ‘invisible eye’ often functions by virtue of very
audible tongues.”).
9. Id. at 1; see also Eric Goldman, The Regulation of Reputational
Information, in THE NEXT DIGITAL DECADE: ESSAYS ON THE FUTURE OF THE
INTERNET 293, 301 (Berin Szoka & Adam Marcus eds., 2010) (noting the
“inadequate production incentives” of reputational information).
10. See Yannis Bakos et al., Does Anyone Read the Fine Print? Consumer
Attention to Standard-Form Contracts, 43 J. LEGAL STUD. 1, 2 (2014) (“Defenders

[p. 6]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1244 WAKE FOREST LAW REVIEW [Vol. 54
These concerns with reputation failure are consistent with some
important trends in the empirical data.11 Amazon is a case in point;
despite the cornucopia of products listed there, reviews follow an
unusual distribution.12 One might expect that among so many
products, some reviews would be exciting and others disappointing,
but that the majority would be middling. Evidence from millions of
products contradicts this expectation; reviews concentrate in the
extremes and are scarce in the middle.13 Additional evidence
suggests that the properties of the products themselves do not drive
this unusual distribution. For example, there is little agreement—
quite often, disagreement—between the rating of the same products
by professionals and consumers.14 There is also little agreement
among consumers on different platforms regarding the same
products.15 And when consumers are asked to rate products in lab
settings, their reviews show a remarkably different distribution.16
Despite reputation failures, consumers reportedly rely on
reputational information. In a recent survey, 82 percent of American
adults said they sometimes or always read reviews before making
new purchases, and more than two-thirds of those who routinely use
reviews described them as “generally accurate.”17 Similarly, a survey
of online users found that, on average, users rated the credibility of
the last review they read as a 4.2 out of five, or roughly 84 percent,
on average.18
Not all reputation failures are severe but ignoring the risk of
failure is a serious omission. In some cases, sophisticated consumers
might be able to mitigate part of the distortionary effect of reputation
failure by interpreting reputational information using a combination
of freedom of contract have generally rejected intervention by relying on
reputational constraints.”).
11. See infra Subpart III.B for a more general discussion of the evidence.
12. See infra note 171.
13. At least, one would expect a unimodal distribution, but this is
contradicted in the data. See infra Subpart III.B.
14. Bart de Langhe et al., Navigating by the Stars: Investigating the Actual
and Perceived Validity of Online User Ratings, 42 J. CONSUMER RES. 817, 821
(2016) (studying correlations between online reviews and scores provided by the
magazine Consumer Reports and finding that “[t]he average correlation is 0.18,
and 34% of correlations are negative”).
15. See Georgios Zervas et al., A First Look at Online Reputation on Airbnb,
Where Every Stay is Above Average 10 (Jan. 28, 2015) (unpublished manuscript),
https://papers.ssrn.com/sol3/papers2.cfm?abstract_id=2554500 (finding that the
ranking on Airbnb only explains 17 percent of the correlation (R2) between crosslisted properties on TripAdvisor).
16. See infra Subpart II.B.2.
17. Aaron Smith, Online Reviews, PEW RESEARCH CTR. (Dec. 19, 2016),
http://www.pewinternet.org/2016/12/19/online-reviews/ (explaining that even
among the general population, 51 percent of US adults described reviews as
generally giving an accurate picture).
18. Cindy Man-Yee Cheung et al., Is This Review Believable?, 13 J. ASS’N.
INFO. SYS. 618, 624 (2012).

[p. 7]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1245
of heuristics, statistical analysis, multisource analysis, and
experience.19 However, it is important to recognize that the power of
these methods is limited; there is only so much signal that can be
extracted from a biased and noisy sample.20 To illustrate this claim,
I employ a method known as a Monte-Carlo simulation, which
illustrates the limits of such heuristics.21
Like other forms of market failure, the existence of reputation
failure has various legal implications. Most directly, reputation
failures call for greater scrutiny of consumer transactions and stricter
regulation of product safety and quality. Such regulation can come in
the form of mandatory warranties, broader disclosure obligations,
good-faith requirements, product liability duties, etc.22
Legal interventions, however, need not be limited to the
consequences of reputation failure. The law can also improve the
quality of reputation itself, thus avoiding the failure of reputation in
the first place. To this end, I propose here a new framework of
synthesizing legal institutions and markets, called Reputation-byRegulation. Policymakers can significantly improve consumer
welfare while preserving consumer autonomy by focusing on
designing rules that improve and increase the flow of reliable
reputational information to the market. Building channels through
which reputational information can effectively flow to the market can
solve reputation failure and allow consumers to choose freely and
effectively for themselves. To guide future policymaking, this Article
illustrates Reputation-by-Regulation through five concrete types of
effective legal interventions.
The analysis in this Article should also inform economic analysis
more generally. It is very common, even in leading economic models,
to assume that reputation is an inherent feature of the market.23
Sellers sell, buyers buy, and reliable reputational information
miraculously emerges.24 These studies could benefit from explicit
19. See, e.g., id. at 627-28.
20. The problem combines low-response rate, self-selection bias, middlecensoring, and middle-truncation. For a statistical analysis of some of these
issues, see S. Rao Jammalamadaka & Srikanth K. Iyer, Approximate Self
Consistency for Middle-Censored Data, 124 J. STAT. PLAN. & INFERENCE 75, 76-85
(2004).
21. For examples in law, see Dan M. Kahan et al., Whose Eyes Are You Going
to Believe? Scott v. Harris and the Perils of Cognitive Illiberalism, 122 HARV. L.
REV. 837, 870-76 (2009).
22. See Ranchordás, supra note 4, at 459–61 (discussing regulations imposed
and suggested within the sharing economy); see also Stemler, supra note 4, at
703–11 (offering additional regulatory solutions).
23. See Simon Board & Moritz Meyer-Ter-Vehn, Reputation for Quality, 81
ECONOMETRICA 2381, 2384–86 (2013) (discussing the treatment of reputation in
various economic models).
24. See, e.g., David S. Ardia, Reputation in a Networked World: Revisiting
the Social Foundations of Defamation Law, 45 HARV. C.R.-C.L. L. REV. 261,
267―68 (2010) (“Reputation is an emergent property of social interactions.”). In

[p. 8]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1246 WAKE FOREST LAW REVIEW [Vol. 54
recognition of the “microfoundations” of reputation—the incentives
that lead individuals to share their experiences with others, and their
consequences for the reliability of reputational information.25
This Article unfolds in five parts. Part II explores the role of
reputation in our thinking about the proper scope of intervention in
consumer markets. It highlights how central is the idea that
reputation is reliable. Part III develops the “microfoundations” of
reputation. It surveys the motivations to produce reputational
information and explains how these lead to systemic bias. Finally,
Part IV explores the legal-market interface and the use of the law to
improve reputational information flows.
II. LAW VS. MARKETS AND ASYMMETRIC INFORMATION
This Part surveys debates on regulation, deregulation, and the
role of reputation. It shows how critical reputation is to these debates
and yet how very little attention is given to the reliability of
reputational information. As this Article will show, imperfections
and failures in reputational information undermine the validity and
persuasiveness of many market self-ordering arguments.
A. Law vs. Markets
Most commercial transactions involve asymmetric information
between sellers and buyers. Buying a refrigerator or a car, hiring a
contractor, and seeking a financial advisor are all quotidian
transactions that involve what economists call “experience” goods—
i.e., goods where the consumer can only observe quality after
consumption or usage.26 The concern with such transactions is that
the recent edition of the Palgrave Dictionary of Economics, there are early signs
of recognition of these problems: “[T]he literature on this area [of reputation in
large groups] is in its infancy; very little can be said with much certainty now.”
Martin W. Cripps, Reputation, in THE NEW PALGRAVE DICTIONARY OF ECONOMICS
11,569 (2018); see also infra Subpart II.B.
25. See, e.g., Roy Shapira, Reputation Through Litigation: How the Legal
System Shapes Behavior by Producing Information, 91 WASH. L. REV. 1193, 1203
(2016) (describing the perception of reputation in the scholarship as the product
of a simple process). This reductive view uncritically assumes, with only a few
exceptions, that reputation information is reliable. For prominent examples in
economics, see, e.g., Heski Bar-Isaac & Steven Tadelis, Seller Reputation, 4
FOUND. & TRENDS IN MICROECONOMICS 273, 282 (2008) (reviewing the literature);
Board & Meyer-Ter-Vehn, supra note 23, at 2386–87. A leading paper in
evolutionary psychology argues that reputation is an evolutionary solution to the
tragedy of the commons with respect to public goods, not acknowledging that the
creation of reputation itself suffers from the same problem. See Manfred
Milinsky et al., Reputation Helps Solve the ‘Tragedy of the Commons’, 415 NATURE
424, 424 (2002); see also Martin A. Nowack & Karl Sigmund, Evolution of Indirect
Reciprocity, 437 NATURE 1291, 1291 (2005) (noting the evolutionary roots of
reputation).
26. See Gary T. Ford et al., An Empirical Test of the Search, Experience and
Credence Attributes Framework, in 15 ADVANCES IN CONSUMER RESEARCH 239,

[p. 9]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1247
they invite opportunistic behavior, as sellers might take advantage of
the information asymmetry by promising high but delivering low.27
This possibility, left unchecked, would lead to misallocation of
resources, abuse of buyers, and increased buyer trepidation, stalling
economic activity overall.28
The conventional response to such problems is direct regulation.
To give but a few examples, the law sets ex ante minimum quality
regulations;29 mandates price controls;30 imposes good-faith duties;31
compels mandatory disclosure;32 creates implied warranties of
merchantability and fitness;33 requires licensing, training, and
testing;34 enforces truth-in-advertising requirements;35 and imposes
tort and criminal liability for violations of any of these standards.36
All of these measures are meant to curb abuse of asymmetric
information and to facilitate trust in the market. With regulation,
241 (Micheal J. Houston ed., 1988). A subtle issue is that while individual
preferences are idiosyncratic, they are still predictive. A consumer contemplating
the purchase of a microwave may be different in any number of ways from past
consumers, yet, learning that other consumers disliked the microwave would still
inform the consumer’s private decision.
27. Timothy J. Muris, Opportunistic Behavior and the Law of Contracts, 65
MINN. L. REV. 521, 522–26 (1981) (discussing the role of opportunistic behavior in
contract law). Muris seems to consider reputation as a better solution but
suggests that it is unhelpful when sellers can conceal their past misbehavior. Id.
at 526–27; see KATALIN JUDIT CSERES, COMPETITION LAW AND CONSUMER
PROTECTION 155–56 (2005) (explaining the need for consumer protection and the
various private law tools meant to achieve it); Henry Smith, Equity as SecondOrder Law: The Problem of Opportunism (Harvard Pub. L. Working Paper No.
15-13, 2015) (arguing that equity’s primary goal in private law is to control the
problem of opportunism).
28. George A. Akerlof, The Market for “Lemons”: Quality Uncertainty and the
Market Mechanism, 84 Q.J. ECON. 488, 488 (1970).
29. See, e.g., 7 C.F.R. § 996.13 (2018) (requiring that peanuts considered
“Segregation 1” shall not have more than 2.49 percent damaged kernels).
30. See, e.g., ALA. CODE § 8-8-1 (1975) (setting a cap of 8 percent on the price
of credit).
31. U.C.C. § 1-201(b)(20) (AM. LAW INST. & UNIF. LAW COMM’N, amended
2011); RESTATEMENT (SECOND) OF CONTRACTS § 205 (AM. LAW INST. 1981).
32. See, e.g., Moehling v. W. E. O’Neil Constr. Co., 170 N.E.2d 100, 107 (Ill.
1960) (finding a fiduciary duty of disclosure of material facts in a real estate
transaction between an agent and its principal). See generally Anthony T.
Kronman, Mistake, Disclosure, Information, and the Law of Contracts, 7 J. LEGAL
STUD. 1 (1978) (exploring optimal disclosure rules).
33. See, e.g., Choi & Spier, supra note 5, at 735 (explaining that courts are
“generally hostile toward[s]” attempts to opt-out of product liability).
34. See, e.g., MISS. CODE ANN. § 73-9-29 (1972) (requiring practicing dentists
to attend a board-certified educational program and pass an exam before
receiving a license to practice in the state).
35. See, e.g., U.S. FOOD & DRUG ADMIN., INGREDIENTS DECLARED AS
EVAPORATED CANE JUICE: GUIDANCE FOR INDUSTRY (2016) (providing that a
sweetener cannot be described as “evaporated cane juice” because it may mislead
consumers to believe that it is juice rather than sugar).
36. See, e.g., 815 ILL. COMP. STAT. ANN. 505/2, 7 (West 2016) (imposing
criminal and civil liability for consumer fraud).

[p. 10]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1248 WAKE FOREST LAW REVIEW [Vol. 54
consumers can trust—to an extent—that a car seat, for example, will
be sufficiently safe even if the producer is unknown or that an
unfamiliar merchant will not engage in unfair business practices.
Free-market advocates contest the need for such laws, which
necessarily limit freedom of contract, and instead argue that law is
“only one of many social institutions and practices amid which
markets function.”37 One such institution, arguably the most central
one, is reputation.38 Part of Adam Smith’s genius was the insight that
reputation facilitates trust in markets, even if actors are selfinterested; as Smith explained, the baker will soon realize that selling
low-quality bread will diminish his reputation and thus future profits,
making dishonesty an unprofitable business strategy.39 Drawing on
this insight, influential scholars such as Milton Friedman, Richard
Posner, and Richard Epstein have called to deregulate markets and
rely instead on internal market discipline.40
These debates are longstanding and tend to track traditional
political stances on the government, top-down regulation, and the free
market.41 Recently, however, a sea change has swept many
progressives and left-leaning thinkers towards the deregulatory
camp.42 The rise of the sharing economy and information technology
has inspired widespread belief in the impending death of asymmetric
information.43 In the past, private ordering through reputation was
understood to be the domain of small-knit communities, such as
Orthodox Jewish diamond traders or cattle ranchers in small
counties, as these small communities could effectively exchange
37. Lewis A. Kornhauser, Reliance, Reputation, and Breach of Contract, 26
J.L. & ECON. 691, 702 (1983); see also OLIVER E. WILLIAMSON, THE ECONOMIC
INSTITUTIONS OF CAPITALISM: FIRMS, MARKETS, RELATIONAL CONTRACTING 163–68
(1985).
38. See FRIEDRICH A. HAYEK, LAW, LEGISLATION AND LIBERTY, VOLUME 1:
RULES AND ORDER 46–48 (1973); Barak D. Richman, Firms, Courts, and
Reputation Mechanisms: Towards a Positive Theory of Private Ordering, 104
COLUM. L. REV. 2328, 2333–34 (2004) (dating the literature on private ordering
to the early 1990s).
39. See SMITH, supra note 1; see also ALBERT HIRSCHMAN, RIVAL VIEWS OF
MARKET SOCIETY 106–07 (1992).
40. See Lucian A. Bebchuk & Richard A. Posner, One-Sided Contracts in
Competitive Consumer Markets, 104 MICH. L. REV. 827, 827–28 (2006); Epstein,
supra note 5, at 131; Milton Friedman’s Free to Choose: Who Protects the
Consumers?, supra note 1.
41. Venerable traditions in political theory—Godwinian utopia and
Smithian natural liberty—believe reputation can effectively constrain selfinterested behavior. Klein, supra note 8, at 2–3. As economist Benjamin Klein
observed, “[i]f one puts small confidence in the efficacy and integrity of external
authority—in particular, governmental institutions—then the hope for selfpolicing gains in relevance.” Id. at 2.
42. See, e.g., Christian Britschgi, Progressives and Libertarians Team Up to
Deregulate Airports, REASON (July 19, 2019), https://reason.com/2019/07/19
/progressives-and-libertarians-team-up-to-deregulate-airports/printer/.
43. See Tabarrok & Cowen, supra note 3.

[p. 11]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1249
gossip and other word-of-mouth reputational information.44 The rise
of information technology now promises gossip at scale. Drug users,
involuntary regulatory entrepreneurs as they are, utilize message
boards to spread reputational information on which cocaine dealer
uses cheap fillers.45 Less daring consumers use popular reputation
platforms such as Amazon and Uber to guide their decisions on
everyday purchases.46 The meteoric rise of these platforms enthused
many about a future where regulatory interventions are moot, as
consumers do the work organically for their peers47 without coercive
44. See ROBERT C. ELLICKSON, ORDER WITHOUT LAW: HOW NEIGHBORS SETTLE
DISPUTES vii (1991); Bernstein, Opting Out, supra note 2, at115, 130, 139–40
(analyzing the behavior of Jewish Orthodox diamond traders in New York); see
also Stewart Macaulay, Non-Contractual Relations in Business: A Preliminary
Study, 28 AM. SOC. REV. 55, 55, 63 (1963) (studying the behavior of managers in
Wisconsin).
45. Drug dealers trade under carefully maintained brand names, although
they suffer from a short half-life. See Nick Janetos & Jan Tilly, Reputation
Dynamics in a Market for Illicit Drugs 1 (U. Pa., Working Paper No.
arXiv:1703.01937v1, 2017), https://arxiv.org/pdf/1703.01937.pdf.
46. See Cheung et al., supra note 18, at 619; Smith, supra note 17.
47. A sample of contemporary examples of pro-reputation attitudes includes
The Disrupter Series: How the Sharing Economy Creates Jobs, Benefits
Consumers, and Raises Policy Questions: Hearing Before the Subcomm. on
Commerce, Mfg., & Trade of the H. Comm. on Energy & Commerce, 114th Cong.
2 (2016) (statement of Hon. Michael C. Burgess, Chairman, Subcomm. on
Commerce, Mfg., & Trade) (“Sharing platforms are inherently good, providing
reputation feedback loops.”); Barriers to Opportunity: Do Occupational Licensing
Laws Unfairly Limit Entrepreneurship and Jobs: Hearing Before the Subcomm.
on Contracting and Workforce of the H. Comm. on Small Business, 113th Cong.
11 (2014) (statement of Hon. Richard Hanna, Chairman, Subcomm. on
Contracting and Workforce) (“Bad actors get in, people find out about their
reputations, good or bad, they grow or leave the market.”); FTC, THE “SHARING”
ECONOMY: ISSUES FACING PLATFORMS, PARTICIPANTS & REGULATORS 32 (2016),
https://www.ftc.gov/system/files/documents/reports/sharing-economy-issuesfacing-platforms-participants-regulators-federal-trade-commission-staff
/p151200_ftc_staff_report_on_the_sharing_economy.pdf (“[A] seller’s favorable
reputation can provide important leverage for regulators seeking to ensure
consumers are protected when shopping online.”); Jürgen Backhaus, Company
Board Representation, in THE ELGAR COMPANION TO LAW AND ECONOMICS 155, 155
(Jurgen Backhaus ed., 1999); RICHARD A. POSNER, THE ECONOMICS OF JUSTICE 288
(1981); Strahilevitz, supra note 1, at 71; Omri Ben-Shahar, Consumer Protection
Without Law, REGULATION 26, 26 (Summer 2010); David Charny, Illusions of a
Spontaneous Order: “Norms” in Contractual Relationships, 144 U. PA. L. REV.
1841, 1841–42 (1996); Alex Geisinger, Are Norms Efficient? Pluralistic Ignorance,
Heuristics, and the Use of Norms as Private Regulation, 57 ALA. L. REV. 1, 1–2,
29–30 (2005); Robert E. Scott, A Theory of Self-Enforcing Indefinite Agreements,
103 COLUM. L. REV. 1641, 1644, 1692 (2003). In economics, see, e.g., Benjamin
Klein & Keith B. Leffler, The Role of Market Forces in Assuring Contractual
Performance, 89 J. POL. ECON. 615, 616 (1981). On complementarity between law
and reputation, see Kishanthi Parella, Reputational Regulation, 67 DUKE L.J.
907, 910–18 (2018); Shapira, supra note 25, at 1203. But see Eric A. Posner,
Recent Books On International Law, 101 AM. J. INTL. L. 509, 510 (2007) (reviewing
ROBERT E. SCOTT & PAUL B. STEPH, THE LIMITS OF LEVIATHAN: CONTRACT THEORY
AND THE ENFORCEMENT OF INTERNATIONAL LAW (2006)).

[p. 12]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1250 WAKE FOREST LAW REVIEW [Vol. 54
and often misguided government intervention.48 Given this new
zeitgeist, there is little in the way of effective opposition to the recent
policy trends that have dramatically scaled down consumer
protection.49
To give a sense of where the contemporary battle lines are
drawn, consider the influential debate between Professors Richard
Epstein and Oren Bar-Gill (who also serves as a reporter for the new
Restatement of Consumer Contracts).50 Both scholars debate the old
question of asking how severely contract law should limit the freedom
of contract in the name of other interests.51 Epstein concedes the
existence and importance of cognitive constraints on consumer
decision-making, which he agrees can lead to “serious mistakes.”52
Still, in his view, “second-order rationality” in the form of reputation,
among other sources, can overcome these shortcomings.53
Remarkably, Epstein sees reputation as a valid response to the new
problems posed to the traditional model by behavioral economics.54
In contrast, Bar-Gill argues that because goods are sometimes
unique, or consumers use them in unique ways, there is too little
information that is transferable among consumers.55 Under such
conditions, Bar-Gill contends, there is still a need for regulatory
interventions in private contracts, such as safety standards or
immutable warranties.56
The most surprising feature of this debate is not the
disagreement but the broad agreement that underlies it. For
standardized goods (or standard uses), reputation, in addition to
48. See Rory Van Loo, The Corporation as Courthouse, 33 YALE J. REG. 547,
569 (2016) (noting the trend where “[t]he consumer legal system is evolving
toward a similar reliance on reputation-based governance mechanisms”). On the
problem of authority, see MICHAEL HUEMER, THE PROBLEM OF POLITICAL
AUTHORITY 100 (2013) (arguing that courts have a limited role and “may not go
on to coercively impose paternalistic or moralistic laws”); see also Charny, supra
note 47, at 1845–47 (highlighting the centralization inherent to informal systems
of ordering); Duncan Kennedy, The Role of Law in Economic Thought: Essays on
the Fetishism of Commodities, 34 AM. U. L. REV. 939, 944–49 (1985) (critiquing
the tendency to see the market as natural and state interventions as artifice).
49. See sources cited supra note 5.
50. See Bar-Gill, supra note 5, at 749–54; Richard A. Epstein, The
Neoclassical Economics of Consumer Contracts, 52 MINN. L. REV. 803, 803–10
(2003).
51. See Bar-Gill, supra note 5, at 749–54; Epstein, supra note 50, at 808–10.
52. Epstein, supra note 5, at 111.
53. Id.
54. Epstein, supra note 50, at 811.
55. Bar-Gill, supra note 5, at 756 (“[Epstein] forcefully argues that mistakes
with respect to the value of a standardized product are unlikely to persist in the
marketplace. But not all products are standardized . . . . With a nonstandardized good, the information obtained by one consumer might not be
relevant to another consumer who purchased a different version of the
nonstandard good.”).
56. Id. at 793–94.

[p. 13]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1251
other background pressures, can be sufficiently potent to curb
opportunistic behavior. Hence, contract and consumer law need not
worry about intervening in areas where product reputation is
abundant.57 Similar attitudes were expressed by other leading
figures in the field and in the context of torts, occupational licensing,
and even drug regulation.58 As shall become clear, the existence of
reputation failures undermines this view.
B. The Supposed Reliability of Reputation in Legal Thought
These debates highlight the centrality of the belief that
reputation is an effective, dependable, credible, and reliable
regulator.59 Advocates argue that the loss of reputation is immediate,
independent of lengthy and uncertain trials, and stems from the
interactions of the parties themselves.60 Additionally, reputation
allows parties to tap into assets that the legal system cannot reach.61
For these reasons, advocates believe that what the law does slowly
and inaccurately, reputation can do quickly and precisely.62
57. See Schwartz & Scott, supra note 5, at 557; see also Robert A. Hillman &
Jeffrey J. Rachlinski, Standard-Form Contracting in the Electronic Age, 77
N.Y.U. L. REV. 429, 441 (2002) (arguing that reputational pressures restrain
firms from enforcing some exploitative terms in standard-forms contracts).
58. See supra sources cited note 5.
59. See Klein, supra note 8, at 2 (noting that reputation enforcement does
not involve coercion); Barak D. Richman, Norms and Law: Putting the Horse
Before the Cart, 62 DUKE L.J. 739, 740 (2012) (“Among the most salient features
of modern courts are that they are expensive, slow, and inaccurate.”); see also
Scott Baker & Albert H. Choi, Reputation and Litigation: Why Costly Legal
Sanctions Can Work Better Than Reputational Sanctions, (Va. Law & Econ.,
Research Paper No. 2013-02, 2013).
60. See Richman, supra note 38, at 2335.
61. Id. at 2332 (explaining that efficient enforcement is more important than
efficient administration in explaining why merchant communities prefer private
ordering to contractual enforcement).
62. See, e.g., John Barton, The Economic Basis of Damages for Breach of
Contract, 1 J. LEGAL STUD. 277, 277 (1972); Lisa Bernstein, Private Commercial
Law in the Cotton Industry: Creating Cooperation Through Rules, Norms, and
Institutions, 99 MICH. L. REV. 1724, 1725 (2001) (arguing that the informal order
at the cotton industry “work[s] extraordinarily well,” that this system resolves
disputes “expeditiously and inexpensively,” and that arbitration awards “are
widely respected and complied with promptly”); Juan Jose Ganuza, Fernando
Gomez & Marta Robles, Product Liability versus Reputation, 32 J.L. ECON. &
ORG. 213, 213 (2016); Lawrence Lessig, The New Chicago School, 27 J. LEGAL
STUD. 661, 665 (1998). For earlier examples of reputation advocates, see, e.g.,
Avner Greif, Informal Contract Enforcement: Lessons from Medieval Trade, in 2
THE NEW PALGRAVE DICTIONARY OF ECONOMICS AND THE LAW 287, 287–88 (Peter
Newman ed., 1998); Janet T. Landa, A Theory of the Ethnically Homogeneous
Middleman Group: An Institutional Alternative to Contract Law, 10 J. LEGAL
STUD. 349, 349 (1981).

[p. 14]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1252 WAKE FOREST LAW REVIEW [Vol. 54
As central as reputation is to these debates, it is perplexing to see
how little attention is given to reputation’s nature.63 Instead, the
literature mostly relies on a simplistic “emergentist” view of
reputation that ignores all questions of how reputation comes to be.64
The reputation of goods and services is described as simply emerging
from complex market interactions.65 A seller sells a widget and once
enough consumers purchase and use it, the widget “automatically”
gains a reputation for quality.66 Somehow, reputation emerges. From
where? How? By whom? These issues are hardly ever addressed.
Instead, reputation is taken to be, as Shapira pointedly notes, “[a]
frictionless, uncomplicated process in which individuals somehow get
access to information.”67
It is hard to overstate just how common it is for people to perceive
that reputation is reliable and how much this perception influences
policy. Take, for example, economic models of market transactions;
there, it is common to assume that “[t]he moment that a person
cheats, it becomes common knowledge that the person lacks integrity,
and hence there is no cooperation for the rest of the game.”68 In the
foundational Klein-Leffler model of reputation, consumers are
explicitly assumed to “costlessly communicate [quality information]
among one another.”69 Such spontaneous reputational information is
then thought to propagate “throughout the community without
63. LAWRENCE MCNAMARA, REPUTATION AND DEFAMATION 19 (2007) (“[O]nly
a few works are concerned with the nature of reputation.”); Laura A. Heymann,
The Law of Reputation and the Interest of the Audience, 52 B.C. L. REV. 1341,
1345 (2011); Robert C. Post, The Social Foundations of Defamation Law:
Reputation and the Constitution, 74 CALIF. L. REV. 691, 692 (1986) (describing
reputation as a “mysterious thing”).
64. A view is emergentist if it identifies a phenomenon only at the complex
level. For example, the quality of “saltiness” does not describe the taste of
chlorine or sodium, yet their combination creates a salty ionic compound; a grain
of sand has no “pileness” to it, but once enough grains are collected, a pile
emerges; or, more contentiously, no neuron has self-awareness, yet their
collection seems to cause conscience. This omission is hardly unique to law; for a
recent example in other fields, see Wenqi Shen et al., Competing for Attention:
An Empirical Study of Online Reviewers’ Strategic Behavior, 39 MGMT. INFO. SYS.
Q. 683, 684 (2015) (“[M]ost of the existing literature has overlooked the question
of how online reviewers are incentivized to write reviews.”).
65. For a review of legal conceptions of reputation, see Post, supra note 63,
at 691.
66. See Richman, supra note 59, at 750.
67. Shapira, supra note 25, at 1203.
68. W. Bentley MacLeod, Reputations, Relationships and the Enforcement of
Incomplete Contracts 31 (Ctr. for Econ. Studies & Ifo Inst. for Econ. Research,
Working Paper No. 1730, 2006); see also Lewis A. Kornhauser, Reliance,
Reputation, and Breach of Contract, 26 J.L. & ECON. 691, 697 (1983) (depicting
an ideal model of reputation where “buyers have perfect knowledge of the seller’s
performance rate”).
69. Klein & Leffler, supra note 47, at 617. They do admit the possibility of
imperfect recall of reputational information. Id.

[p. 15]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1253
institutional help.”70 In legal scholarship, this conception arises most
clearly in defamation law jurisprudence where reputation is thought
of as a right—natural, static, or inherent, like property or dignity—
that comes into being by immaculate conception and must be
“protected” against those who seek to besmirch it.71
Some scholars have started to recognize cracks in the traditional
paradigm.72 As they note, reputational information may be costly to
obtain, noisy,73 distorted by the incentives of intermediaries,74 or
ineffectual (if, for example, a sellers’ presence in the market is shortlived).75 Economists Paul Milgrom and John Roberts focus on the
difficulty of creating reputation when opportunistic behavior is hard
to detect.76 Similarly, Professor Alan Schwartz notes the potential
costs of reputation, explaining that “the innocent party [to a failed
transaction] will incur costs in informing others that it was not at
fault, and third parties will incur costs learning about which of the
contract parties is unreliable.”77 Finally, Professor Bar-Gill
emphasizes the possibility that in some markets there will be an
insufficient volume of reputational information.78 Still, even
admitting reputation’s potential noise does not amount to a claim that
recognizes the inherent systematic distortion of reputational
information.79 Rather, the noise is thought to disappear, as new
information accumulates and “increases the diagnosticity” and
persuasiveness of reputation.80 Moreover, these are exceptions. The
dominant view in the policy and scholarship is still very much
emergentist, expressing great trust in the reliability of reputation.
The emergentist view is problematic for several reasons, not the
least of which is its lack of any theoretical underpinnings that explain
when—and when not—reputation will come to be. Why is
70. See Richman, supra note 59, at 750.
71. See Post, supra note 63, at 692; Yonathan Arbel & Murat Mungan, The
Uneasy Case for Expanding Defamation Law, 71 ALA. L. REV. (forthcoming 2019)
(arguing that audiences are actively involved in assigning meaning to statements
and that defamation law might exacerbate the harmful effect of lies).
72. Some notable examples include Goldman, supra note 9; Stemler, supra
note 4; Van Loo, supra note 48, at 583 (noting the existence of potential
informational market failures due to manipulations of consumer reviews).
73. See Alan Schwartz, The Enforcement of Contracts and the Role of the
State, in LEGAL ORDERINGS AND ECONOMIC INSTITUTIONS 105, 105 (Fabrizio
Cafaggi et al. eds., 2007) (“Reputation is a noisy signal.”).
74. See Shapira, supra note 25, at 1219.
75. See Douglas W. Diamond, Reputation Acquisition in Debt Markets, 97 J.
POL. ECON. 828, 829 (1989).
76. PAUL MILGROM & JOHN ROBERTS, ECONOMICS, ORGANIZATION, AND
MANAGEMENT 265 (1992).
77. See Schwartz, supra note 73, at 105.
78. See Bar-Gill, supra note 5, at 756.
79. See, e.g., MILGROM & ROBERTS, supra note 76, at 259–67.
80. Adwait Khare et al., The Assimilative and Contrastive Effects of Wordof-Mouth Volume: An Experimental Examination of Online Consumer Ratings, 87
J. RETAILING 111, 112 (2011).

[p. 16]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1254 WAKE FOREST LAW REVIEW [Vol. 54
information about some products abundant whereas information for
other products is sparse? Another problem is that many jurists have
come to think of reputation as a right that belongs to individuals,81
rather than the byproducts of dynamic social processes.82 As a result,
much of the discussion about reputation neglects its social value.83
But most disconcerting is the implication that reputation is generally
reliable—that it fairly describes the quality of the underlying good
without systematic bias. After all, if reputation simply emerges,
there is no process by which it will be “tainted.” Thus, the proposition
that reputation can be unreliable cuts at heart, nerve, and sinew of
these influential works.
III. REPUTATION FAILURE: MICROFOUNDATIONS, DISTORTIONS, AND
SOCIAL WELFARE
When my neighbor complained that his newly purchased lawn
mower was shoddy, he created reputational information.84 When user
“daniel” wrote on Amazon that a play tent has “the stability of a house
of cards,” he or she created reputational information.85 When
musician Dave Caroll uploaded his song “United Breaks Guitars” to
YouTube, he created reputational information, and powerful
information at that: United’s stock price fell ten cents, representing a
market cap loss of $180 million.86
Reputation is information. It is a kind of statistical information
which helps consumers predict their own experiences based on the
81. See, e.g., Joseph Blocher, Reputation as Property in Virtual Economies,
118 YALE L.J. POCKET PART 120, 120 (2009), https://www.yalelawjournal.org
/forum/reputation-as-property-in-virtual-economies (“[R]eputation is not merely
valuable; it is the new New Property.”). Moreover, many disagreed with the
Supreme Court’s decision in Paul v. Davis, 424 U.S. 693, 694 (1976) (holding that
harm to reputation, by itself, is not deprivation of liberty or property for
Fourteenth Amendment purposes). See Eric J. Mitnick, Procedural Due Process
and Reputation Harm: Liberty as Self-Invention, 43 U.C. DAVIS L. REV. 79, 89–97
(2009); see also Marrero v. City of Hialeah, 625 F.2d 499, 514 (5th Cir. 1980)
(concerning a Florida law that considers business reputation a property interest).
82. See POSNER, supra note 47, at 252–53 (1981) (“It makes no sense to treat
reputation as a ‘right.’ Reputation is what others think of us.”).
83. See Heymann, supra note 63, at 1342.
84. See Nick Emler, Gossip, Reputation, and Social Adaptation, in GOOD
GOSSIP 135 (R. F. Goodman & A. Ben-Ze’ev eds., 1994) (“Reputations do not exist
except in the conversations that people have about one another.”).
85. daniel, Customer Review of “AMASKY tm Large Space Children Game
Play Tent,” AMAZON (Apr. 20, 2016), https://www.amazon.com/gp/customerreviews/R373NHZZ854AU4/ref=cm_cr_arp_d_rvw_ttl?ie=UTF8&ASIN=B00YB
TFY52.
86. See Chris Ayres, Revenge is Best Served Cold – on YouTube, TIMES
(London) (July 22, 2009, 1:00 AM), https://www.thetimes.co.uk/article/revengeis-best-served-cold-on-youtube-2dhbsh6jtp5; Gulliver, Did Dave Carroll Lose
United Airlines $180m?, ECONOMIST (July 24, 2009), https://www.economist.com
/gulliver/2009/07/24/did-dave-carroll-lose-united-airlines-180m.

[p. 17]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1255
distribution and valence of experiences of past consumers.87 As such,
reputational information is like a poll. But as the examples highlight,
this kind of information does not simply emerge; rather, it is the fruit
of deliberate action by disparate individuals who decide to take time
and effort to share reviews, opinions, gossip, and other word-of-mouth
information.
The most basic observation to make is that such peer-to-peer
reputational information is a public good.88 While everyone benefits
from having this public resource, producers of reputational
information are not directly compensated for their contributions.
Private costs and public benefits are a recipe for the well-known
free-rider problem—like national defense, clean air, and a vaccinated
society—where there is a constant concern with overconsumption and
undersupply.89
Once reputation is seen as public good, a deep puzzle is exposed:
What motivates individuals to share reputational information? Who
does so? And, critically, to what effect?90
This Part explores these “microfoundations” of reputation and
their consequences. It shows how people share experiences for
reasons that are mostly private and self-serving.91 As a result, future
consumers are often exposed to a highly unrepresentative and biased
sample of limited credibility. Inferences drawn from such samples
can be highly misleading, even for those consumers who are aware of
them and try to account for them. As data scientists would say: Bias
in, bias out.92
87. See Goldman, supra note 9, at 294; Shapira, supra note 25, at 1201.
88. See Cowen, supra note 7. Once reputational information exists, it is hard
to prevent people from using it (i.e., it is non-excludability); nor does use of this
resource diminish it (i.e., it is non-rivalry). See Larry Downes, The Economics of
Information: From Dismal Science to Strange Tales, in THE NEXT DIGITAL DECADE
273, 277–78 (Berin Szoka & Adam Marcus eds., 2010). Professional publications
solve these problems by commoditizing the information they produce, which is
subject to copyright and other protections. See, e.g., N.Y. GEN. BUS. LAW § 397
(McKinney 1961) (prohibiting the unconsented use of nonprofits’ test results by
nonprofits).
89. Today, reputational platforms reap most of the benefit of reputation
aggregation, but reputation’s direct producers receive very little reward. These
issues are sometimes conceptualized as a tragedy of the commons. See Garrett
Hardin, The Tragedy of the Commons, 162 SCI. MAG. 1243, 1243–48 (1968).
90. By and large, legal scholars have glossed over this question. One notable
exception is Robert D. Cooter, Decentralized Law for A Complex Economy: The
Structural Approach to Adjudicating the New Law Merchant, 144 U. PA. L. REV.
1643, 1669 (1996) (arguing that individuals disseminate reputational
information due to an internalized social norm).
91. For an exploration of consumer activism in the marketplace, see
Yonathan A. Arbel & Roy Shapira, Consumer Activism: From the Informed
Minority to the Crusading Minority, DEPAUL L. REV. (forthcoming 2019) and
Arbel & Shapira, supra note 2.
92. Sandra G. Mayson, Bias In, Bias Out, 128 YALE L.J. 2218, 2224 (2019).

[p. 18]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1256 WAKE FOREST LAW REVIEW [Vol. 54
A. The Microfoundations of Reputation
1. The Costs of Gossip
Creating and sharing reputational information involves effort,
time, and, in some cases, the risk of legal liability. “Bianca S.” must
have spent the better part of her lunch break writing a 143-word
review of a cleaning service attached alongside six photos of her
home.93 One anonymous Amazon user probably spent at least a few
minutes writing a 291-word review of his experiences with Kevlar
gloves,94 and “The Amazing Lucas” in all likelihood spent a few hours
creating and editing a seven-minute review of the movie It.95 Not only
is creating reviews time consuming, it is also sometimes emotionally
difficult to say negative things about others.96 Beyond these costs, as
will be elaborated below, there is a growing tendency among some
service providers to sue consumers for negative reviews, using factual
inaccuracies and misstatements to ground their lawsuits.97 Such
lawsuits can involve months of litigation, a serious disruption, and—
in some rare cases—large judgments.98
These broadly-defined costs suggest that there must be
countervailing motivations to produce reputational information, or
else people—as distinct from sellers, advertisers, and affiliates—will
not generate reputational information. Citing the utility of
reputation to the operation and efficiency of the market, as well as
the welfare of future consumers, is insufficient as these are public
benefits.99 What needs to be determined, then, are the specific private
benefits—what the sharing individual gets from incurring these
costs.
2. Internal Drives: On Spite and Gratitude
Starting in the 1960s, psychologists began investigating the
psychological drives impelling individuals to participate in
93. Bianca S., Customer Review of “Joanna Cleaning Service,” YELP (July 15,
2016), https://www.yelp.com/biz/joanna-cleaning-service-brooklyn2?hrid=XjR0XP21gLXqvOoHH99vEA.
94. PsychSchematics, Customer Review of “SINNAYEO-Kevlar Cat Bird
Dog Reptile Barbecue, Grill, hearth Leather Gloves Animal Handling Gloves,”
AMAZON (Mar. 4, 2017), https://www.amazon.com/gp/review
/R2ZUV15SLM6H0M.
95. The Amazing Lucas, it movie 2017 review WHY THIS MOVIE IS BAD,
YOUTUBE (Sept. 12, 2017), https://www.youtube.com/watch?v=PEKLB9j6-nY.
96. Many religions prohibit calumny and detraction. See, e.g., Joseph
Delany, Detraction, in 4 THE CATHOLIC ENCYCLOPEDIA 757, 757–58 (Charles G.
Herbermann et al. eds., 1908); YISRAEL MEIR KAGAN, SEFER CHAFETZ CHAYIM
(Yedidya Levy trans., 2008).
97. See infra Subpart IV.B.5.
98. See Steven Tadelis, Reputation and Feedback Systems in Online
Platform Markets, 8 ANN. REV. ECON. 321 (2016).
99. See Jeffrey L. Harrison, A Positive Externalities Approach to Copyright
Law, 13 J. INTELL. PROP. L. 1, 8–10 (2012).

[p. 19]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1257
word-of-mouth activity.100 In an influential paper, psychologist
Ernest Dichter highlighted four internal motivations: self-centered
perceptions, quality of experience, altruistic motivations, and
message involvement.101 In the presence of these factors, individuals
would be motivated to create and share reputational information.102
Over time, however, this theory encountered difficulties. Most
problematic were the empirical findings that consumers are more
likely to share reputational information when they had a favorable
experience.103 Yet other evidence showed the exact opposite:
dissatisfaction yields greater propensity to share.104 In 1998,
business professor Eugene Anderson reconciled these findings by
explaining that the underlying issue is the extremity of experience
rather than its valence.105 Further, current work in psychology
explains the creation of favorable and negative reviews as a distinct
activity motivated by different internal impetus. For example, D.S.
Sundaram et al. argued that positive word-of-mouth results from
altruism,106 product involvement, self-enhancement, and a desire to
help the company, whereas negative word-of-mouth is due to anxiety
reduction, vengeance, altruism, and advice-seeking purposes.107
Critically, these internal drivers are related to the quality of the
experience. The standard Expectation Disconfirmation Theory holds
that the gap between expectation and reality creates a sense of
100. The propensity to help others, as well as the idea that one must
“retaliate” against wrongs and “reward” generosity, has strong roots in
evolutionary psychology and social norms. See ELINOR OSTROM, TRUST AND
RECIPROCITY: INTERDISCIPLINARY LESSONS FOR EXPERIMENTAL RESEARCH 41–44
(2003); Yonathan A. Arbel & Yotam Kaplan, Tort Reform Through the Back Door:
A Critique of Law and Apologies, 90 S. CAL. L. REV. 1199, 1212 n.60 (2017); see
also Jeffrey L. Harrison, Spite: Legal and Social Implications, 22 LEWIS & CLARK
L. REV. 991, 993 (2018) (“[P]erhaps actions that appear spiteful are actually not
self-regarding but have deontological significance in that the detractor acts out
of sense of duty.”).
101. Ernest Dichter, How Word-of-Mouth Advertising Works, 44 HARV. BUS.
REV. 147, 148 (1966); see also Kyung Hyan Yoo & Ulrike Gretzel, What Motivates
Consumers to Write Online Travel Reviews?, 10 INFO. TECH. & TOURISM 283, 286
(2008) (discussing motivations).
102. See Yoo & Gretzel, supra note 101, at 286–88.
103. See John H. Holmes & John D. Lett, Jr., Product Sampling and Word of
Mouth, 17 J. ADVERT. RES. 35, 36 (1977).
104. See Marsha L. Richins, Negative Word-of-Mouth by Dissatisfied
Consumers: A Pilot Study, 47 J. MARKETING 68, 76 (1983).
105. See Eugene W. Anderson, Customer Satisfaction and Word of Mouth, 1
J. SERV. RES. 5, 11 (1998).
106. One oddity with altruistic motivations is that consumers tend to review
products that were already extensively reviewed despite the low information
value of such reviews. See Jonathan Lafky, Why Do People Rate? Theory and
Evidence on Online Ratings, 87 GAMES & ECON. BEHAV. 554, 567 (2014); Fang Wu
& Bernardo A. Huberman, Opinion Formation Under Costly Expression, 1 ACM
TRANSACTIONS ON INTELLIGENT SYS. & TECH. 1, 3 (2010).
107. See D.S. Sundaram et al., Word-Of-Mouth Communications: A
Motivational Analysis, 25 ADVANCES IN CONSUMER RES. 527, 527–28 (1998).

[p. 20]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1258 WAKE FOREST LAW REVIEW [Vol. 54
disequilibrium in the consumer that manifests as feelings of spite or
gratitude that motivate action.108 A recently developed, empirically
successful variant of this theory is the tractable “brag and moan”
model. This model stipulates that independent of expectations,
extreme experiences would motivate individuals to share their
opinions with others.109 The problem is, however, that many if not
most products and services are not extreme in the experiences they
generate.110 The theory holds that these tepid experiences will tend
to be suppressed because they are “boring” and do not evoke any sense
of spite or gratitude.111 Empirical research strongly supports this
prediction.112
3. Social Pressures and Herd Behavior
Aristotle was not wrong: man is a social animal. As such, we are
in some sense “programmed” to cooperate, reciprocate, and engage in
social behavior.113 It is no coincidence that reputation-creating
activities stand at the center of so many social activities.114 In social
gatherings, individuals gossip, share experiences, and impart
opinions with other members of the social community.115 Such
activities have a clear social function. When transgressors violate
community norms, gossip and related activities allow members of the
community to learn of the violation and take concerted social action
against the transgressor, such as avoidance, disrespect, and in
extreme cases, shunning and excommunication.116 Fearing this,
108. See William O. Bearden & Jesse E. Teel, Selected Determinants of
Consumer Satisfaction and Complaint Reports, 20 J. MARKETING RES. 21, 21–27
(1983).
109. Nan Hu et al., Can Online Reviews Reveal a Product’s True Quality?
Empirical Findings and Analytical Modeling of Online Word-of-Mouth
Communication, EC ‘06 PROC. 7TH ACM CONF. ELECTRONIC COM. 324, 327 (2006)
(attempting to discern whether consumer spite (or gratitude) is premised on a
desire to punish the seller or to warn future buyers failed to reach any clear
conclusions); Lafky, supra note 106, at 563.
110. Lafky, supra note 106, at 556–57.
111. Id.
112. See infra Subpart II.A.
113. See, e.g., OSTROM, supra note 100, at 28 (summarizing experimental
studies showing human tendency to reciprocate at the expense of self-interest).
114. See, e.g., Nicholas Emler, Gossip, Reputation, and Social Adaptation, in
GOOD GOSSIP 117, 117 (Robert. F. Goodman & A. Ben-Ze’ev eds., 1994) (exploring
the role of gossip).
115. See Jonah Berger, Word of Mouth and Interpersonal Communication: A
Review and Directions for Future Research, 24 J. CONSUMER PSYCHOL. 586, 588–
90 (2014) (arguing that word-of-mouth activities are means to a variety of social
ends, such as self-enhancing one’s image, signaling a positive identity, and filling
conversational space).
116. See, e.g., ROBERT C. ELLICKSON, ORDER WITHOUT LAW: HOW NEIGHBORS
SETTLE DISPUTES 130 (1991).

[p. 21]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1259
community members feel a strong pressure to conform, thus
maintaining social norms.117
The standard view of social pressures fails to consider the many
ways that social forces can also undermine cooperation.118 If a seller
misbehaves in a market, it is important that buyers share this
information. But proper socialization often consists of masking one’s
true feelings, forgiving slights, and avoiding offending others even at
the cost of distorting reality (i.e., “white lies”). Forgiveness and
charity can play a negative role by leading consumers to avoid
pursuing action.119 Even reciprocity can be problematic, as
discovered inadvertently by eBay’s engineers.120 In one iteration,
buyers and sellers could rate each other after every transaction.121
This led reviewers to post (artificially) favorable reviews in the hope
that their counterparts will positively review them.122 Consumers
explained their behavior as being motivated by fear of retaliation: “[I]f
I left a bad review, I might be afraid of being retaliated against.”123 A
similar issue arises with Uber and Lyft, where both drivers and
passengers rate each other.124 In addition, individuals often mask
their opinions due to the Social Desirability Bias,125 the pressure to
strategically project socially acceptable opinions.126 A hijab-wearing
interviewer would hear more women reporting themselves as
religious than an unveiled one.127 Similarly, this bias often leads
people to misreport tax compliance, porn consumption, homosexual
117. Id. at 57.
118. This conversion—of how social tendencies undermine cooperation—is
similar to Adam Smith’s conversion—the idea that selfish behavior can promote
cooperation. SMITH, supra note 1, at 326–27.
119. See Arbel & Shapira, supra note 2 (discussing findings showing that most
consumers avoid reacting to seller failure).
120. See Chrysanthos Dellarocas et al., Self-Interest, Reciprocity, and
Participation in Online Reputation Systems 3 (MIT Sloan, Working Paper No.
4500-04, 2004).
121. Id.
122. Id. at 1.
123. See FTC, supra note 47, at 42; Edelman & Geradin, supra note 4, at 316
(“Some users seem to fear retaliation through a review platform.”); see also
Bryant Cannon & Hanna Chung, A Framework for Designing Co-Regulation
Models Well-Adapted to Technology-Facilitated Sharing Economies, 31 SANTA
CLARA HIGH TECH. L.J. 23, 38 (2015).
124. See, e.g., SAUL KASSIN ET AL., SOCIAL PSYCHOLOGY 285 (10th ed. 2017)
(discussing the role of reciprocity with respect to reviews); Stemler, supra note 4,
at 692 (discussing the effects of reciprocity in the sharing economy).
125. See Maryon F. King & Gordon C. Bruner, Social Desirability Bias: A
Neglected Aspect of Validity Testing, 17 PSYCHOL. & MARKETING 79, 82 (2000)
(“Today, [Social Desirability Bias] is considered to be one of the most common and
pervasive sources of bias affecting the validity of experimental and survey research findings in psychology.”).
126. See Lisa Blaydes & Rachel M. Gillum, Religiosity-of-Interviewer Effects:
Assessing the Impact of Veiled Enumerators on Survey Response in Egypt, 6 POL.
& RELIGION 459, 462 (2013).
127. Id. at 476.

[p. 22]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1260 WAKE FOREST LAW REVIEW [Vol. 54
activities, recycling, and charity.128 It is hard to overstate the
tendency of individuals to misstate opinions given this bias.129
Social pressures often result in individuals herding around
popular opinions.130 Herding is the convergence of opinions, a
phenomenon very familiar in public debates, whereby participants
“often shift their public statements in accordance with reputational
incentives.”131 Herding is also well documented in the context of
reviews.132 Experiments by University of Washington Professor Ann
Schlosser found that exposing subjects to past negative reviews
increases the likelihood that the subject will also voice a negative
review.133 Similarly, reviews for popular movies tend to lump around
leading opinions.134 Interestingly, some individuals exhibit an
antiherding behavior whereby they strategically express
nonconforming opinions, possibly in an attempt to lead the herd.135
While it is hard to assess the overall effect of social motivations in the
abstract, the effect seems large. As marketing professors Wendy Moe
and David Schwiedel concluded: “[A] vocal subset of the customer
base may dominate the ratings environment, consequently steering
the subsequently posted evaluations and deterring some customers
from contributing to the environments.”136
128. See Samuel Himmelfarb & Carl Lickteig, Social Desirability and the
Randomized Response Technique, 43 J. PERSONALITY & SOC. PSYCHOL. 710, 710–
17 (1982) (reviewing empirical findings).
129. See King & Bruner, supra note 125, at 82.
130. See Timur Kuran & Cass R. Sunstein, Availability Cascades and Risk
Regulation, 51 STAN. L. REV. 683, 727–30 (1999); Maria del Mar Rueda et al., Use
of Randomized Response Techniques When Data Are Obtained from Two Frames,
9 APPLIED MATHEMATICS & INFO. SCI. 389, 389 (2015). Social motivations are
complex and their effects can go in many different directions, including
antiherding, as in Radu Jurca et al., Reporting Incentives and Biases in Online
Review Forums, 4 ACM TRANSACTIONS ON WEB 1, 1–3, 14, 20, 21, 22, 25 (2010).
131. Cass R. Sunstein, Deliberative Trouble? Why Groups Go to Extremes, 110
YALE L.J. 71, 78 (2000).
132. See Stemler, supra note 4, at 693–94 (discussing evidence of herding in
online reviews).
133. See Ann E. Schlosser, Posting Versus Lurking: Communicating in a
Multiple Audience Context, 32 J. CONSUMER RES. 260, 264 (2005) (“[R]eading a
negative review triggers posters’ concerns with the social outcomes of their public
evaluations, thereby causing them to lower their public ratings strategically.”).
134. See Young-Jin Lee et al., Do I Follow My Friends or the Crowd?
Information Cascades in Online Movie Ratings, 61 MGMT. SCI. 2241, 2256 (2015).
135. Wendy W. Moe & David A. Schweidel, Online Product Opinions:
Incidence, Evaluation, and Evolution, 31 MARKETING SCI. 372, 383 (2012); Shen
et al., supra note 64, at 689–90 (finding that raters choose to review less reviewed
books in order to stand out and gain more attention where there are reviewer
rankings systems).
136. Moe & Schweidel, supra note 135, at 385. Similarly, others find that
attention seeking is another important social motivator (where there are
reviewer ranking systems). See Shen et al., supra note 64, at 685. Additionally,
maintaining an online social identity (rather than anonymity) was found to lead
to more quality content. Zhongmin Wang, Anonymity, Social Image, and the

[p. 23]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1261
4. Material Incentives: Shilling and Cherry-Picking
Material rewards are the most direct form of incentivizing
individuals to share reputational information. The familiar version
of that is “shilling,” also known as “fake reviews” or “astroturfing,”
which involves the provision of payments in exchange for (unfounded)
positive reviews.137 Shilling is reported to be quite common, with
some estimating that as much as 30 percent of online reviews are
fake.138 In 2013, for example, Samsung was fined $340,000 because
it paid for fake reviews—both positive reviews for their own products
and negative reviews for their competitors.139 Various websites offer
full reimbursement of the purchase of certain products in exchange
for positive reviews, which are then reported by the unwitting
platforms as being made by “verified users.”140 Firms also use
negative rewards, i.e., sanctions, to deter consumers from sharing
negative reviews.141 Until the recent passing of the Consumer Review
Fairness Act, and perhaps continuing despite the law, firms would
include nondisparagement clauses in contracts with consumers.142 In
addition, firms sometimes threaten consumers with legal action for
defamation or use copyright law to argue that a review infringes on
Competition for Volunteers: A Case Study of the Online Market for Reviews, 10
B.E. J. ECON. ANALYSIS & POL’Y 1, 1–31 (2010).
137. See generally FTC, supra note 47, at 41–42 (reviewing evidence on
shilling and reporting some attempts by reputation platforms to curb shilling);
Kaitlin A. Dohse, Fabricating Feedback: Blurring the Line Between Brand
Management and Bogus Reviews, 2013 U. ILL. J.L. TECH. & POL’Y 363, 370–71
(reviewing some of the services that offer bogus reviews).
138. Nan Hu et al., Manipulation of Online Reviews: An Analysis of Ratings,
Readability, and Sentiments, 52 DECISION SUPPORT SYS. 674, 681 (2012)
(estimating fake reviews at 10 percent); Karen Weise, A Lie Detector Test for
Online Reviewers, BLOOMBERG BUSINESSWEEK (Sept. 29, 2011, 6:09 PM),
https://www.bloomberg.com/news/articles/2011-09-29/a-lie-detector-test-foronline-reviewers.
139. See Andreas Munzel, Malicious Practice of Fake Reviews: Experimental
Insight into the Potential of Contextual Indicators in Assisting Consumers to
Detect Deceptive Opinion Spam, 30 RECHERCHE & APPLICATIONS MARKETING 24,
41 (2015).
140. See, e.g., AMZDISCOVER, https://www.amzdiscover.com/blog/best-100amazon-review-groups-to-help-you-test-products/ (last visited Dec. 3, 2019);
AMZRC, https://amzrc.com/ (last visited Dec.3, 2019). A more extensive list of
websites like these is on record with the author.
141. Brad Tuttle, Guess Who’s Getting Some Pretty Awful Reviews: User
Review Sites, TIME (Sept. 21, 2013), http://business.time.com/2013/09/21/guesswhos-getting-some-pretty-awful-reviews-user-review-sites/.
142. Consumer Review Fairness Act of 2016, 15 U.S.C. § 45b(b)(1) (2012)
(voiding standard form contracts that include anti-disparagement clauses). See
Lucille M. Ponte, Protecting Brand Image or Gaming the System? Consumer
“Gag” Contracts in an Age of Crowndsourced Ratings and Reviews, 7 WM. & MARY
BUS. L. REV. 59, 59 (2016) (surveying the use of anti-disparagement clauses before
the law).

[p. 24]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1262 WAKE FOREST LAW REVIEW [Vol. 54
their copyright and should be taken down.143 Shilling strategies are
highly diverse, sophisticated, and reportedly quite potent.144
A related but less understood problem is that of “cherry-picking.”
Companies often selectively choose consumers who are most likely to
disseminate either favorable or unfavorable information and reward
them.145 As is very familiar, businesses offer thinly veiled bribes to
unhappy consumers in the form of reimbursements, free meals, or
“heartfelt” apologies.146 Celebrities and other influencers are also
more likely to receive special treatment in the hope that they will
share their (unrepresentative) experiences with their many
followers.147
Both shilling and cherry-picking result from strategic behavior
on behalf of firms. Both result in and emphasize more extreme
opinions, at the extreme of middling ones.148 This is because it would
not pay to invest in promoting middling reviews.149
B. Reputational Distortions
Placing reputation within a framework of individual rationality
allows us to draw meaningful conclusions about the integrity,
evolution, and credibility of reputational information.150 Based on the
143. See infra Subpart IV.B.
144. See Dohse, supra note 137, at 370–71 (reviewing online shilling
techniques and the services). For an updated list of shilling services and news,
see Opinion Spam Detection: Detecting Fake Reviews and Reviewers,
https://www.cs.uic.edu/~liub/FBS/fake-reviews.html (last visited Dec. 3, 2019).
145. See generally Shmuel L. Becher & Tal Z. Zarsky, Minding the Gap, 51
CONN. L. REV. 69, 90 (2018) (demonstrating how firms treat consumers based on
the threat the consumers pose to the firms’ revenue).
146. See generally Arbel & Kaplan, supra note 100, at 1216 (exploring the
corrosive effects of apologies on deterrence).
147. See Becher & Zarsky, supra note 145, at 90–91 (finding that firms often
consider consumers’ “online influence over peers” when deciding how to handle
complaints).
148. See, e.g., Miguel Helft, Charges Settled over Fake Reviews on iTunes, N.Y.
TIMES (Aug. 26, 2010), http://www.nytimes.com/2010/08/27/technology/27ftc.html
(discussing false reviews that “typically gave the games four or five stars”).
149. Astroturfing is a form of advertising, although a highly misleading one.
For some economic dynamics of reputation and advertising, see Kyle Bagwell,
The Economic Analysis of Advertising, in 3 HANDBOOK OF INDUSTRIAL
ORGANIZATION 1701, 1703 (Mark Armstrong ed., 2007); Phillip Nelson,
Advertising as Information, 82 J. POL. ECON. 729, 730 (1974); see also Lingfang
(Ivy) Li et al., Buying Reputation as a Signal of Quality: Evidence from an Online
Marketplace 2 (Nat’l Bureau of Econ. Research, Working Paper No. 22584, 2016)
(finding that quality sellers tended to participate more often in a program where
they offer rebates for (all) reviews of their products).
150. Professor Abbey Stemler recently provided an insightful account of such
biases in the context of the sharing economy where intimate interactions between
peers occur often (such as sharing a stranger’s house or car). Stemler, supra note
4, at 674. Unlike her account, I focus on developing the microfoundations of
reputation of consumer goods generally and explore how sophisticated, rational
consumers would process flawed reputational information.

[p. 25]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1263
framework developed in the last Subpart, three systemic distortions
will be expounded here, relating to participation, selection, and
content biases. The implications for consumer action are explored in
the following Part, but the overall arc of the argument is that in the
presence of these distortions, a reputation failure emerges, which
undermines the reliability of reputational information.
1. Reputational Sluggishness
Reputational sluggishness is the consequence of feeble, yet
existing, motivations to contribute to the public good of reputation.
On the one hand, reputation creators do not benefit financially from
creating reputational information.151 It is hard to commoditize
opinions and the transaction costs of doing so are prohibitive.152 On
the other hand, there are drivers that incentivize individuals to create
reputational information even in the absence of monetary
compensation. Altruism, desire for social recognition, gratitude, and
anger all provide reasons for people to create reputational
information that benefits others.153 Sluggishness emphasizes the
concern that for many individuals, or in many circumstances, these
benefits are insufficient. As a result, participation rates in reputation
creation are going to be low, leading reputational information to be
more slowly developed than is generally recognized.154
Empirical data, while wanting, suggests the broad scope of this
issue. One study found a sharing rate of fifteen out of a thousand
consumers.155 More optimistic estimates suggest a rate of one in
ten.156 In my analysis of product review data from Amazon, I found
that among electronics products with at least one review, the median
product only had two reviews.157 Moreover, few elect to write long
151. Today, reputational platforms reap most of the benefit of reputation
aggregation, but reputation’s direct producers receive very little reward. See also
Goldman, supra note 9, at 301.
152. Blockchain and cryptocurrencies may be promising solutions to such
transactions as they offer—in theory—trivial transaction costs. In the future, it
may be possible to commoditize opinions and employ a pay-per-use model.
153. See Lafky, supra note 106, at 555.
154. See generally Thomas R. Palfrey & Jeffrey E. Prisbrey, Altruism,
Reputation and Noise in Linear Public Goods Experiments, 61 J. PUB. ECON. 409,
410 (1996) (explaining that “altruistic behavior is illusionary or, at best, of minor
importance”).
155. Eric T. Anderson & Duncan I. Simester, Reviews Without a Purchase:
Low Ratings, Loyal Customers, and Deception, 51 J. MARKETING RES. 249, 251
(2014).
156. See, e.g., Andrew Thomas, The Secret Ratio That Proves Why Customer
Reviews Are So Important, INC., https://www.inc.com/andrew-thomas/the-hiddenratio-that-could-make-or-break-your-company.html (last visited Dec. 3, 2019)
(explaining that only one in ten satisfied customers will leave a review).
157. The mean was considerably higher at sixteen; the result of a few products
amassing many reviews. The analysis was based on data collected by Ruining
He & Julian McAuley, Ups and Downs: Modeling the Visual Evolution of Fashion

[p. 26]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1264 WAKE FOREST LAW REVIEW [Vol. 54
verbal reviews.158 Another indication comes from eBay. There, for
every negative review there are three times as many complaints to
customer service, which strongly indicates that many negative
reviews are either not generated or are deleted.159 In fact, this three
to one ratio seems like a lower bound on the scope of suppression of
opinions by users because it is reasonable that there would be many
more negative reviews than there would be active complaints. A more
speculative source of data, but interesting nonetheless, comes from
the sanitation reputation of restaurants in Los Angeles. In a series
of studies, researchers attempted to establish the effect of a law that
required restaurants to disclose their sanitation ratings.160 In a naïve
model of reputation the impact of such a law on food-borne illnesses
should be relatively small.161 If a person contracts such an illness,
then conditions are ripe for the word to travel fast: a food-borne illness
is highly salient, it is moderately easy to establish its cause, and it is
of great interest to prospective diners.162 Then, mandatory disclosure
of sanitation levels would then not be expected to have a significant
effect on food-borne illness because the information would already
exist throughout the market. Despite that, the study found that the
law had a powerful effect, with a sharp decline in hospitalizations due
to foodborne illnesses.163 The law’s effectiveness is amenable to a few
explanations, but one is that the reputation system was too congested
to work properly before the law—despite the ideal background
conditions.
Trends with One-Class Collaborative Filtering, U.C. SAN DIEGO JACOBS SCH.
ENGINEERING, http://cseweb.ucsd.edu/~jmcauley/pdfs/www16a.pdf.
158. Most reviews on Amazon for electronics are in the range of one hundred150 characters, or about half a paragraph. See Max Woolf, A Statistical Analysis
of 1.2 Million Amazon Reviews, MAX WOOLF’S BLOG (June 17, 2014),
http://minimaxir.com/2014/06/reviewing-reviews/.
159. See Nosko & Tadelis, supra note 4, at 9 (concluding “there are a
substantial number of transactions that went badly for which negative feedback
was not left”).
160. Ginger Zhe Jin & Phillip Leslie, The Effect of Information on Product
Quality: Evidence from Restaurant Hygiene Grade Cards, 118 Q.J. ECON. 409, 410
(2003) [hereinafter Jin & Leslie, Effect of Information]; see also Ginger Zhe Jin &
Phillip Leslie, Reputational Incentives for Restaurant Hygiene, 1 AM. ECON. J.:
MICROECONOMICS 237, 238 (2009) [hereinafter Jin & Leslie, Reputational
Incentives]; Paul A. Simon et al., Impact of Restaurant Hygiene Grade Cards on
Foodborne-Disease Hospitalizations in Los Angeles County, 67 J. ENVTL. HEALTH
32, 32 (2005). The local health department collected these ratings long before
restaurants were required to disclose them. Jin & Leslie, Effect of Information,
supra, at 410; Jin & Leslie, Reputational Incentives, supra, at 238.
161. See Simon et al., supra note 160, at 32 (explaining that some studies have
not found a connection between low department of health inspection scores and
foodborne-disease outbreaks at restaurants).
162. See Jin & Leslie, Reputational Incentives, supra note 160, at 238 (“Local
customers can learn about a restaurant’s hygiene quality by repeatedly
patronizing the restaurant, by talking to friends who have patronized the
restaurant, or through exposure to local news reports about the restaurant.”).
163. Jin & Leslie, Effect of Information, supra note 160, at 439–40.

[p. 27]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1265
2. Regression to the Extreme
So far, we saw that only some consumers would choose to produce
reputational information, but this leaves the question of who those
consumers are. If the sample of consumers who produce reputation
is randomly selected, then we would expect the outliers to experience
what statisticians call “regression to the mean,” i.e., the eventual
balancing of outliers towards the mean of the group.164 Indeed, the
regression to the mean will be impeded by sluggishness, but there is
the possibility of self-correction over time with a randomly selected
sample. Unfortunately, the selection of consumers is all but random.
Regression to the extreme is the propensity of reputational data
to emphasize, rather than eliminate, outlier experiences over time.
Internal motivations select against middling reviews because those
reviews are based on experiences that are too “boring” to generate the
requisite sense of spite or gratitude that will overcome the costs of
producing reputational information.165 Additionally, reciprocity
norms would lead consumers to overly represent positive experience,
in hopes of receiving reciprocal reviews from sellers,166 and herding
would tend to silence nonpopular reviews that might betray the
consumer’s lack of sophistication.167 If a bottle of French wine
receives paeans, an individual consumer may be embarrassed to
reveal that she did not like it, noted no accents of “forest floor,” and
was not seduced by its “interplay of plump grapes and jazzy oak.”168
Lastly, financial incentives select against middling reviews because
shilling and cherry-picking foster creation of extreme opinions. All
these tendencies lead to “regression to the extreme”: the propensity of
reputational data to emphasize, rather than eliminate, outlier
experiences over time.169
Product reviews consistently provide strong evidence of
regression to the extreme. One might expect that most products sold
on the market would follow some generalized, bell-shaped (Gaussian)
distribution—after all, very few products are really outstanding or
truly atrocious. Instead, most reviews on a large variety of online
platforms form a so-called “J-shaped distribution,” with most reviews
164. Stephen M. Stigler, Regression Towards the Mean, Historically
Considered, 6 STAT. METHODS MED. RES. 103, 103–05 (1997).
165. Nan Hu et al., Overcoming the J-shaped Distribution of Product Reviews,
52 COMM. ACM 144, 145 (2009).
166. See Stemler, supra note 4, at 692.
167. Id. at 693.
168. Wine Description of Lewis, Cabernet Sauvignon Napa Valley 2014,
WINE SPECTATOR TOP 100, http://top100.winespectator.com/wine/wine-no-1/;
Wine Description of Orin Swift, Machete California 2014, WINE SPECTATOR TOP
100, http://top100.winespectator.com/wine/6-orin-swift/.
169. See Hillel J. Bavli, The Logic of Comparable-Case Guidance in the
Determination of Awards for Pain and Suffering and Punitive Damages, 85 U.
CIN. L. REV. 1, 17 (2017).

[p. 28]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1266 WAKE FOREST LAW REVIEW [Vol. 54
amassed in the extremes.170 On Amazon, more than 72 percent of the
products have an average rating of at least four stars.171 In Airbnb
listings, the average rating is 4.7 stars.172 Studies repeatedly find
that middling reviews are rare and that even products with an
average rating of two or three stars have only a few middling
reviews.173 Further evidence suggests this pattern is not unique to
online settings but carries over to offline settings.174 Figure 1 shows
the distribution of 1.2 million electronic products listed on Amazon,175
while Figure 2 shows a comparison of three specific products.176
170. See, e.g., Chrysanthos Dellarocas & Ritu Narayan, A Statistical Measure
of a Population’s Propensity to Engage in Post-Purchase Online Word-of-Mouth,
21 STAT. SCI. 277, 279–80 (2006); Yi-Chun Ho et al., Disconfirmation Effect on
Online Rating Behavior: A Structural Analysis, 28 INFO. SYS. RES. 626, 630
(2017); Hu et al., supra note 165, at 144–45 (detailing evidence from Amazon and
arguing that the J shaped distribution “contradicts the law of ‘large numbers’
that would imply a normal distribution”); Hu et al., supra note 109, at 328
(finding that 54 percent of all products on Amazon have a review distribution
that is neither normal or bimodal, and 35 percent have a unimodal, nonnormal
distribution); Woolf, supra note 158.
171. See Wu & Huberman, supra note 106, at 3; Woolf, supra note 158.
172. See Zervas et al., supra note 15, at 3.
173. See Lafky, supra note 106, at 556.
174. See Eugene W. Anderson, Customer Satisfaction and Word of Mouth, 1
J. SERV. RES. 5, 6–7 (1998).
175. See Woolf, supra note 158.
176. Amazon.com Customer Reviews: Culture War? The Myth of a Polarized
America, AMAZON, www.bit.do/RBProd1 (last visited Dec. 3, 2019); Amazon.com
Customer Reviews: $20 PlayStation Store Gift Card [Digital Code]: Video Games,
AMAZON, www.bit.do/RBProd2 (last visited Dec. 3, 2019); Amazon.com Customer
Reviews: BIC Cristal For Her Ball Pen, 1.0, Black, 16ct, AMAZON,
www.bit.do/RBProd3 (last visited Aug. 28, 2019).

[p. 29]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1267
FIGURE 1. DISTRIBUTION OF RATINGS OF 1.2 MILLION ELECTRONIC
PRODUCTS LISTED ON AMAZON
FIGURE 2. REVIEWS OF THREE SAMPLE PRODUCTS

[p. 30]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1268 WAKE FOREST LAW REVIEW [Vol. 54
Although the J-shaped distribution strongly supports the idea of
regression to the extreme, alternative explanations should also be
considered. One alternative is that market dynamics push lowquality products out of the market.177 Because such pressures leave
only high-quality products on the market, reviews would be
(accurately) amassed in the right tail. While worthy of further
investigation, this explanation appears unlikely for a number of
reasons. Even if the products on the market are of high quality, there
should still be some middling reviews,178 and, in particular, there
should be more middling reviews than negative reviews.179 This
explanation suggests that high ranking is indicative of quality, but in
fact, rankings of identical products listed on different platforms are
often negatively correlated, such that a high ranking of the same
product in one place does not predict a high ranking elsewhere.180
A more general issue, and one that clouds other alternative
explanations, is that voluntary consumer reviews systematically
diverge from other types of evaluations. Across a variety of products,
professional testing of the same products listed on consumer websites
shows low correlation with consumer reviews.181 Not only does the
average quality differ, but a systematic difference in the distribution
of opinions also exists. While consumers’ reviews follow the noted Jdistribution, professional reviews of the same products follow a bellshaped distribution.182 One can almost hear the exasperation in the
voice of the researchers who concluded that “critics are more normal
than normal users.”183 If professional reviewers are competent, one
would expect a strong correlation between their judgments and
consumer reviews; the lack of such correlations suggests that at least
one of these sources of information is amiss. To test whether the
professional reviewers are actually inaccurate, consider the following
experiment. 218 participants, none of them a professional critic, were
177. A related theory is that consumers select products that they will probably
like and so it is expected that there will be a concentration of satisfied consumers.
See Nilesh Dalvi et al., Para ‘normal’ Activity: On the Distribution of Average
Ratings, 7 PROC. 7TH INT’L AAAI CONF. WEBLOGS & SOC. MEDIA 110, 111, 115
(2013). This theory, however, needs to explain why, in the absence of regression
to the extreme, there are so few middling reviews, which are to be expected in
light of possible consumer mistakes among similarly highly rated products.
178. See Hu et al., supra note 165, at 145–46.
179. See Stigler, supra note 164, at 104–05.
180. See de Langhe et al., supra note 14, at 826.
181. See generally Roberto Centeno et al., On the Inaccuracy of Numerical
Ratings: Dealing with Biased Opinions in Social Networks, 17 INFO. SYS.
FRONTIERS 809, 809 (2015) (discussing how reputation rankings within current
social networks are likely skewed due to subjectivity issues); Dalvi et al., supra
note 177 (discussing a theory that consumers select products that they will
probably like, thus leading to concentrations of satisfied customers); de Langhe
et al., supra note 14, at 818 (comparing online reviews to reviews of the same
products in Consumer Reports).
182. See Centeno et al., supra note 181, at 811.
183. See Dalvi et al., supra note 177, at 114.

[p. 31]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1269
asked to review a product that—unbeknownst to them—was also
listed on Amazon.184 The experimental reviews followed the bell
curve, unlike their Amazon counterparts, as illustrated in reproduced
Figure 3 below:185
FIGURE 3. AMAZON REVIEWS VS. REVIEWS BY TEST PARTICIPANTS
60%
50%
40%
30%
20%
10%
0%
0 1 2 3 4 5 6
Amazon Experiment
3. Reputation Integrity
The last issue is the concern that even when consumers have an
incentive to report experiences, they may face adverse incentives
about the content of their reviews. For multiple reasons, individuals
may misstate the quality of their own experience. Consider, for
example, social pressures to conform, financial incentives, a desire to
avoid confrontation, an endowment effect, personal style, and other
similar considerations.186 All of these may lead individuals to report
their experiences more or less favorably than they actually were.
Threats to the integrity of reputational information are hard to
measure through data itself, but there is evidence that exposure to
the opinions of others will make others report more or less favorably
about their own experiences. Sociologist Ronald Burt noted, in the
184. Hu et al., supra note 165, at 145–46.
185. The graph reproduces the data presented in Id. at 146.
186. Ronald Burt, Gossip and Reputation, 9–13 (2008)
https://faculty.chicagobooth.edu/ronald.burt/research/files/GR.pdf. This source
was taken from Ronald Burt’s faculty website; it was a preprint of a chapter to
appear in Management et réseaux sociaux: ressource pour l'action ou outil
degestion?, edited by Marc Lecoutre and Lievre Pascal, Editions Hermes-Lavoisier, and published in 2008.

[p. 32]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1270 WAKE FOREST LAW REVIEW [Vol. 54
context of gossip and stories told about others, that “[a]ccuracy is a
nicety more than a requirement for the stories.”187 Thus, he finds
significant echo chamber effects, where people shape the valance of
reputational information on the basis of context and audience rather
than merit.188 One such experimental finding is made by Tory
Higgins, who gave research subjects a description of a person called
Donald.189 The key was that the descriptions were very ambiguous
about whether Donald had positive or negative characteristics.190
Then, a confederate entered the room and said that he “kinda likes”
or “kinda dislikes” Donald, and asked the subjects for their opinion.191
The subjects then offered a distorted view of Donald that accorded
with the confederate’s disposition.192
* * *
In sum, the theory of the microfoundations of reputation suggests
that (1) there will be a trend in reputational data towards more
extreme reviews, (2) that the integrity of information will be
compromised, and (3) that the volume of reputational data will be
constrained by reputational sluggishness. The predictions of this
theory are consistent with available data, which show that significant
divergence exists between voluntary consumer reviews and other
measures of quality.193 While there is much to be desired in the way
of additional evidence, the existing data come from millions of
different products and across various platforms. All in all, the case
for distortions seems robust, given current knowledge.
C. Flawed Information, Flawed Decisions
When consumers make purchase decisions, one key type of
information they seek is data on the experiences of past consumers.
From the perspective of a prospective consumer, it is useful to know
how frequently the product or service resulted in a favorable
experience along some dimension (e.g., quality of food, promptness of
service, durability).194 Consumers seek to extrapolate from these data
to predict their own individual experience despite the obvious
187. Id. at 1.
188. Id. at 9–10.
189. E. Tory Higgins & William S. Rholes, “Saying is Believing”: Effects of
Message Modification on Memory and Liking the Person Described, 14 J.
EXPERIMENTAL SOC. PSYCHOL. 363, 366 (1978).
190. Id. at 366–67.
191. Id. at 367.
192. Id. at 368–70, 374–77. For more examples of social pressures, see
discussion and notes supra Subpart III.A.3.
193. See supra notes 172–87 and accompanying text.
194. Xinxin Li & Lorin M. Hitt, Self-Selection and Information Role of Online
Product Reviews, 19 INFO. SYS. RES. 456, 459–60 (2008).

[p. 33]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1271
differences in taste among individuals.195 Moreover, consumers care
about more than the average expected experience—whether it is good
or bad on average—but also about its variance, e.g., a phone that
generally functions perfectly but will on rare occasions explode could
have the same average rating as a phone that is consistently
mediocre.196 To the risk-averse consumer, the high-variance phone
would be inferior.197 To optimize decisions, then, consumers would
like to have access to the distribution of past consumer experiences,
rather than just their average.
Understood this way, the deleterious effect of sluggishness,
integrity, and regression to the extreme become apparent as they
make estimation less accurate. This loss of accuracy is because both
the quantity and quality of reputational information themselves are
jeopardized. The goal of this Part is to study these effects, using both
theory and a Monte Carlo simulation. One key insight from the
simulation, which is worth emphasizing here and throughout, is that
not all reputation failures are born equal. Some may lead to small
distortions that are largely inconsequential. Understanding the
circumstances under which reputation failures are most severe is key
to policymaking but, unfortunately, is largely outside the ambit of
this paper.
1. Informational Distortions
Sluggishness makes quality estimation difficult because it limits
the quantity of available data. Because consumers lack incentives to
share reputational information data exist for only a fraction of all
consumers. This limited quantity of information has two related
adverse effects. First, sluggishness leads outlier, unrepresentative
experiences to appear more common than they actually are—there is
not enough “regular use” data to contradict them.198 If, by chance
alone, one of these consumers had an extreme but unrepresentative
experience, an outlier, this will taint the perception of the product.
Sluggishness prolongs the time it takes for more reviews to
accumulate and correct the noise.199 Second, less information also
195. Id. at 459.
196. Hayley Tsukayama, How Samsung Moved Beyond Its Exploding Phones,
WASH. POST (Feb. 23, 2018), https://www.washingtonpost.com/business/howsamsung-moved-beyond-its-exploding-phones/2018/02/23/5675632c-182f-11e8b681-2d4d462a1921_story.html.
197. See Eyal Zamir, Loss Aversion and the Law, 65 VAND. L. REV. 829, 872
(2012) (articulating that individuals who perceive losses as more painful than
potential gains are less inclined to pursue the potential gains).
198. Consider an example of a product that has, on average, a three-star
quality. If the first consumer, by chance, ranks it at one star, then it would take
two higher-than-average consumers rankings of four stars to correct this
impression.
199. For an analysis of the dynamic evolution of reputation, including the
possibility that early adopters may be systematically different in their

[p. 34]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1272 WAKE FOREST LAW REVIEW [Vol. 54
means less information regarding the distribution of experiences.
Even if the average is accurate, consumers also care about the
distribution, but sluggishness limits the volume of available
information.200
The problem of sluggishness is the same familiar problem of
surveys with small sample sizes. Figure 4 illustrates, using randomly
generated values, how much sluggishness can distort one’s
impression of products. Both graphs track the distribution of reviews
that were given by a sample of all consumers who chose to share their
reviews of the same product. In the top graph, only ten consumers
chose to do so, whereas in the bottom graph, one hundred consumers
shared their reviews. The dashed line indicates the full distribution
of all consumer experiences while the full line denotes the estimated
distribution based on the limited number of consumer reviews. As
can be seen, a smaller sample distorts one’s view of both the mean
and the distribution.
preferences and views from standard consumers, see generally Li & Hitt, supra
note 194.
200. See Li & Hitt, supra note 194, at 457–58, 463 and accompanying text.

[p. 35]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1273
FIGURE 4. ESTIMATED AVERAGE VS. REAL AVERAGE AS THE NUMBER
OF OPINIONS INCREASE
The next distortion concerns the quality of information and is
caused by regression to the extreme. Because incentives to report
experiences are weakest when the experiences are rote or bland, very
few reviews fall in the middle range, leaving only extreme reviews
reported.201 Trying to infer quality based on such a sample involves
a thorny statistical problem known as “middle censoring.” From a
statistical perspective, most methods of estimation assume that the
201. See Hu et al., supra note 165, at 145.

[p. 36]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1274 WAKE FOREST LAW REVIEW [Vol. 54
sample is taken from a random sample.202 If, instead, subjects selfselect—as is the case here—then this bias could undermine the
validity of statistical inferences. Figure 5 demonstrates the potential
implications of regression to the extreme. Similarly, the figure
collects, using randomly generated data, different consumer
reviews—with a sample of one hundred participating consumers—
but it omits the reviews of people with middling reviews who
presumably lacked an incentive to share. A prospective consumer,
seeking to decide whether to buy the product, only observes the filled
bars (the empty bars are not visible and illustrate the distribution of
unreported experiences). The full lines again mark the consumer’s
best guess about the mean and distribution based on this limited
information.203 The difference between the estimated mean and the
mean of all experiences (dashed) is highlighted by the arrow. As can
be seen, naïve estimation methods would yield widely inaccurate
outcomes.
FIGURE 5. ESTIMATED AVERAGE VS. REAL AVERAGE AS THE NUMBER
OF OPINIONS INCREASE
Dashed = Estimated Average; Full = Average of all consumers.
The shorter the gap between the curves and the means, the more
accurate the estimate
Notedly, Figure 5 illustrates that the extremes do not “even
out.”204 It may seem that in a large sample, extreme results on one
202. See DAVID S. MOORE & GEORGE P. MCCABE, INTRODUCTION TO THE
PRACTICE OF STATISTICS (5th ed., 2006).
203. See infra Subpart III.C.2.
204. See Bavli, supra note 169, at 17.

[p. 37]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1275
end will be balanced by extreme results on the other end. This logic,
however, only applies to symmetric distributions—not the positivelybiased J distribution here.205
Finally, there are problems with the integrity of information.
Herding (or antiherding) is a highly path-dependent phenomenon.
Some products will appear to draw more and more favorable reviews,
but this can be the result of chance that led the group of first
consumers to have a favorable experience.206 Or, a product may
appear overly negative simply because a random group of first
consumers experienced some rare issues.
In sum, for a prospective consumer to accurately estimate the
quality of the underlying good, both the quantity and quality of
reputational information are essential. Sluggishness and regression
to the extreme make reputational information scarce and biased.
Consequently, the estimated mean and distribution of reported
reviews systematically diverge from the actual mean and
distribution, thus making them unreliable as sole guides for
consumer decision-making. The question still remains, however,
whether consumers can adjust for these distortions.
2. Overcoming Bias
Distorted information is likely to have a strong effect on
consumers. Survey after survey, consumers express strong
confidence in reputational information, describing it as a reliable
source of information.207 While it is unclear whether consumers take
reputation at face value, the level of their confidence is at least
suggestive of the former. Moreover, evidence shows a strong
monotonic relationship between ratings and sales—a half-star
increase in a restaurant rating resulting in a 19 percent higher
likelihood that the restaurant would sell out and another half-star
increase resulted in a 5 percent–9 percent increase in revenues.208
Consumers can also be affected by distorted information through a
205. See Hu et al, supra note 165, at 144 (“[T]he average is statistically
meaningful only when it is based on a unimodal distribution, or when it is based
on a symmetric bimodal distribution. However, since product systems have an
asymmetric bimodal (J-shaped) distribution, the average is a poor proxy of
product quality.”).
206. See Stemler, supra note 4, at 693 (discussing evidence of herding in
online reviews).
207. See generally Rosie Murphey, Local Consumer Review Survey: Online
Reviews Statistics & Trends, BRIGHTLOCAL (Dec. 7, 2018),
https://www.brightlocal.com/learn/local-consumer-review-survey/ (finding that
89 percent of consumers read online reviews for local businesses and that 91
percent of eighteen-to thirty-four-year-old consumers trust online reviews as
much as they trust personal recommendations).
208. See Michael Anderson & Jeremy Magruder, Learning from the Crowd:
Regression Discontinuity Estimates of the Effects of an Online Review Database,
122 ECON. J. 957, 966 (2012); Michael Luca, Reviews, Reputation, and Revenue:
The Case of Yelp.com 2 (Harv. Bus. Sch., Working Paper No. 12-2016, 2016).

[p. 38]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1276 WAKE FOREST LAW REVIEW [Vol. 54
variety of behavioral, cognitive limitations,209 most notably
anchoring. Anchoring is the well-replicated psychological
phenomenon that describes how the introduction of arbitrary and
irrelevant numbers affects the outcomes of negotiations, evaluations,
and work performance ratings.210 If the consumer is exposed to
inflated reviews, then this can anchor an inflated sense of value.211
In addition, even if consumers learn that the data is biased, it is
unclear that they can effectively discount it.212 In a set of studies,
researchers investigated how individuals reacted when they learn
that they receive biased advice.213 Participants were asked to
estimate the cost of a house in Pittsburgh.214 To aid them, they were
given an estimate by a local realtor who knew the local market.215
However, the realtor also had an incentive to exaggerate her
estimate, because her commission was based on the final sale price.216
Surprisingly, the control group that was unaware of the realtor’s bias
had more accurate estimates than the treatment group which was
informed that the realtor had a conflict of interest.217 Finally, even if
consumers were capable of mentally detaching from these effects, it
is unclear that most have the statistical literacy to effectively
discount online data.218 The median American will likely not
understand what it means to be median.219 In a famous experiment,
respondents insisted that a person described to them is less likely to
209. See Bar-Gill, supra note 5, at 749.
210. See, e.g., Adrian Furnham & Hua Chu Boo, A Literature Review of the
Anchoring Effect, 40 J. SOCIO-ECON. 35, 35 (2011).
211. See id.
212. See Daylian M. Cain et al., When Sunlight Fails to Disinfect:
Understanding the Perverse Effects of Disclosing Conflicts of Interest, 37 J.
CONSUMER RES. 836, 845, 847 (2011) (finding an absence of the ability to
effectively discount the biased information when disclosed).
213. George Loewenstein et al., The Limits of Transparency: Pitfalls and
Potential of Disclosing Conflict of Interest, 101 AM. ECON. REV.: PAPERS & PROC.
423, 425 (2011).
214. See id.
215. See id.
216. See Cain et al., supra note 212, at 840–41 (finding that groups that were
aware of the bias had a higher variation in results).
217. See id. at 845, 847.
218. See, e.g., Laurent E. Calvet et al., Measuring the Financial Sophistication
of Households, 99 AM. ECON. REV. 393, 393 (2009) (“Many households invest in
ways that are hard to reconcile with standard financial theory and that have been
labelled as investment mistakes.”); Mark Grinblatt, et al., IQ, Trading Behavior,
and Performance, 104 J. FIN. ECON. 339, 360 (2012) (finding that measured levels
of IQ affect stock market sophistication).
219. See Pranjal Gupta & Judy Harris, How E-WOM Recommendations
Influence Product Consideration and Quality of Choice, 63 J. BUS. RES. 1041, 1042
(2010) (explaining how research finds that consumers sometimes lack motivation
to process information, sometimes trusting even a single data point).

[p. 39]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1277
be a bank teller than be both a bank teller and a feminist.220 But this,
of course, cannot be. Obviously there are more bank tellers than there
are bank tellers who are also feminists, but still, people find it difficult
to intuit statistical judgments.221
Consequently, the law is generally skeptical of consumers’
abilities to correct biased data, even in situations where consumers
may be aware of the existence of distortions and where third-party
services may be used to correct them.222 Such are, for example, the
limits on contractual misrepresentation, investor fraud through
pump-and-dump strategies, false advertising, defamation, and false
lights. The pump-and-dump scheme is especially telling because it
involves the dissemination of wrong reputational information about
firms.223 Even though investors may be thought to be, on average,
somewhat more sophisticated than consumers and even though it
may be clear to those investors that pump-and-dump schemes take
place, the law still chooses to proscribe such activities, fearing that
consumers will not be able to compensate for such misleading
strategies adequately.224
While it is clear that reputation failure could have a strong effect
on most consumers, in the rest of this Part, I focus on a harder
question. Can consumers—at least those that are rational,
sophisticated, and informed—overcome these distortions? After all,
it is fair to assume that many consumers at least suspect that
reputational information should not be taken at face value. This Part
investigates these issues, using both examples and computer
simulations to evaluate three central consumer strategies: cardinal
evaluations—i.e., choosing a product based on its score or mean;
ordinal ranking—i.e., choosing the relatively better-rated product;
and evaluating qualitative information—i.e., choosing on the basis of
the content of reviews.
Cardinal Evaluations. Suppose that a consumer is trying to
estimate the quality of a hypothetical product based on the valence of
reviews. She knows that reviews in the middle, rated two or three
220. Amos Tversky & Daniel Kahneman, Extensional Versus Intuitive
Reasoning: The Conjunction Fallacy in Probability Judgment, 90 PSYCHOL. REV.
293, 297 (1983).
221. But see Berit Brogaard, Linda the Bank Teller Case Revisited, PSYCHOL.
TODAY (Nov. 22, 2016), https://www.psychologytoday.com/us/blog/thesuperhuman-mind/201611/linda-the-bank-teller-case-revisited.
222. See Truth In Advertising, FTC, https://www.ftc.gov/news-events/mediaresources/truth-advertising (last visited Dec. 3, 2019) (showing the Federal Trade
Commission aims to protect consumers by enforcing federal law which says
information given to consumers must be truthful and not misleading).
223. See Pump and Dump, INVESTOPEDIA, https://www.investopedia.com/
terms/p/pumpanddump.asp (last updated Apr. 26, 2019).
224. In the unregulated space of Bitcoin, a recent research paper found that
the market price of bitcoin rose up in 2013 tenfold due to manipulative trading
tactics by a single trader. Neil Gandal et al., Price Manipulation in the Bitcoin
Ecosystem, 95 J. MONETARY ECON. 86, 87 (2018).

[p. 40]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1278 WAKE FOREST LAW REVIEW [Vol. 54
stars, are suppressed, so she only sees extreme reviews. Figure 6 lists
the information that is available to her. Armed with the knowledge
that middling reviews are censored, what can she say about the
quality of the underlying product? What would she believe the mean
to be? How confident should she be?225
FIGURE 6. FREQUENCY OF REVIEWS UNDER REGRESSION TO THE
EXTREME
Bars show the number of reviews in each ranking category
Next, based on this analysis, which of Figures 7 and 8 best
represents the quality of the underlying product?
225. As I shall argue, consumers care about more than the mean, but average
star rating is often the first filter consumers use in their searches.

[p. 41]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1279
FIGURES 7 & 8. FREQUENCY OF POTENTIAL REVIEWS UNDER
REGRESSION TO THE EXTREME
Bars, full and empty, show the number of reviews in each
ranking category
Of course, these questions are not answerable. In fact, these
figures are only two of many possible distributions. There is just not
enough information to make accurate so-called cardinal evaluations,
i.e., determinations of the quality of the product on the basis of review
valence.226 In particular, estimating the mean on the basis of a
226. If one has enough data about the relationship between full reviews (as in
reviews solicited from all consumers) and voluntary reviews, it may be possible
to make some educated guesses to fill in the missing data. Whether this will be

[p. 42]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1280 WAKE FOREST LAW REVIEW [Vol. 54
truncated sample is a risky proposition, doubly so when the sample is
small due to sluggishness, and triply so when the data is misstated
due to integrity bias. Still, consumers often try to estimate quality on
the basis of review valence, especially by limiting their searches to
products above a certain mean.
Ordinal Comparisons. Suppose that the consumer reluctantly
accepts that means are problematic, and instead seeks to compare
among products, reasoning that if all are subject to biases, at least
the comparison of the means would reveal which one is superior.227
The following Table describes two products that the consumer is
trying to compare; the shaded area is middling information that is not
available to her. On the basis of available information, she takes the
mean of product A to be 2.8 and that of product B to be also 2.8. She
also notes that they both have the same distribution of reviews. She
concludes that the two products are of equal value. Based on your
knowledge of the shaded information, is this a correct conclusion?
TABLE 1. ORDINAL COMPARISON WITH TRUNCATED DATA
To give a sense of the scope of mistakes based on interproduct
comparisons, I conducted various Monte Carlo simulations, reported
in Figures 9, 10, and 11.228 Monte-Carlo simulations are a computerassisted technique used in finance, physics, and computer science to
track complex interactions in domains where the parameter space is
possible, the accuracy of such process and its transferability across domains
remains to be seen.
227. It is worth recalling that statistical tests of significance of mean
difference (Student’s T-Test) are unhelpful when the sample is not randomly
chosen. See MOORE & MCCABE, supra note 202, at 463. Nonparametric tests also
depend on various assumptions that may not hold in these contexts. Id.
228. The computer simulation is on file and can be replicated by the reader.
Given the use of randomness, actual results may vary slightly among runs, but
given the large volume of trials, this deviation will not affect any of the
conclusions here.

[p. 43]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1281
large and uncertain.229 By running thousands of simulated
experiments, each with random deviations, the experimenter can
learn of trends in the data.230 Methodologically, the Monte Carlo
simulation is akin to a quasi experiment; it is useful in demonstrating
the existence of certain phenomena and indicating their potential
magnitude, although it is not at the epistemic level of natural
experiments as it only studies possibilities rather than actual
quantities.231
The first simulation generated two products of arbitrarily chosen
mean quality (although they shared all other statistical features).
Each product was “tried” by one hundred different “consumers,”
which means that each consumer has a random experience based on
the quality of the product. The higher mean product was more likely
to generate a favorable experience. After trying the product, each
consumer could report the experience by rating it from one to five
stars in half-star increments. Because of regression to the extreme,
the consumers are coded not to share experiences in the range of two
to three stars. Once all the information accumulates, a new consumer
comes and tries to decide which product to purchase on the basis of
interproduct comparisons. She chooses the one with the higher
reported mean, based on the reasoning noted above. The code then
counts every instance where the consumer was misled into choosing
the inferior product. This process, for the same products, was
repeated ten thousand times. This gives an account of the frequency
of mistakes, given products of different means. To see how a higher
difference would affect the frequency of mistakes, the simulation then
ran the same process but increased the mean of product B by a slight
amount. The following figures report this simulation, and two others,
explained below.
229. See Ankita Bihani, A New Approach to Monte Carlo Simulation of
Operations, 8 INT’L J. ENGINEERING TRENDS & TECH. 218, 218 (2014). A recent
example is the use of a Monte-Carlo Simulation to evaluate the rarity of
intelligent life in the universe. See Anders Sandberg, Eric Drexler & Toby Ord,
Dissolving the Fermi Paradox, CORNELL U. (June 8, 2018), https://arxiv.org/pdf
/1806.02404.pdf.
230. See Bihani, supra note 229, at 218–19.
231. See Robert L. Harrison, Introduction To Monte Carlo Simulation 2 (Jan.
1, 2011) (unpublished manuscript) (on file with the National Institutes of
Health), https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2924739/pdf
/nihms219206.pdf.

[p. 44]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1282 WAKE FOREST LAW REVIEW [Vol. 54
FIGURE 9. DIFFERENCE IN MEAN
FIGURE 10. DIFFERENCE IN SALES VOLUME

[p. 45]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1283
FIGURE 11. DIFFERENCE IN STANDARD DEVIATION
Figure 9 shows how the mean quality of the underlying goods
affects consumer mistakes.232 For products that are not clearly
distinguishable, i.e., they have a somewhat similar mean, consumer
mistakes are widespread. For example, if the difference in mean is
0.1 stars, then under the parameters of the simulation consumers
would choose the wrong products in 25 percent of the cases. For
products that are even harder to distinguish, with only a 0.05 stardifference, the ratio of mistakes rises to 35 percent. On the flip side,
the more different the products are, the fewer mistakes consumers
commit, despite regression to the extreme. When the difference is
0.25 stars, the ratio of mistakes falls below 10 percent, and when it is
0.5 stars, it mostly disappears.
Figure 10 reports the same simulation, but this time it holds the
mean difference constant at 0.3 stars and only varies the sale volume.
One product is reviewed throughout by one hundred consumers,
whereas the other is reviewed by a variable number of consumers.
When the second product is reviewed by only fifteen consumers, the
ratio of mistakes raises to about 20 percent. This is important, in part
because the difference in means is relatively significant (0.3 stars),
and in part because most products have very few reviews, making this
scenario likely.233
232. Part of the distortion of reputational information is also due to the use of
integers or half integers (i.e., a consumer reports a 2.5-star review, whereas the
actual experience is 2.34). Despite its flaws, the integer ranking system is almost
universal.
233. As noted, the median product only has two reviews. See supra note 157
and accompanying text.

[p. 46]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1284 WAKE FOREST LAW REVIEW [Vol. 54
Finally, Figure 11 reports the same simulation but this time
holding the mean and sale volume equal, and only changing the
variability of experiences. From a consumer perspective, in choosing
between two products with equal means, the one with the lower
variability would be preferred due to risk aversion. Here we find the
most considerable degree of mistakes. When one product yields
consistent experiences and the other variable ones, the ratio of
mistake is very high—when the variance of one product is 0.1 stars,
but the variance of the other is 0.5 stars, consumers mistakenly prefer
the inferior product in over 80 percent of the cases. Only when both
products are highly variable does the ratio of mistakes start to fall.
Taken together, and under some important caveats, these
simulations demonstrate the potential scope of consumer errors given
reputation failure. At the same time, the simulations also
demonstrated a broad range of cases where reputation failure is
unlikely—specifically, environments where there is a great difference
in product quality, when the sale volume is large, and when quality
variability is large or constant across products. This conclusion is
important in evaluating the social harm and the contexts in which it
is likely to arise from reputational distortions. An important caveat,
however, is that these simulations are based on stylized examples and
use arbitrarily chosen parametric values. This limits the
interpretation of the results reported here. On the other hand, reallife considerations tend to increase the problematic nature of reviews
relative to the simulations. For example, satisfied consumers may be
more or less likely to report their experiences than disgruntled
consumers.234 After all, the tendency to complain is different from the
tendency to praise, and spite is not simply gratitude multiplied by
negative one. Other practical complications involve consumers
ascribing different meaning to star reviews (for some, a four-star
review means high-valence, while for others, it will indicate a
negative experience); the possibility that some products will have
bimodal or other nonstandard distributions; the dynamic effect
caused by buyers experimenting less with low-reviewed products; and
firms investing different amounts of efforts in shilling and cherrypicking. These considerations would tend to make not only the
simulations less reliable but also any quantitative approach to the
data.
Qualitative Analysis. Suppose now that the consumer seeks to
only read reviews and focus only on qualitative content. In particular,
she is trying to decide between two brands of toilet paper sold on one
234. See Interview with anonymous medium-sized seller on Amazon of
organic products for babies (Jan. 19, 2018) (claiming that disgruntled consumers
are more prone to writing reviews than very happy ones); cf. Chrysanthos
Dellarocas & Charles A. Wood, The Sound of Silence in Online Feedback:
Estimating Trading Risks in the Presence of Reporting Bias, 54 MGMT. SCI. 460,
460 (2008) (finding that satisfied consumers are more prone to write reviews).

[p. 47]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1285
platform. It turns out that there are almost 8000 different reviews
for the two brands.235 How long would it take her to read them all?
How confident should she be in her ability to spot fake reviews?
Suppose that she finds some regularity in reviews that she deems
suspect, so she dismisses them; how long would it take for financially
motivated sellers to produce reviews that avoid the pattern?
Stated more generally, qualitative analysis does not scale, and
trying to peruse all the reviews of more than a few products can often
be unmanageable. Yet, limiting attention to a few potential products
is also unworkable—what would be the selection criteria? If it is
reputation (e.g., only products with 4.5 stars), this runs into exactly
the same issues discussed above. Moreover, the ability to spot fake
reviews—consumer overconfidence notwithstanding—is in fact quite
limited.236 Finally, even if a consumer can find a useful guiding
heuristic, it will be exploitable. If consumers only trust the reviews
of serial reviewers, for example, a seller may derive a sizeable
financial benefit from bribing this serial reviewer.237 If consumers
mostly care about negative reviews, for another example, then a seller
will pay to invest heavily in shilling against competitors’ products.238
If consumers mostly care about the volume of sales, the seller may
artificially inflate sales by giving away products.239 Stated more
generally, heuristics beget loopholes which beget exploitation by
opportunistic sellers.
* * *
To summarize, this Part demonstrated how the microfoundations
of reputation result in informational distortions. Because the reasons
to share information are often private and self-serving, three types of
information distortions emerge—sluggishness, regression to the
235. ANGEL SOFT Toilet Paper Bath Tissue, 48 Double Rolls, 260+ 2-Ply
Sheets Per Roll, AMAZON, https://www.amazon.com/Angel-Soft-Toilet-DoubleTissue/dp/B00FFJ2LXU/ (last visited Dec. 3, 2019); Cottonelle Ultra ComfortCare
Big Roll Toilet Paper, Bath Tissue, 12 Toilet Paper Rolls, AMAZON,
https://www.amazon.com/Cottonelle-Ultra-ComfortCare-Toilet-Tissue/dp
/B01AFRSQGW/ (last visited Dec. 3, 2019).
236. See, e.g., Ponte, supra note 142, at 64–65 (“[I]t is becoming challenging
to decipher more sophisticated forms of fake online reviews.”); Bing Liu, Opinion
Spam Detection: Detecting Fake Reviews and Reviewers, U. ILL. CHI.
https://www.cs.uic.edu/~liub/FBS/fake-reviews.html#reviews (last visited Dec. 3,
2019) (providing an example to test one’s ability in detecting fraud reviews).
237. See, e.g., Jason Murdock, Amazon Sellers Are Bribing Users with Cash
and Gift Vouchers for Five-Star Reviews, Investigation Reveals, NEWSWEEK (July
5, 2019, 12:01 AM), https://www.newsweek.com/amazon-fake-reviews-whichinvestigation-bribes-cash-vouchers-five-star-reviews-1447606.
238. See, e.g., Jacob Shamsian, Beauty Brands Are Reportedly Paying $85,000
to Influencers Who Trash Their Competitors on YouTube, INSIDER (Aug. 30, 2018,
1:11 PM), https://www.insider.com/brands-reportedly-paying-influencers-tocriticize-makeup-competitors-2018-8.
239. See Ponte, supra note 142, at 134.

[p. 48]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1286 WAKE FOREST LAW REVIEW [Vol. 54
extreme, and integrity bias. As a result, peer-to-peer reputational
information tends to provide a false sense of the true quality of the
underlying product. While consumers can try to account for these
distortions, many will lack the requisite sophistication, and even
consumers who do account for the distortions might not be able to do
so given the potentially corrupting effect of the distortions. This is
not to say that distortions are always strong or that consumer
heuristics are not helpful. Still, reputation failures have important
normative implications which I now move to discuss.
IV. LEGAL IMPLICATIONS OF REPUTATION FAILURE
What does the law have to say or do about reputation failures? I
start with the most direct legal interventions that are needed when
the symptoms of reputation failure are present and acute (which is
not always the case). I then move to outline a more ambitious
program: Reputation-by-Regulation. The key idea is to shift attention
from symptoms—such as consumer mistakes—to causes. Legal
institutions can be improved to facilitate the creation of quality
reputational information, thus mitigating some of the root causes of
reputation failure.
A. Reputation Failure and Contemporary Debates in Contracts
and Torts
There are many who call for deregulation on the basis of the rise
of reputational information.240 There has also recently been growing
support among sharing-market enthusiasts, liberals and
conservatives alike, who believe that online platforms open avenues
for effective self-regulation “outside the law.”241
Recognizing reputation failures highlights the dangers of relying
on existing market mechanisms. Naïve reliance on reputation-based
market mechanisms often leads to perverse outcomes in the presence
of acute reputation failures. When consumers select products on the
basis of biased or distorted reputational information, they are likely
to make persistent mistakes242—a social deadweight loss.243 These
mistakes have negative dynamic effects because they make the
production of quality products less rewarding and the production of
unsafe products more rewarding.244 Such a dynamic can devolve into
240. See supra notes 4–6 and accompanying text.
241. See supra note 47 and accompanying text (providing deregulatory
examples among lawmakers and scholars).
242. See supra Subpart III.C.1.
243. See Cannon & Chung, supra note 123, at 39.
244. See Michael Spence, Consumer Misperceptions, Product Failure and
Producer Liability, 44 REV. ECON. STUD. 561, 561 (1977) (“The effect of consumer
misperceptions is that demand votes are miscast, and the supply-side produces
the wrong products.”); see also Oren Bar-Gill, Seduction by Plastic, 98 NW. L. REV.
1373, 1399 (2004) (studying the market effects of consumer mistakes).

[p. 49]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1287
what economists call a “lemon market,” where sellers decide to stop
selling quality goods even though consumers would want to buy them
because consumers cannot distinguish between high and low-quality
goods.245
In the presence of persistent and systematic consumer errors,
recent scholarship has shown that some type of regulation can be
welfare-enhancing even when accounting for the limitations of a topdown regulator.246 If sellers cannot be trusted to meet consumer
expectations, then setting boundaries for permissible dealings may
improve matters.247 Thus, laws and regulations such as lemon laws,
implied warranties, safety audits and recalls, restaurant safety
grading, and many other measures may be needed more than critics
would admit. It also helps draw the boundaries of the sharing
economy and the continued need for traditional reputational sources,
such as professional critics or professional review media (Consumer
Reports, for example).
In sum, reputation failure fits in the family of market failures. It
is a market friction that can justify intervention in consumer markets
in order to improve consumer and social welfare. Of course,
reputation failures do not give a carte blanche for regulation. These
failures vary in scope and severity and in some cases, the costs of
product regulation by an outside regulator may outweigh the
benefits.248 Still, modern debates on deregulation—especially those
involving the sharing economy—fail to recognize that reputation is
subject to systematic failures.249 Bringing this insight into modern
debates should temper some deregulatory trends.
B. Reputation-by-Regulation
Today, the regulatory schema is one of competition between legal
ordering and market ordering “outside the law.”250 Policymakers are
told to choose between heavy-handed regulation and unbridled trust
245. See Akerlof, supra note 28, at 489.
246. See e.g., Oren Bar-Gill, Algorithmic Price Discrimination: When Demand
Is a Function of Both Preferences and (Mis)Perceptions, 86 U. CHI. L. REV. 217,
235–36 (2019); Oren Bar-Gill & Kevin E. Davis, (Mis)perceptions of Law in
Consumer Markets, 19 AM. L. & ECON. REV. 245, 280–81 (2017).
247. About CPSC, U.S. CONSUMER PROD. SAFETY COMM’N,
https://www.cpsc.gov/About-CPSC/ (last visited Dec. 3, 2019) (“CPSC is charged
with protecting the public from unreasonable risks of injury or death associated
with the use of the thousands of types of consumer products under the agency’s
jurisdiction.”).
248. Harold Demsetz, Information and Efficiency: Another Viewpoint, 12 J.L.
ECON. 1, 19 (1969) (describing the “!,” the fallacious assumption that against
market failures there is a perfect regulator).
249. See Stemler, supra note 4, at 687–88.
250. Ronald J. Gilson et al., Braiding: The Interaction of Formal and Informal
Contracting in Theory, Practice, and Doctrine, 110 COLUM. L. REV. 1377, 1379
(2010).

[p. 50]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1288 WAKE FOREST LAW REVIEW [Vol. 54
in the market.251 The analysis presented here suggests a novel third
way, a complementarity model between these two options. The key
insight is that the law has an active role to play ex ante in designing
the rules of the game, such that the information that flows to the
market is more reliable and abundant. I name this family of
strategies Reputation-by-Regulation to indicate how closely related
reputation is to regulation. Rather than an organic and “natural”
outgrowth of market relations, reputation is deeply influenced by
background legal institutions. Drawing awareness to Reputation-byRegulation helps expose the role legal institutions play in the
development of reputational information and highlights alternative
institutional strategies. The rest of this Part suggests a menu of five
options that illustrate how the law can take various degrees of
involvement in removing reputational bottlenecks.
To provide some initial motivation to Reputation-by-Regulation,
it is worth noting that when reputation works it has an important
advantage over standard disclosures rules in that it communicates
with consumers in their own terms, thus avoiding some of the
critiques brought against mandatory disclosure over the last few
years.252 As a peer-to-peer mechanism, consumers directly transmit
the information that they find pertinent using their language and
emphasizing their use patterns. For example, consumer comments
on fuel economy can track normal use patterns more accurately than
the abstract categories of pure city or highway miles required by
law.253 Similarly, annual percentage interest rates that credit issuers
must disclose may be less intelligible to consumers than actual
examples of costs per use.254 As these examples highlight, there is
great potential in Reputation-By-Regulation.
1. Leveraging Market Players: The Role of Reputational
Platforms
It goes without saying that market players will often have an
incentive and ability to deal with market problems themselves. In
the last two decades, for example, reputational platforms specializing
in the aggregation of peer-to-peer reputational information have
blossomed.255 Such platforms have realized the value of “reputation
for reputation,” i.e., the value of garnering consumer trust which can
251. Id. at 1398.
252. See Omri Ben-Shahar & Carl E. Schneider, The Failure of Mandated
Disclosure, 159 U. PA. L. REV. 647, 651 (2011). See generally OMRI BEN-SHAHAR
& CARL E. SCHNIEDER, MORE THAN YOU WANTED TO KNOW: THE FAILURE OF
MANDATED DISCLOSURE 6 (2014) (providing general background on the use and
limitations of mandatory disclosure systems).
253. 16 C.F.R. § 259.4 (2019).
254. 12 C.F.R. § 226.5 (2019).
255. Spencer E. Ante, How Amazon is Turning Consumer Opinions into Gold,
BUS. WEEK, Oct. 26, 2009, at 47, 47.

[p. 51]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1289
then be monetized using various business models.256 Although it may
seem natural now, in the early days of the internet it was far from
evident that a shopping website would want to display information
that could portray some of its traded products in a negative light.257
The sentiment at the time was that “[l]etting consumers rant about
products in public was a recipe for retail suicide.”258 It was also
incredible that consumers could trust the advertised opinions of
complete strangers.259 Still, a small online bookseller by the name of
Amazon took a bold step and adopted a system of consumer
feedback.260 The rest, is, well, history.261
Reputational platforms are metaregulators and, within their
limits, should be enlisted to address some of the problems of
reputation failure. Given that consumer trust is a source of the “new
oil”—internet traffic—platforms have an incentive to develop
metareputation for being honest curators of reputational
information.262 Indeed, some platforms have already taken
voluntarily action to stamp out fake reviews.263 Amazon uses a
variety of algorithms to detect suspect reviews, prohibits the
provision of incentives-for-reviews, and sues violators.264 Moreover,
Amazon also lists some reviews as those done by “Verified
Purchasers” to further limit manipulation, although this has led to a
cottage industry of payments for purchases coupled with fake
reviews.265
256. Id.
257. Id. at 47.
258. Id.
259. See Tadelis, supra note 98.
260. Ante, supra note 255, at 47. For a review of eBay’s history and success,
see Tadelis, supra note 98, at 321.
261. For historical examples of reputation systems, see supra note 5.
262. See Eric Goldman, The Regulation of Reputational Information, in THE
NEXT DIGITAL DECADE: ESSAYS ON THE FUTURE OF THE INTERNET 293, 294–95
(Berin Szoka & Adam Marcus eds., 2010) (reviewing examples of online
reputational platforms).
263. See Communications Decency Act of 1996, 47 U.S.C. § 230 (2012)
(immunizing websites from liability for restricting material that the website
considers to be obscene or otherwise objectionable, “whether or not such material
is constitutionally protected”).
264. See Community Guidelines, AMAZON, https://www.amazon.com/gp/help
/customer/display.html?nodeId=14279631 (last visited Dec. 3, 2019) (limiting
reviews made with a financial motive). Amazon is not the only platform that
protects its reviews. For other examples from other providers see Content
Guidelines, YELP, https://www.yelp.com/guidelines (last visited Dec. 3, 2019)
(prohibiting biased contributions); What Constitutes a First-Hand Traveler
Review? TRIPADVISOR, https://www.tripadvisorsupport.com/hc/en-us/articles
/200614837-What-constitutes-a-first-hand-traveler-review- (last visited Dec. 3,
2019) (limiting reviews in various ways, including “second-hand information”).
265. Various websites offer refunds, sometimes with commission, for reviews.
See list of websites cited supra note 140. A further list of sites is on record with
the author.

[p. 52]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1290 WAKE FOREST LAW REVIEW [Vol. 54
To the extent that such systems work, they are desirable and
helpful. But reputational platforms are also limited in their policing
powers. For the most part, platforms only rely on contractual
agreements between themselves, sellers, and buyers.266 Thus, their
ability to investigate and sanction fake reviews is very limited.
Platforms also risk harmful public relations implications if they take
actions that consumers deem too aggressive.267 Moreover, platforms’
ability to correct consumer misstatements, investigate cherrypicking, or validate information is also limited. There is not much
TripAdvisor can do to enforce its ban of reviews by family members of
an owner’s hotel.268
A deeper problem is that platforms do not always have the
incentive to act in the public interest. Platforms face a conflict of
interest because profits and sales can be in tension with consumer
trust. The existence and type of a conflict depends on the specific
business model, but any platform that profits from the transactions it
facilitates may be tempted to promote higher margin items.269 As a
result, the platform may list these products first, suppress negative
reviews of its own products, or otherwise manipulate the market for
its own advantage.270 Even if gross violations of consumer trust can
be detected, small violations of consumer trust—“‘fudging’ on the
margin”—will be all but impossible to detect by consumers.271 When
Amazon, for example, lists its own products prominently on the first
page and before higher-rated products, it trades off some of the trust
consumers place in it, the trust that it will feature best products first,
against its own profits.272 Uber, facing pressures from drivers,
266. See, e.g., Amazon Services Business Solutions Agreement, AMAZON,
https://sellercentral.amazon.com/gp/help/external/1791?language=en_US&ref=e
fph_1791_cont_G521 (last visited Dec. 3, 2019) (defining sellers’ agreement);
Conditions of Use, AMAZON, https://www.amazon.com/gp/help/customer
/display.html/?nodeId=508088&tag=zxcv123-20 (last visited Dec. 3, 2019)
(defining buyers’ agreement).
267. See David Streitfeld, Giving Mom’s Book Five Stars? Amazon May Cull
Your Review, N.Y. TIMES (Dec. 23, 2012) https://www.nytimes.com/2012/12/23
/technology/amazon-book-reviews-deleted-in-a-purge-aimed-atmanipulation.html.
268. See How Does TripAdvisor Catch Fake Reviews?, TRIPADVISOR,
https://www.tripadvisor.com/TripAdvisorInsights/w3688 (last updated July 13,
2018).
269. See David Adam Friedman, Do We Need Help Using Yelp? Regulating
Advertising on Mediated Reputation Systems, 51 U. MICH. J.L. REFORM 97, 111,
161 (2017) (exploring the conflict of interest).
270. See id. at 122; Julia Angwin & Surya Mattu, Amazon Says It Puts
Customers First. But Its Pricing Algorithm Doesn’t, PROPUBLICA (Sept. 20, 2016,
8:00 AM), https://www.propublica.org/article/amazon-says-it-puts-customersfirst-but-its-pricing-algorithm-doesnt; Julie Creswell, How Amazon Steers
Shoppers to Its Own Products¸ N.Y. TIMES (June 23, 2018),
https://www.nytimes.com/2018/06/23/business/amazon-the-brand-buster.html.
271. See Friedman, supra note 269, at 111.
272. See id. at 130; Angwin & Mattu, supra note 270.

[p. 53]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1291
systematically censors reviews from passengers who give four-star or
less reviews more than a few times.273 A large class action was
brought and settled against Angie’s List in which the primary
allegation was that Angie’s List reviewer ratings were influenced by
payments from providers.274 Yelp has been the subject of extensive
litigation for allegedly manipulating reviews against businesses that
were not willing to pay advertising fees.275 According to an
investigation by the Wall Street Journal, the Federal Trade
Commission (“FTC”) received hundreds of complaints against Yelp
alleging that businesses received unfair reviews after refusing to
advertise on the website.276 Another platform, Consumer Affairs, has
similarly been subject to litigation for allegedly presenting reviews of
certain paying members in a more favorable light.277 A ProPublica
report also suggests that Amazon may be unfairly manipulating
listings in order to promote its own goods.278
Exacerbating the conflict of interest is the right courts granted to
platforms to almost arbitrarily curate reviews. The Ninth Circuit
recently considered whether a platform could arbitrarily choose the
reviews it presents to consumers.279 The court held that the reviewee
has no right to have any review posted at all and as such, cannot
compel the platform to publish reviews it does not want to publish.280
This decision licenses platforms to present reviews according to their
own discretion—with a minimal check on their behavior.281
Additionally, the fight against fake reviews exacerbates the problem
273. David Lumb, Uber Refines Its Rating System to Appease Both Drivers
and Riders, ENGADGET (Nov. 21, 2017), https://www.engadget.com/2017/11/21
/uber-refines-its-rating-system-to-appease-both-drivers-and-rider/.
274. Conditional Amended Class Action Complaint, at 2, 4, 10, Moore v.
Angie’s List, Inc., No. 2:15-cv-01243 (E.D. Pa. July 11, 2016).
275. See Curry v. Yelp Inc., No. 14–cv–03547–JST, 2015 WL 1849037, at *1,
*2 (N.D. Cal. Apr. 21, 2015); Reit v. Yelp!, Inc., 907 N.Y.S.2d 411, 412–13 (N.Y.
Sup. Ct. 2010); Rolfe Winkler, Yelp Says FTC Won’t Act on Complaints About Its
Reviews, WALL ST. J.: DIGITS (Jan. 6, 2015, 4:27 PM), https://blogs.wsj.com/digits
/2015/01/06/yelp-says-ftc-wont-act-on-complaints-about-its-reviews/ (reporting
on a closed investigation by the FTC against Yelp). Note that these cases were
ultimately dismissed; see also Eric Goldman, Court Says Yelp Doesn’t Extort
Businesses, FORBES (Sept. 3, 2014, 12:20 PM), https://www.forbes.com/sites
/ericgoldman/2014/09/03/court-says-yelp-doesnt-extort-businesses
/#c62f7d76e4ab (“For years, Yelp has been dogged by allegations that it
manipulates user reviews.”).
276. See Angus Loten, Yelp Reviews Brew a Fight over Free Speech vs.
Fairness, WALL ST. J. (Apr. 2, 2014, 7:31 PM), https://www.wsj.com/articles/noheadline-available-1396479922.
277. See Consumer Cellular, Inc. v. ConsumerAffairs.com, No. 3:15-CV-1908PK, 2016 WL 3176602, at *1 (D. Or. Feb. 29, 2016).
278. See Angwin & Mattu, supra note 270 (“About three-quarters of the time,
Amazon placed its own products and those of companies that pay for its services
in [a prominently placed] position.”).
279. Levitt v. Yelp! Inc., 765 F.3d 1123, 1126 (9th Cir. 2014).
280. Id. at 1133.
281. Id. at 1133–34.

[p. 54]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1292 WAKE FOREST LAW REVIEW [Vol. 54
because screening of reviews, often done algorithmically, relies on
necessarily opaque standards since disclosing the algorithm would
invite abuse by sellers.282 At the same time, these opaque algorithms
give the platform more power to abuse consumer trust.283
To the extent regulators are worried about these issues, a few
options to leverage market players present themselves: regulation,
investigation, and accreditation.
a. Regulating Platforms
On the regulatory side, consumer agencies and legislators can
create a unified set of rules that governs what constitutes fair and
reasonable treatment of consumer peer-to-peer reputational
information.284 A platform should not promote its own products, or
higher-margin products, when it presents consumer-sourced
reputational information. A clear first step in this direction is to
revise the holding that platforms are free to arbitrarily curate
reviews.285 Additionally, platforms should be considered as forum
providers for speech, limiting their ability to arbitrarily censor
reviews. Platforms should also be required to publish their review
curation, aggregation, and display standards. Another possibility is
to require platforms to release certain key statistical information
such as the volume of sales. Additionally, platforms might be
required to display the ratio of consumers who did not rate the
product to those who did in order to aid consumers (or assistive
technology) in drawing better statistical inferences. Alternatively,
they might simply be required to create and follow their own choice
of standards of regulation—thus creating metaregulation of sorts,
relying on the platforms to find the right balance of substantive rules.
A useful source of inspiration (although with some caution) is the
evolving international standards of platform regulation developed by
the International Consumer Protection Enforcement Network
(“ICPEN”), a network of consumer protection authorities from nearly
sixty countries.286 The ICPEN standards include disclosure
282. See David Adam Friedman, Addressing the Commercialization of
Business Reputation, 80 LAW & CONTEMP. PROBS. 73, 83 (2017).
283. See id. at 79 (relating the conflict of interest to the business model of the
platform).
284. See Strahilevitz, supra note 1, at 64, 69, 71 (arguing that the government
should subsidize and encourage transparency among reputational platforms);
Van Loo, supra note 48, at 585–99 (developing an account of the regulation of
platforms by agencies, highlighting the need for regulation in certain key areas).
285. See Levitt, 765 F.3d at 1126, 1134.
286. INT’L CONSUMER PROT. & ENF’T NETWORK, ONLINE REVIEWS &
ENDORSEMENTS: ICPEN GUIDELINES FOR REVIEW ADMINISTRATORS 7 (2016)
https://icpen.org/sites/default/files/2017-06/ICPEN-OREGuidelines%20for%20Review%20Administrators-JUN2016.pdf (guiding
reputational platforms to be “equal and fair in the collection of reviews,” be “alert
and proactive in the moderation of reviews,” and “transparent in publication of
reviews”).

[p. 55]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1293
standards governing platforms’ methods for curating and aggregating
reviews, requirements for platform functionality of review sorting
according to consumer criteria, and obligations to present negative
reviews of a platforms’ own products.287 It should be noted, however,
that some of these regulations can run into potential First
Amendment constraints.288 At the same time, their usefulness and
importance merits serious consideration.
b. Policing Platforms
To ensure the integrity of reputational information, it may be
necessary for an external agency to regularly inspect and police the
inner workings of reputational platforms.289 This is already done, to
some extent, by the FTC.290 However, there are still some important
informational gaps.291 The agency should first have access to all
reviews posted to the platform, including their timestamps, IP
addresses, and external information relating to the product price and
type. Then, it should access the (anonymized) reviewer data itself—
past transactions and past review history—in order to identify shell
accounts. Then, the agency should trace typical consumer searches
and the corresponding results: what products are featured, in which
order, and by which criteria. Finally, the agency should review the
platform’s algorithms for identifying faux reviews and test their
operation. Such investigations are crucial given that many of the
processes for collecting and curating reputational information are
opaque and that opaqueness may be necessary to avoid manipulation
by other market players. In addition to audits, the agency should also
investigate claims of unfair treatment by market players.
c. Platform Accreditation
Accreditation is perhaps the least intensive form of regulatory
intervention. Accreditation will involve using a badge to indicate that
the platform is monitored by the agency and that it complies with its
own standards. Receiving accreditation may be entirely voluntary,
287. Id. at 7–11.
288. See Zauderer v. Office of Disciplinary Counsel of Supreme Court of Ohio,
471 U.S. 626, 650–51 (1985) (permitting disclosure requirements for attorneys);
Milavetz, Gallop & Milavetz, P.A. v. United States, 559 U.S. 229, 252–53 (2010)
(same). But cf. In re R.M.J., 455 U.S. 191, 203 (1982) (holding that misleading
commercial speech is not protected); Cent. Hudson Gas & Elec. Corp. v. Pub. Serv.
Comm’n of N.Y., 447 U.S. 557, 563–64 (1980) (applying intermediate scrutiny).
289. See Van Loo, supra note 48, at 585–98 (suggesting the need for external
policing).
290. See Brogaard, supra note 221.
291. On existing agency powers to conduct audits, see Yonathan A. Arbel,
Adminization: Gatekeeping Consumer Contracts, 71 VAND. L. REV. 121, 144–46
(2018) (exploring the use of agencies to audit market players for consumer abuse).

[p. 56]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1294 WAKE FOREST LAW REVIEW [Vol. 54
thus sidestepping any potential First Amendment concerns.292 Of
course, if the agency finds at any time that the platform is not in
compliance, it may strip its badge.
Receiving a badge would garner consumer trust. Such a system
can thus be valuable even if it is entirely voluntary because it will be
in the platform’s interest to receive accreditation. Remember that
consumers may be suspicious even when platforms act honestly
because the platform has superior information regarding its own
internal practices. The badge would allow platforms to credibly
communicate their honesty to the public.
2. Professional Publications
An alternative or supplement to a system of accreditation
involves using professional rating agencies and publications. Some
successful examples include Consumer Reports, US News, PC
Magazine, Michelin Restaurant Review, and the New York Times
Book Review section.293 Like reputation platforms, these services are
also premised on the idea of a reputation for reputation, i.e.,
monetizing consumer trust in their reputation production services.294
Moreover, they have some advantages over amateur consumer
reviews in that they have both the facilities and knowledge to
extensively test products.295 Such publications are demonstrably
valuable, as consumers continue to use them despite their cost and
the rise of free online consumer-generated information.296 Indeed,
292. The use of a government agency, rather than a market player, is
motivated by the infinite regress problem noted by Brian Galle, whereby
consumers might worry that the contracted auditor is itself compromised. See
Brian D. Galle, Self-Regulation of Social Enterprise, in RESEARCH HANDBOOK ON
SOCIAL ENTERPRISE LAW 7–8 (forthcoming); see also Oren Perez et al., The
Dynamic of Corporate Self-Regulation: ISO 14001, Environmental Commitment,
and Organizational Citizenship Behavior, 43 LAW & SOC’Y REV. 593, 593–94
(2009) (studying self-regulation under voluntary accreditation).
293. See, e.g., Book Review, N.Y. TIMES, http://www.nytimes.com/section
/books/review (last visited Dec. 3, 2019); Electronics: Ratings & Buying Guides,
CONSUMER REP., http://www.consumerreports.org/cro/index.htm (last visited
Aug. 27, 2019); The MICHELIN Star Restaurant Rating System, MICHELIN,
http://guide.michelin.com/th/en/to-the-stars-and-beyond-th (last visited Aug. 27,
2019).
294. Paul Resnick et al., Reputation Systems, 43 COMM. ACM 45, 47–48
(2000).
295. See, e.g., Jeff S. Bartlett & Gabe Shenhar, How Consumer Reports Tests
Cars, CONSUMER REP., https://www.consumerreports.org/cars-how-consumerreports-tests-cars/ (last visited Dec. 3, 2019) (noting that Consumer Reports has
a 327-acre test center where it test drives cars for hundreds of thousands of
miles).
296. Consumer Reports, for example, charges thirty-nine dollars per year for
a digital membership. See Buying Smart is Just the Start, CONSUMER REP.,
http://www.consumerreports.com/join?INTKEY=1810GH0MB (last visited Dec.
3, 2019). In recent years, various professional and semiprofessional critics
started using platforms such as YouTube to broadcast reviews. See, e.g., 10
YouTube Film Critics You Need to Be Watching, WHATCULTURE,

[p. 57]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1295
their continued existence is a possible testament to the existence of
reputation failure in peer-to-peer reputational information.297
Still, reliance on such publications is not without its limitations.
Professional critics do not always care about the same things as lesssophisticated consumers.298 Professional publications can only cover
a sliver of the product-space, and it is unlikely that they can ever
approach the comprehensiveness of consumer-sourced reviews.299
But, most acutely, as consumers place more confidence in such
publications, it becomes more profitable for sellers to bribe those
reviewers to publish favorable reviews.300
The government may increase the use of such services by either
subsidizing them or otherwise requiring testing in some areas.
Notably, the state already supports these publications by protecting
their copyright and intellectual property. Such protections can be
extended through broader, more aggressive copyright protections,
subject—of course—to a full cost-benefit analysis.
3. Fighting Fake Reviews
As noted earlier, fake reviews are the scourge of the reputation
system. The more consumers use and trust reviews, the more it pays
to invest in creating fake reviews.301 And while reputation platforms
have some incentive to fight fake reviews, their efforts tend to fall
short.302
A useful market-based solution here is the leveraging of a
competitor’s interest. Importantly, I propose that fake reviews will
http://whatculture.com/film/10-youtube-film-critics-you-need-to-be-watching
(last visited Dec. 3, 2019). These critics monetize user engagement through ads
or, one worries, side payments from sellers. See Jeff Rose, How Much Do
YouTubers Really Make?, FORBES (Mar. 21, 2019), https://www.forbes.com/sites
/jrose/2019/03/21/how-much-do-youtubers-really-make/#8b3cec37d2b2.
297. While professional reviews continue to exist, consumers rely more often
on peer-to-peer reputational information. See Mehdi Ghazisaeedi et al.,
Trustworthiness of Product Review Blogs: A Source Trustworthiness Scale
Validation, 6 AFR. J. BUS. MGMT. 7498, 7498 (2012).
298. See WEBER SHANDWICK & KRC RES., BUY IT, TRY IT, RATE IT: STUDY OF
CONSUMER ELECTRONICS PURCHASE DECISIONS IN THE ENGAGEMENT ERA 6, 8
https://www.webershandwick.com/uploads/news/files/ReviewsSurveyReportFIN
AL.pdf.
299. Compare Woolf, supra note 158 (examining over 1.2 million consumer
product reviews on Amazon), with All Products A-Z, CONSUMER REP.
https://www.consumerreports.org/cro/index.htm (last visited Dec. 3, 2019)
(noting that the service has only reviewed just over 9,000 products).
300. See Roomy Khan, From Fake Reviews to Unvetted Sellers: Here’s Why
Amazon Marketplace Needs More Oversight, FORBES (Apr. 1, 2019),
https://www.forbes.com/sites/roomykhan/2019/04/01/amazon-marketplace-achaotic-bazaar-unvetted-sellers-to-fake-reviews-where-is-the-oversight/.
301. See supra notes 145–47.
302. See David Adam Friedman, supra note 269, at 142 (questioning the
“effectiveness of these internal initiatives to discourage and eliminate false
reviews”); Stemler, supra note 4, at 707–10.

[p. 58]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1296 WAKE FOREST LAW REVIEW [Vol. 54
be considered a form of false advertising and subject to the Lanham
Act of 1940 or state-level antitrust laws, which permit competitors to
bring private suits against sellers for posting fake reviews.303 Some
courts have authorized the use of this provision to impose liability for
fake reviews.304 This tool is helpful—competitors may lose market
share when a firm fakes its own reviews or use reviews to attack
another—but it is also quite limited. Only competitors may employ
this tool, and they themselves have limited resources to investigate
claims. More fundamentally, the cost of fighting false advertising by
a competitor is private, but the benefit accrues to all the firms that
compete in the space.
In contrast to private market players, the FTC, the Consumer
Financial Protection Bureau, and some state agencies have broad
investigative powers.305 These agencies can investigate fraud and
have both the authority and resources to do so effectively.306 They
can also use their powers to fine market players for unlawful
behavior, creating the strong deterrence needed to effectively combat
the generation of fake reviews.307 Importantly, fake reviews can be
considered a form of false advertising and subject to the Lanham Act
or state-level antitrust laws,308 although the exact mechanisms are
beyond the scope of this Article.309 Reputation failure provides strong
reasons for further investment in resources in these measures.
One issue in combating fake reviews is the First Amendment
protection of speech. Historically, the First Amendment was not
thought to cover fraudulent speech.310 In In re R.M.J.,311 the Supreme
Court held that states are free to regulate advertising that is
inherently misleading.312 And while the recent decision in United
States v. Alvarez313 allowed some protection of fraudulent speech in
the context of the Stolen Valor Act, such protection is very limited.314
303. 15 U.S.C. § 1125(a) (2012); see, e.g., CAL. BUS. & PROF. CODE § 17200
(West 2019).
304. Romeo & Juliette Laser Hair Removal, Inc. v. Assara I LLC, No.
08cv0442(DLC), 2016 WL 815205, at *21–23 (S.D.N.Y. Feb. 29, 2016).
305. Dodd-Frank Wall Street Reform and Consumer Protection Act § 1052, 12
U.S.C. § 5562 (2012); 15 U.S.C. § 46 (2012).
306. 12 U.S.C. § 5562; 15 U.S.C. § 46.
307. See also Arbel, supra note 291, at 171–72.
308. See 15 U.S.C. § 1125(a); CAL. BUS. & PROF. CODE § 17200; Assara I LLC,
2016 WL 815205, at *21–23.
309. On the regulation of platforms, see, e.g., Max N. Helveston, Regulating
Digital Markets, 13 N.Y.U. J.L. & BUS. 33, 83–84 (2016).
310. See Va. Bd. of Pharmacy v. Va. Citizens Consumer Council, Inc., 425 U.S.
748, 771 (1976).
311. 455 U.S. 191 (1982).
312. Id. at 207.
313. 567 U.S. 709 (2012).
314. Id. at 719, 730 (allowing regulation of fraudulent speech); see also
Donaldson v. Read Magazine, Inc., 333 U.S. 178, 190 (1948) (holding that the
government’s power “to protect people against fraud” has “always been
recognized in this country and is firmly established”).

[p. 59]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1297
Fake reviews are by their nature misleading, thus it seems that welltailored regulations meant to apply this standard would be justified.
In addition, the law should limit businesses’ ability to offer incentives
for favorable reviews, i.e., cherry-picking consumers. Such an
approach could sidestep many of the thorny constitutional tensions
while advancing the goal of combating reputation failure.
To be clear, it is not expected that regulatory action alone will be
capable of eliminating fake reviews. Still, decisive regulatory action
can significantly curtail the profitability of this practice. It should
also be noted that investing in some of the other measures proposed
here would also be helpful in fighting fake reviews. It is much easier
and cheaper to cultivate a favorable view of one’s restaurant when
competing with a handful of reviews; it is much more complex to do
so when there are dozens of reviews.
4. Fostering Positive Incentives
As argued earlier, because reputation is a public good, consumers
often lack sufficient incentive to create it—a problem most acute with
respect to middling experiences and unpopular opinions. The nascent
law regulating consumer benefits exhibits considerable confusion
about this basic point and takes an overly strong stance against
incentivizing reviews.315 Here, again, the microfoundations
framework helps delineate the proper limits of providing incentives
and behavioral nudges to consumers.
To promote transparency in the market and curb false
advertising, the FTC announced new guidelines in 2015 that regulate
incentivized reviews.316 The context of these guidelines is facially
reasonable: the ascendency of social media has created a new form of
endorsement—reviews by “influencers,” or individuals who amass
many followers. Companies are estimated to be spending billions of
dollars paying influencers to endorse products on their social media
accounts.317 In response, the FTC sought to require influencers to
disclose their financial interests.318
The result, however, is the proverbial throwing out the baby with
the bathwater. Consumers need incentives and nudges to produce
accurate reputational information, which are outcomes of consumers
not internalizing the benefits of reputational information.319 Direct
incentives consist of free products, discounts, payments, and
commissions. Nudges, such as prompts to rate the previous
315. 16 C.F.R. § 255 (2019).
316. Id.
317. See Suzanne Kapner and Sharon Terlep, Online Influencers Tell You
What to Buy, Advertisers Wonder Who’s Listening, WALL ST. J. (Oct. 20, 2019, 8:59
PM) https://www.wsj.com/articles/online-influencers-tell-you-what-to-buyadvertisers-wonder-whos-listening-11571594003.
318. 16 C.F.R. § 255.5 (2019).
319. See supra Subpart III.B.

[p. 60]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1298 WAKE FOREST LAW REVIEW [Vol. 54
experience before engaging in a new transaction or reminders to rate
and review, also increase the creation of consumer reputational
information.320
Despite the importance of such incentives and nudges, the FTC
guidelines impose onerous disclosure requirements that are triggered
almost indiscriminately without attention to context. For example, if
a restaurant chooses to offer free meals on its opening night so as to
incentivize traffic, every person dining there has to disclose her
financial stakes when discussing her experience—even if the
restaurant never asked for any review, much less a favorable one.321
The same goes for a “dollar-off” coupon, sweepstakes promotions, or
even charity donations.322 The imposition of such broad duties is not
only onerous but it also has unwanted secondary effects. Like the
harried student highlighting the entire textbook, there is danger in
indiscriminate disclosure. Using the same disclosure standards for
content-neutral and content-biased reviews can be misleading. Even
worse, mandating such disclosures may exacerbate the problem of
regression to the extreme. Research finds that disclosing financial
incentives may create a “moral license” to exaggeratedly extol the
virtues of the product.323
There is a readily available alternative. The developing
international standard permits the use of content-neutral
incentives.324 Under this standard, businesses may legitimately offer
incentives to reputation creators if it ensures that the resulting
opinion arises independently of the incentive.325 A content-neutral
incentive may include offering free products under an agreement that
clearly states that the user has full discretion over the content of the
review and that future promotions will not be made contingent on her
response.326 Or businesses can provide discounts to consumers who
review a product, so long as the review is anonymized by a trusted
third-party.327 Research on the effect of such incentives is scant, but
it suggests that content-neutral reviews are effective. A recent study
compared incentivized reviews to organic ones, both qualitatively and
320. Stemler, supra note 4, at 684–85 (discussing creation of reputational
information by encouraging or requiring users to leave feedback after a
transaction is complete).
321. FTC, THE FTC’S ENDORSEMENT GUIDES: WHAT PEOPLE ARE ASKING 5
(2017).
322. Id. at 4.
323. See Loewenstein et al., supra note 213, at 424–25.
324. See Int’l Consumer Prot. & Enf’t Network, Online Reviews &
Endorsements: ICPEN Guidelines for Review Administrators, 8 (June 2016).
325. See id.
326. See id. (stating that financial or material benefits should be given by
review administrators to all types of reviews); see, e.g., FTC, supra note 321.
327. See Int’l Consumer Prot. & Enf’t Network, supra note 324, at 8; see, e.g.,
Maria Petrescu et al., Incentivized Reviews: Promising the Moon for a Few Stars,
41 J. RETAILING AND CONSUMER SERVS. 288, 292 (2018) (stating reviewers are
given incentives such as discounted products from third party companies).

[p. 61]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1299
quantitatively.328 Not surprisingly, incentivized reviews put less
emphasis on price; but importantly, there was no difference in rating
between the content-neutral-incentivized and organic reviews.329
Thus, incentives can have desirable effects for creating reliable
reputational information.
5. Controlling Costs: First Amendment and Reputation
The last set of solutions builds on the key insight that the costs
of reputation generation are wholly private, but the benefits are
partly public. This Subpart advocates the expansion of free speech
safeguards provided by the First Amendment to consumer reviews.
Today, with increasing frequency, lawsuits are brought against
consumers for providing reviews.330 Businesses latch onto factual
inaccuracies (some small or innocent) and sue using a variety of
doctrines including defamation, tortious interference, injurious
falsehoods (commercial disparagement),331 and false light.332 One
report finds that “negative reviews have become the subject of dozens
of lawsuits across Texas in recent years,” and there is a growing sense
that this happens across the nation.333 The consequences of such
lawsuits can be dire: a woman complaining online about the services
of her divorce attorney was charged with $350,000 in damages.334
Admittedly, such judgments are relatively exceptional. However, the
threat is not just liability but also litigation. Indeed, according to
Professor Lyrissa Lydsky, a primary reason such lawsuits are
brought is not to collect damages but to silence.335 And this menacing
effect is amplified by consistent media coverage of such lawsuits.336
328. Petrescu et al., supra note 327, at 291, 293.
329. See id. at 294 (finding that providing incentives does not affect the
“satisfaction ratings assigned to the product in the form of ‘stars’ from one to five”
although they do find some evidence of “potential linguistic and sentiment
differences found in the qualitative analysis”).
330. See, e.g., Brittany Glas, Think Twice Before You Post a Negative Review
Online, KXAN (Feb. 14, 2018), https://kxan.com/news/local/austin/think-twicebefore-you-post-a-negative-review-online/.
331. See LOUIS ALTMAN & MALLA POLLACK, 3 CALLMANN ON UNFAIR
COMPETITION, TRADEMARKS. & MONOPOLIES, §11:13 (4th ed., 2019). States differ
considerably; see also Gillon v. Bernstein, 218 F. Supp. 3d 285, 301 (D.N.J. 2016)
(considering whether an unfavorable review amounts to product disparagement).
332. RESTATEMENT (SECOND) OF TORTS § 623A, § 623A cmt. g, § 652E, § 652E
cmt. b (AM. LAW INST. 1977).
333. Glas, supra note 330.
334. Blake v. Giustibelli, 182 So. 3d 881, 884 (Fla. Dist. Ct. App. 2016); see
also Samson Habte, Court Affirms $350k Verdict for Lawyer Smeared on Avvo,
BLOOMBERG L. (Jan. 27, 2016), https://www.bloomberglaw.com.
335. See Lyrissa Barnett Lidsky, Silencing John Doe: Defamation & Discourse
in Cyberspace, 49 DUKE L.J. 855, 858–60 (2000) [hereinafter Lidsky, Silencing
John Doe]; see also Lyrissa Barnett Lidsky, Anonymity in Cyberspace: What Can
We Learn from John Doe?, 564 B.C. L. REV. 1373, 1374 (2009).
336. See, e.g., Beth Landman & Julia Marsh, I Wrote a Negative Yelp Review
– and It Made My Life a Nightmare, N.Y. POST (May 28, 2018),

[p. 62]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1300 WAKE FOREST LAW REVIEW [Vol. 54
To compound the matter further, in handling these lawsuits,
consumers face the common risk of de-anonymization.337 Nor is state
legislation very helpful. Anti-Strategic Lawsuits Against Public
Participation (“Anti-SLAPP”) legislation meant to combat abusive
lawsuits is not broadly adopted or consistently applied.338 Finally,
even slight increases in cost can dissuade reviewers (reduce the
number of reviews). One experiment, for example, tested how small
costs affect behavior and found that “[r]emoving a cost of only
$0.25 . . . leads to a more than 50 percentage point increase in the
frequency of rating.”339
Some examples might be helpful in appreciating the chilling
effect of litigation risk. One New Jersey consumer, Jane Perez,
complained online about her contractor, stating, “My home was
damaged: the ‘work’ had to be re-accomplished . . . he invoiced me for
work not even performed.”340 The contractor sued Perez for $750,000
in damages for defamation.341 The contractor finally lost the suit, but
along the way, Ms. Perez deleted her review and had to defend herself
through an expensive five-day jury trial.342 Or, take the case of Las
Vegas consumer Pamela Boling.343 She sought the assistance of a tax
professional to help demonstrate her economic hardships to tax
authorities. However, the service she received was below her
expectations, so she turned to Yelp and wrote a review concluding
“this is MALPRACTICE!”344 Soon after, the business filed a
defamation lawsuit against her. To ward off the lawsuit, she spent
$40,852 in litigation costs.345 Although she ultimately won the case
https://nypost.com/2018/05/28/i-wrote-a-negative-yelp-review-and-it-made-mylife-a-nightmare/.
337. See, e.g., Yelp, Inc. v. Hadeed Carpet Cleaning, Inc., 770 S.E.2d 440, 441
(Va. 2015) (ruling, by the Supreme Court of Virginia, that authority exists for
disclosing the identity of the consumers); see also Lori A. Roberts, Brawling with
the Consumer Review Site Bully, 84 U. CIN. L. REV. 633, 653–56 (2016) (reviewing
the procedures involved in de-anonymizing consumers). Every month, Yelp
receives about six subpoenas to reveal the identity of consumers and many more
requests are filed with the courts. See Loten, supra note 276.
338. For a review of state legislation, see State Anti-SLAPP Laws, PUB.
PARTICIPATION PROJECT, https://anti-slapp.org/your-states-free-speech-protection
(last visited Dec. 3, 2019); see also Aaron Smith, Note, SLAPP Fight, 68 ALA. L.
REV. 303, 305 (2016) (surveying anti-SLAPP legislation and exploring the
uncertainty surrounding their applicability in federal courts).
339. Lafky, supra note 106, at 561.
340. Complaint at 3–4, Dietz Dev., LLC v. Perez, 2012 Va. Cir. LEXIS 139
(Va. Cir. Ct. Dec. 7, 2012) (No.2012-16249).
341. Id. at 9–10.
342. See Paul A. Levy, Ruminations About Dietz v. Perez, PUB. CITIZEN (Mar.
28, 2014, 6:22 PM), http://pubcit.typepad.com/clpblog/2014/03/ruminationsabout-dietz-v-perez.html.
343. Defendant Pamela Boling’s Motion for Costs and Attorneys’ Fees Under
NRS 41.670 at 1, IQTAXX, LLC v. Pamela Boling, 2016 WL 4924268 (Nev. D. Ct.
May 12, 2016) (No. 15-A-728426-C).
344. Id. at 4.
345. Id. at 15.

[p. 63]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1301
and recovered some costs, the process was long, risky, and caused her
to redact her original opinion—suppressing a view that the court
deemed legitimate.346 Finally, consider the case of Stephen Glover,
who posted a negative review about his lawyer, claiming that he was
the “[w]orst ever” because he yelled at him to “GOOGLE IT!” in
response to a question and otherwise acted unprofessionally.347 This
review led to a two-year defamation lawsuit and appeal where Glover,
finally, prevailed.
The chilling effect of lawsuits is related to the lack of legal
safeguards to protect consumer speech.348 Under prevailing
standards, businesses can bring a defamation lawsuit against
consumers if the review contains some factual inaccuracies.349 From
a First Amendment perspective, some courts have been willing to
accept that reviews are a matter of public interest and therefore
should be protected under the First Amendment, but the scope of
protection is slim. In a recent case, the Oregon Supreme Court
explained that reviews are protected only if “a reasonable factfinder
could not conclude that [the consumer’s] review implies an assertion
of fact.”350 In effect, the decision underscores the costs borne by
consumers who pen reviews. Beyond the possibility of an
anticonsumer mistake by judges or juries,351 it is simply difficult for
most consumers—especially those who are emotional—to write
reviews that clearly communicate an opinion or avoid any factual
inaccuracies given inevitable gaps in recollection, errors in phrasing,
or strong emotions.352
If we recognize the social importance of consumer reviews, the
weak positive incentives to produce them, and the risk of liability or
even just litigation, a few solutions present themselves. One
moderate solution is greater adoption and broader implementation of
anti-SLAPP legislation.353 This legislation is useful because it
imposes costs on strategic lawsuits by businesses. But it is also
limited. To win such a suit, the consumer has to prove that the
346. See id.
347. Spencer v. Glover, 397 P.3d 780, 783 (Utah Ct. App. 2017).
348. See DAN B. DOBBS ET AL., THE LAW OF TORTS, 159–61 (2d ed. 2011 & Supp.
2016); W. PAGE KEETON ET AL., PROSSER AND KEETON ON THE LAW OF TORTS 771–
72, 839–40 (5th ed. 1984).
349. DOBBS ET AL., supra note 348, at 167–71; MCNAMARA, supra note 63, at
2; see Sim v. Stretch [1936] 52 Times L. Rep. 669, 669–71 (UK).
350. Neumann v. Liles, 369 P.3d 1117, 1126 (Ore. 2016) (interpreting
Milkovich v. Lorain Journal Co., 497 U.S. 1 (1990)).
351. Nuno Garoupa, Dishonesty and Libel Law: The Economics of the
“Chilling” Effect, 155 J. INSTITUTIONAL THEORETICAL ECON. 284, 285, 289 (1999).
352. Lidsky, Silencing John Doe supra note 335, at 865 (noting the current
protections are “inadequate”).
353. See Smith, supra note 338, at 305. The “SLAPP” in anti-SLAPP stands
for “strategic lawsuit against public participation.” Dan Frosch, Venting Online,
Consumers Can Find Themselves in Court, N.Y. TIMES (May 31, 2010)
https://www.nytimes.com/2010/06/01/us/01slapp.html?module=inline.

[p. 64]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1302 WAKE FOREST LAW REVIEW [Vol. 54
business does not stand a good chance of prevailing, which—given the
current legal standards and the limited resources consumers have—
is tough.354 Other options also include the use of legal aid subsidies
or crowdfunding to defend consumers, arbitration, and other forms of
alternative dispute resolution, and agency audits.355
The most powerful solution would be a First Amendment
protection in the form of a consumer review privilege.356 Today,
political speech enjoys broad protections under the New York Times
v. Sullivan357 standard.358 Despite the recognition that protecting
political speech could foster false allegations, the Supreme Court
expressed a strong preference for the promotion of speech on matters
involving public figures.359 As a result, the Court ruled that unless
plaintiffs can show malice on the defendant’s side, a lawsuit cannot
be brought.360 As a result, such lawsuits are relatively rare. Future
cases, most notably Dun & Bradstreet, Inc. v. Greenmoss Builders,361
emphasized that issues of public interest are also deserving of greater
protection.362 This is explained on the basis of the positive externality
of speech, a feature that consumer reviews also share.363
A consumer review privilege would still permit businesses to
bring lawsuits against false reviews, but they will have to be able to
show malice on the consumer side. Such a privilege will greatly
reduce the business ability to strategically drag consumers to court.364
The privilege would also protect consumers in the lawsuit itself,
although given the high win rate consumers enjoy today, this effect is
admittedly small. Additionally, this privilege will considerably limit
354. See id. at 305, 316–18, 325.
355. See Arbel, supra note 291, at 158; Ronen Perry, Crowdfunding Civil
Justice, 59 B.C. L. REV. 1357, 1359–60 (2018) (describing the use of crowdfunding
to subsidize litigation).
356. See Lyrissa Barnett Lidsky & RonNell Andersen Jones, Of Reasonable
Readers and Unreasonable Speakers: Libel Law in a Networked World, 23 VA. J.
SOC. POL’Y & L. 155, 157–59 (2016) (exploring how policymakers can amend the
rules on expression of opinion and the malice requirements to control speech).
357. 376 U.S. 254 (1964).
358. See Gertz v. Robert Welch, Inc., 418 U.S. 323, 334–37, 340 (1974);
Sullivan, 376 U.S. at 269–70; see also Anthony Lewis, New York Times v.
Sullivan Reconsidered: Time to Return to “The Central Meaning of the First
Amendment”, 83 COLUM. L. REV. 603, 604 (1983) (noting that the Supreme Court
had gone “for 170 years without finding in the first amendment any limits on libel
judgments”). But cf. Milkovich v. Lorain Journal Co., 497 U.S. 1, 18–19 (1990)
(holding that there is no “wholesale defamation exemption for anything that
might be labeled ‘opinion’”).
359. See Sullivan, 376 U.S. at 269–70.
360. See id. at 279–80.
361. 472 U.S. 749 (1985).
362. See id. at 758–59 (1985); see also Snyder v. Phelps, 562 U.S. 443, 453–54
(2011) (testing what counts as public interest).
363. See POSNER, supra note 47, at 297–98.
364. For an early expression of this sentiment, see THOMAS STARKIE, A
TREATISE ON THE LAW OF SLANDER, LIBEL, SCANDALUM MAGNATUM, AND FALSE
RUMOURS xx–xxi (New York, J. & J. Harper 1826).

[p. 65]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
2019] LIMITS OF MARKET DISCIPLINE 1303
the ability of businesses to de-anonymize consumers. As a result, this
privilege would significantly reduce the cost of legal liability making
speech more attractive on the margin. Indeed, existing privileges are
also justified by the positive externalities of speech, so this privilege
would be a natural extension.
Some have objected to protecting consumers’ reviews on the
ground that it encourages reckless or deliberate lies by consumers
against businesses.365 However, objections of this sort have not
sufficiently accounted for the public value of reputation or its
microfoundations.366 They assume that consumers share a desire to
besmirch the reputation of firms but say little about why (or when)
consumers care to tell the truth in the first place. Moreover, they
have not analyzed the dynamic equilibrium that emerges from a lax
defamation regime. In short, people tend to place less trust in
assertions that are made in the absence of defamation law and so the
negative impact of lies would be much abated.367 Thus, the opposition
to consumer review privileges should be revisited. At the very least,
scholars and policymakers should adopt a more skeptical approach to
the social utility of defamation laws, and courts should better
understand the chilling effect of their rulings, even when the case is
finally disposed of in favor the consumer.
V. CONCLUSION
Reputation is fundamental to the operation of many markets.
When reputation works, it works extremely well; it disciplines sellers
at a low cost, saving the need for courts and lawyers. But reputation
can also fail. Today, many are too excited by the rise of the sharing
economy to see that the microfoundations on which it rests are
faltering.
Earlier scholarship—in law, economics, sociology, and biology—
has trusted reputation to work well, at least in certain domains. This
Article explained why careful analysis of the microfoundations of
reputation—the microincentives that lead individuals to create and
share reputational information—suggests the potential of reputation
failures. Such failures have a significant bearing on future
policymaking and contracts scholarship in particular. Most directly,
it invites greater skepticism towards current trends to deregulate
consumer transactions on the basis of faith in the internal regulatory
power of market forces.
365. See Dohse, supra note 137, at 390–91.
366. See Heymann, supra note 63, at 1417–23 (arguing that the public
interest dimension of reputation has been neglected).
367. See Yonathan A. Arbel & Murat Mungan, The Case Against Strong
Defamation Laws 6–7 (Univ. of Ala. Legal Studies Research Paper No. 3311527),
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3311527.

[p. 66]
W03_ARBEL_GRAPHICS_REVISED.DOCX (DO NOT DELETE) 1/30/20 11:10 AM
1304 WAKE FOREST LAW REVIEW [Vol. 54
The most ambitious goal of this Article is to carve a path for
future regulation of consumer markets—Reputation-by-Regulation,
i.e., the use of laws and institutions to improve the flow of
reputational information to the market. This approach holds
considerable promise. Like mandatory disclosures, the law of
reputation seeks to improve markets indirectly by providing
consumers with reliable information that would allow them to make
informed purchasing decisions.368 By identifying and removing
reputational failures, the law can increase consumer welfare without
mandating any specific set of terms, thus preserving autonomy and
freedom of contract.369 Addressing reputation failure should be the
cornerstone of future consumer policy.
368. See, e.g., John C. Coffee, Jr., Market Failure and the Economic Case for
a Mandatory Disclosure System, 70 VA. L. REV. 717, 728–29 (1984); Amy J.
Schmitz, Remedy Realities in Business-to-Consumer Contracting, 58 ARIZ. L. REV.
213, 217, 219 (2016) (“Classical contract doctrine prefers formulistic disclosure
rules . . . . [D]isclosure bolsters freedom of contract by giving consumers an
opportunity to review contract terms before consenting.”).
369. See Andrew T. Bond, Essay, An App for That: Local Governments and the
Rise of the Sharing Economy, 90 NOTRE DAME L. REV. ONLINE 77, 95–96 (2015)
(arguing that reputational incentives allow deregulation).