Court filing
Exhibit A-1 to Declaration of Susan Fahringer - Redacted -… — Brooks v. Thomson Reuters Corporation (Dkt. 186.3)
Summary
Exhibit A-1 to Document 186-3, filed 03/28/23 in Case No. 3:21-cv-01418-EMC, is the redacted, publicly filed version of the Expert Report of Prof. Jane Bambauer in the Brooks and Shabazz case against Thomson Reuters Corporation, filed pursuant to a Court order dated March 21, 2023. The table of contents shows qualifications, scope, a summary of opinions, and a section titled Privacy Does Not Require Absolute Control that responds to Prof. Turow's suggestion. A further section is titled The Limited Set of Permissible Uses that CLEAR Allows Create Benefits, not Harm, for Californians.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Cite this document
“Exhibit A-1 to Declaration of Susan Fahringer - Redacted -… — Brooks v. Thomson Reuters Corporation (Dkt. 186.3).”
Original at ecf.cand.uscourts.gov, No. 3:21-cv-01418-EMC, Doc. 186-3. Archived at Pandemic Darlings, https://pandemicdarlings.com/court-filings/full-docket-brooks-thomsonreuters-cand-374304/2023-03-28-gov-uscourts-cand-374304-186-3-gov-uscourts-cand-374304-186-3/
, SHA-256 d47abfbefdbd2007bdbd259e9c0aa59e59de3a930d86fbdbde7bff0e6d56335c.Full text
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 1 of 51
EXHIBIT A-1
REDACTED - PUBLICLY FILED VERSION OF DOC. 151-1
PURSUANT TO COURT ORDER,
DATED MARCH 21, 2023 (DOC. 178)
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 2 of 51
UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA
SAN FRANCISCO DIVISION
CAT BROOKS and RASHEED Case No. 3:21-cv-01418-EMC
SHABAZZ, individually and on behalf of
all others similarly situated, EXPERT REPORT OF PROF. JANE
BAMBAUER
Plaintiffs,
Judge: Hon. Edward M. Chen
v.
THOMSON REUTERS
CORPORATION,
Defendant.
1
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 3 of 51
TABLE OF CONTENTS
Page
I. Qualifications ............................................................................................................................................. 4
a. Background and Experience ................................................................................................. 4
b. Publications ............................................................................................................................. 4
c. Prior Expert Witness Work................................................................................................... 6
d. Compensation ......................................................................................................................... 6
II. Scope of the Report .............................................................................................................................. 6
III. Summary of Opinions .......................................................................................................................... 6
IV. Privacy Does Not Require Absolute Control ................................................................................... 6
a. Prof. Turow’s Suggestion that Unconsented Collection and Use of
Data Is Always a Violation of Privacy Is an Extreme and Untenable
Position .................................................................................................................................... 7
b. Privacy Principles Take Context and Social Necessities Into Account .......................... 8
c. Privacy Theories and Laws Recognize Some Per Se Legitimate Data
Practices, Some Per Se Illegitimate Data Practices, and Many Practices
That Are Ambiguous and Context-Dependent ...............................................................10
d. Factors to Consider in Context-Dependent Privacy Assessments................................12
e. A Responsible Data Steward Can Avoid Creating Privacy Harm,
Even if Individuals Do Not Have Full Control Over their Data, by
Securing Data and Limiting Its Access and Use ..............................................................15
V. The Limited Set of Permissible Uses that CLEAR Allows Create Benefits, not
Harm, for Californians ......................................................................................................................16
a. For the Direct Benefit of the Data Subject ......................................................................17
i. Use by a Person Holding a Legal or Beneficial Interest
Relating to the Consumer ......................................................................................17
ii. Administering a Transaction that a Consumer Requests or
Authorizes ................................................................................................................18
b. For the Protection of Others (and Often for the Indirect Benefit of
the Data Subject) ..................................................................................................................18
i. Complying with Legal Requirements and Legal Process ..................................19
ii. Protecting Against or Preventing Fraud ..............................................................20
2
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 4 of 51
iii. Use by a Law Enforcement Agency on a Matter Related to
Public Safety.............................................................................................................21
iv. Resolving a Consumer Dispute or Managing Institutional
Risk............................................................................................................................22
c. The Individual and Collective Benefits for Californians Are Highly
Dependent on the Particulars of the Data Subject, the Purpose of the
Search, and the Quality of the Data Relative to Other Available
Options ..................................................................................................................................23
d. Analysis of Benefits and Harms to the Named Plaintiffs ...............................................27
VI. CLEAR Has Strong Protections Against Risks of Misuse or Mistaken
Associations ........................................................................................................................................28
a. CLEAR Minimizes the Risk of Misuse Using Credentialing, Training,
User Interface Design, Query Logs, and Audits ..............................................................29
b. CLEAR Minimizes the Risk of Overcollection and Mistaken
Associations Using Data Quality Procedures, User Interface Design,
and Training ..........................................................................................................................31
VII. Conclusion ............................................................................................................................................33
3
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 5 of 51
I. Qualifications
a. Background and Experience
I am the Dorothy H. and Lewis Rosenstiel Distinguished Professor of Law at the University of
Arizona College of Law. I am the director of the QuantLaw program (using quantitative methods to
study the effects of laws) and the co-director of the TechLaw Program at the College of Law. I am
also the co-deputy director of the Center for Quantum Networks, a multi-institutional engineering
research center funded by the National Science Foundation, where I facilitate research across the
country on the economic and regulatory implications of emerging quantum communications
technologies.
My research focuses on the social costs and benefits of Big Data and how new information
technologies affect privacy, free speech, and competitive markets. I am particularly interested in the
ways that new technologies can upend longstanding assumptions and intuitions about privacy,
discrimination, and governance. To advance the privacy discourse, I synthesize existing
experimental, statistical, and theoretical research about the real-world impact of privacy laws or Big
Data practices so that policy discussions can be grounded in empirical facts rather than speculation.
I have also conducted my own empirical research on consumer preferences and attitudes related to
the collection, disclosure, or repurposing of personal data. Because of my understanding and
experience with data-driven research, I can apply insights from the highly technical literature
produced by economists and computer scientists to the discussion of privacy, constitutional law, and
public policy.
I have served as the Reporter for the Uniform Law Commission’s model data privacy act (the
Uniform Personal Data Protection Act), as the Director of George Mason’s Program on Economics
& Privacy, and as a Board Member of the TPRC Conference on Communications, Information, and
Internet Policy. I am on the program committee for UNESCO’s Privacy in Statistical Databases
conference and journal. I have testified before the Senate Judiciary Committee on the impact of the
GDPR and CCPA on competition and consumer welfare, and I have testified twice before the
Federal Trade Commission on topics related to privacy. My privacy-related work has been quoted in
the Wall Street Journal, the Boston Globe, the Washington Post, the New York Times, the L.A.
Times, TIME, TechDirt, and Sports Illustrated. I am also the co-founder and executive editor of the
Journal of Free Speech Law, the first peer-reviewed law journal on the topic of free speech. I teach
courses on Information Privacy, Torts (including Advanced Torts, which covers the privacy torts),
Criminal Procedure, Free Speech, and a seminar on the social consequences of the Internet. I hold a
B.S. in mathematics and a J.D. from Yale.
b. Publications
My privacy- and Big Data-related work spans from highly technical papers discussing new methods
for anonymizing data and measuring reidentification risk to more broad-ranging pieces related to the
impact of new technologies on Fourth Amendment theory. My research has appeared in peer-
reviewed journals, top law reviews, and mainstream media. I include here a selection of work. My
complete resume is attached as Exhibit A.
4
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 6 of 51
Articles
Reckless Associations, 36 HARVARD JOURNAL OF LAW & TECHNOLOGY __ (forthcoming 2023) (with Saura
Masconale and Simone Sepe)
Plausible Deniability, 2020 PRIVACY IN STATISTICAL DATABASES 91 (2020) (peer-reviewed) (co-authored with
David Sidi).
The Algorithm Game, 94 NOTRE DAME L. REV. 1 (2018) (with Tal Zarsky).
Other People’s Papers, 94 TEXAS LAW REVIEW 205 (2015).
Is Data Speech?, 66 STANFORD LAW REVIEW 57 (2014).
The New Intrusion, 88 NOTRE DAME LAW REVIEW 205 (2012).
Tragedy of the Data Commons, 25 HARVARD JOURNAL OF LAW & TECHNOLOGY 1 (2011) (published as Jane
Yakowitz).
Books
BUBLICK, BAMBAUER & ARELLANO, DOBBS ADVANCED TORTS: ECONOMIC AND DIGNITARY TORTS (2d
Ed.) (2022).
Book Chapters
Sober and Self-Guided Newsgathering, invited chapter in MEDIA AS A GOVERNANCE INSTITUTION (Gus Hurwitz
et al., eds., Cambridge University Press, forthcoming 2023).
Conceptions of Fairness in Artificial Intelligence, invited chapter in THE PHILOSOPHICAL FOUNDATIONS OF IT
LAW (David Mangan, ed., Oxford University Press, forthcoming 2022).
Privacy Versus Research in Big Data, in THE CAMBRIDGE HANDBOOK OF CONSUMER PRIVACY (Evan Selinger
et al., eds., Cambridge University Press, 2018).
Essays and Letters
Reinventing Cambridge Analytica One Good Intention at a Time, LAWFARE (2022).
Facial Recognition as a Least Bad Option, HOOVER INSTITUTE REPORT (2021).
COVID Apps Are Terrible. They Didn’t Have to Be. LAWFARE REPORT (2020) (with Brian Ray).
Are Privacy Disclosures Informational or Ideological?, 65 DEPAUL L. REV. 503 (2017) (Clifford symposium issue).
Assessing Data Intrusion Threats, SCIENCE (letter) (with D. Barth-Jones et al.) (2015) (peer-reviewed).
How the War on Drugs Distorts Privacy Law, 64 STAN. L. REV. ONLINE 131 (2012).
Op-Eds
Bills Meant to Check Big Tech’s Power Could Lead to More Disinformation, WASHINGTON POST (June 6, 2022) (with
Anupam Chander).
5
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 7 of 51
Letting Police Access Google Location Data Can Help Solve Crimes, WASHINGTON POST (March 28, 2022).
c. Prior Expert Witness Work
I have served as an expert witness in one other case—Sander v. Regents of the University of California, in
support of plaintiffs seeking public records.
d. Compensation
My hourly rate is $700. My compensation does not depend in any way on the substance of my
opinion or the outcome of this matter.
II. Scope of the Report
I understand that plaintiffs in this case seek class certification in their lawsuit against Thomson
Reuters stemming from its CLEAR platform. I have been retained to become familiar with the
CLEAR platform—how it works, what purposes it is used for, and what precautions it takes against
unauthorized access or use—and to assess the analysis and conclusions of plaintiffs’ expert,
Professor Joseph Turow. I have limited my report to an analysis, critique, and response to Prof.
Turow’s opinions as set forth in his report dated June 1, 2022 (“Turow Report”) and deposition
testimony. In order to fully respond to the Turow Report, I have provided a methodology to assess
privacy harms as a corrective to the methodology that the Turow Report and Prof. Turow's
testimony puts forward.
In authoring this report, I considered the documents listed throughout the report and in Exhibit B.
The opinions contained in my report are based on the information and documents available to me as
of the date of this report. If additional information or documents become available, I reserve the
right to modify or supplement the opinions expressed in this report based on that newly available
information.
III. Summary of Opinions
The Turow Report wrongly insists that any unconsented collection, disclosure, or use of personal
data constitutes a privacy harm. Prof. Turow’s conception of privacy harm as any loss of control,
no matter how trivial, and no matter how beneficial the data practice may be to the data subject or
others, is much too rigid and is well outside the bounds of scholarly consensus. In part because of
his absolutist definition of privacy, the Turow Report fails to account for socially valuable and
desirable uses of data (even without data subjects’ consent) and therefore cannot illuminate a means
of reconciling privacy with competing individual and collective interests. When the particular uses of
the CLEAR platform are analyzed, it becomes obvious that Prof. Turow’s conclusions that CLEAR
is harmful to Californians—and is harmful to all Californians in exactly the same way and to the
same degree—is detached from any context or real-world consequences. My report provides a
corrective by taking into account the specific uses for which the CLEAR platform is allowed to be
accessed, the safeguards by which Thomson Reuters prevents and detects misuse, and the highly
variable and fact-dependent nature of the impact of CLEAR on each individual California resident.
IV. Privacy Does Not Require Absolute Control
The Turow Report wrongly claims that every unconsented collection or use of personal information
is a privacy injury. In his opinion, the only question that needs to be answered in this case in order
to understand the extent of privacy harm is whether CLEAR interferes with a data subject’s control
of their data. (Turow Report at 13). This is not correct; while consent and data subject control have
a major influence over whether privacy has been respected, avoiding privacy harms has never
required absolute control to be vested in the data subject. To the contrary, scholars and lawmakers
6
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 8 of 51
have understood that the broader social contract requires individuals’ interest in controlling
information about them to be balanced against, and sometimes overcome by, other interests and
needs.
a. Prof. Turow’s Suggestion that Unconsented Collection and Use of Data Is
Always a Violation of Privacy Is an Extreme and Untenable Position
Prof. Turow concludes that Californians suffer harm because “no Californian has control over the
information in or use of dossiers about them” (Turow Report at 13). According to Prof. Turow, the
only legitimate test for determining whether privacy has been respected is whether a person has
consented. (Turow Report at 14).
Regarding data access or collection, the Turow Report claims that compiling and cross-referencing
data is automatically a violation of privacy irrespective of the source of the data, how the data is
used, or what precautions are in place to prevent abuse:
CLEAR’s very business model is to allow companies to take for granted that huge
numbers of datapoints about Californians are compiled, connected, updated (including,
for certain data, in real-time) and made available for Thomson Reuters’ customers at the
touch of a button. . . Before the internet era, melding information from even public
databases was a cumbersome process, and so the material a company could bring
together about a particular individual was limited.” (Turow Report at 17).
Prof. Turow concludes that any collection and organization of data, even publicly available data,
“deprives all Californians of the right to control their personal data” (Id.). 1
With respect to how data is used, the Turow Report claims that a service allowing personal
information to be accessed for any reason would violate privacy unless the target of the search has
provided consent.
[W]hether a person has consented to the sale of their data has nothing to do with
whether Thomson Reuters lists the use as permissible. And so, whatever the function of
a customer certifying a purportedly permissible use, it does nothing to solve the privacy
harms of lack of control or violation of the right to be let alone. No Californian is given
any control over the uses of dossiers about them. There is no use of CLEAR profiles
that is permitted by the Californians whose information is sold via CLEAR.
Turow Report at 14.
Prof. Turow has described actual uses of CLEAR as “hypotheticals” that “get away from the main
point” of his report’s conclusion that unconsented collection of data is a “first order” harm that
overrides all other considerations. (Turow Dep. 269-70). The Turow Report did not consider the
social benefits that can come from specific collection, disclosure, and uses of data for limited
purposes—like the detection of fraud or the provision of government benefits.
1
The Turow Report also suggests that CLEAR automatically violates privacy and causes harm because Thomson
Reuters is not actively engaging in a public awareness campaign and encouraging individuals to opt out or change their
data. (Turow Report at 15-16). But this, too, is an overstatement of what privacy principles would require. The principle
of notice is typically satisfied by providing a privacy policy on a website. Information about CLEAR is available for the
privacy-concerned who seek it out. Moreover, the purpose of a data system would be undermined if, for example,
would-be money-launderers were alerted and permitted to opt out of a database used for Know Your Customer due
diligence, eviscerating the benefits outlined below in Part V.
7
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 9 of 51
This is an extreme position. Prof. Turow is correct in a general sense. Privacy theory and law has
long incorporated an interest and limited right for a data subject to control information about
themselves. For example, the document known as the “HEW Report” promulgated a set of “Fair
Information Practice Principles” or “FIPPs” that have been tremendously influential to modern
privacy law. 2 These principles include many of the themes Prof. Turow emphasizes, such as notice,
consent, and a constraint on repurposing data.
However, Prof. Turow’s definition of privacy (and of privacy harm) is much too broad and rigid.
His capacious conception of privacy harm would render the Google search bar harmful since
Google search allows scattered pieces of information to be drawn together quickly and cheaply.
(Turow Dep. 198; 206-210; and 326-27 (claiming that even collecting articles that a named plaintiff
published in news media or collecting information contained in public county records would be a
privacy harm if done without the data subject’s permission)). And by his logic, the GDPR, FCRA,
HIPAA, the Wiretap Act, and every other federal privacy law cause (or at least fail to prevent)
privacy harms because they permit many disclosures and uses of personal data without consent.
Indeed, even the warrant process prescribed in the Fourth Amendment is inadequate since the
subject of a warrant has not consented to the search that has been judicially authorized.
b. Privacy Principles Take Context and Social Necessities Into Account
The Turow Report is out of sync with generally-accepted theoretical, academic, and historical
frameworks for understanding privacy violations and harms because the report admits of no
exceptions to his claim that privacy requires data subject control. Nearly every privacy scholar
understands that privacy ensures a personal sphere that is shielded, but not absolutely closed off,
from the uses and needs of others. 3 The right of a data subject to control information has always
been understood as a limited one, where a person’s interest in control must be reconciled with, and
sometimes superseded by, other needs of society. These include physical and financial security, the
right of a free press, and compliance with a wide range of legal and ethical duties. Scholars debate
how widely or narrowly to define the scope of control, but have rarely, if ever, advocated for an
absolutist approach to privacy.
For example, the authors of the HEW Report, which articulated the “Fair Information Practice
Principles” that have informed every modern data privacy statute, explicitly rejected formulations of
privacy that required the absolute control of the data subject. 4 As it explained, “it would be
inconsistent with this essential characteristic of mutuality to assign the individual record subject a
unilateral role in making decisions about the nature and use of his record.” 5 Indeed, every law that
has used the FIPPs as their backbone has included exceptions in which data collection, access,
disclosure, or use can occur without the data subject’s consent. The Privacy Act (which governs how
the federal government manages its own collections of personal data) includes exceptions for
criminal law enforcement, the investigation of tax and social security fraud, for purposes of
“compelling circumstances affecting the health or safety of an individual”, and for “routine uses”
2
U.S. DEPT. HEALTH, EDUCATION & WELFARE, RECORDS, COMPUTERS, AND THE RIGHTS OF CITIZENS (1973); Robert
Gellman, Fair Information Practices: A Basic History, (describing the numerous statutory and regulatory laws that have used
the FIPPs as a basic structure) (hereinafter “HEW Report”).
3
See sources cited infra notes 5-15.
4
The HEW Report favored the concept of “mutuality” instead. HEW Report, supra note 2 at 3, 40.
5
Id.
8
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 10 of 51
without getting the data subject’s consent. 6 Many other US 7 and EU 8 laws similarly describe
scenarios in which data can be collected or used without the data subject’s consent. All of these laws
follow the very widely held conceptual framework that the data subject’s control over information
must yield to other interests in at least some circumstances in order to ensure that privacy rights do
not lead to absurd and harmful results.
Likewise, leading privacy law scholars including Alan Westin 9, Daniel Solove 10, Helen Nissenbaum11,
and Neil Richards 12 have recognized that information control is not always in the best interest of
society or of the data subjects themselves, and privacy does not and should not require consent in
every single conceivable case. 13 Prof. Turow cites Alan Westin repeatedly for the proposition that
privacy loss automatically occurs when people lose control over their data (Turow Report at 7;
Turow Dep. 28, 117), but Westin himself described control as a “central” but not essential factor. 14
He went on to explain that “the needs of social order may require certain levels of exposure and
confession even if these are involuntary. Consent is thus to be analyzed in the specific context of the
purpose of [data collection] and the use to be made of the information so obtained.” 15 Prof. Turow,
too, in his deposition testimony, acknowledged that the loss of control over one’s information is a
privacy harm “in particular contexts” and provided an example—disclosures made in the course of
medical treatment, or disclosures made in order to comply with a legal requirement—in which losing
control over personal data would not constitute a privacy harm. (Turow Dep. 101-102; 203-204).
It is telling that Prof. Turow and the named plaintiffs could not explain why people and companies
are allowed to collect or use individuals’ names without consent without undermining their theories
that loss of control over personal information always constitutes a privacy harm In response to the
question “Does the right to control one’s information, that is the subject of your report in this case,
extend to an individual’s right to prevent a company like a court reporting company from repeating
their name?” Turow replied “It’s a bad question.” (Turow Dep. 202). Pressed further, he explained
that he would have to know:
6
5 U.S.C. § 552a.
7
The Health Insurance Portability and Accountability Act (“HIPAA”), Pub. L. No. 104-191, 110 Stat. 1936 (codified as
amended in scattered sections of 18, 26, 29, and 42 U.S.C.); The Family Educational Rights and Privacy Act (FERPA)
(20 U.S.C. § 1232g); and the Electronic Communications Privacy Act, 18 U.S.C. §§ 2510–2523 (2006).
8
Europe’s General Data Protection Regulation (“GDPR”) allows data to be processed without consent in order to
comply with legal obligations, prevent fraud, to ensure network security, and to detect possible threats to public safety.
GDPR, OJ L 119, 04.05.2016, Art. 6 (2018).
9
ALAN WESTIN, PRIVACY AND FREEDOM 418, 431-34 (1967).
10
Daniel J. Solove, Introduction: Privacy Self-Management and the Consent Dilemma, 126 HARV. L. REV. 1880 (2013) (“in order
to advance, privacy law and policy must face the problems with privacy self-management and start forging a new
direction.”)
11
HELEN NISSENBAUM, PRIVACY IN CONTEXT (2010).
12
NEIL RICHARDS, WHY PRIVACY MATTERS 71 (2022) (“my own definition [of privacy] excludes other ways we could
talk about privacy, such as its being a right to control our personal information or to conceal disreputable information
about ourselves.”)
13
Kirsten Martin, Understanding Privacy Online: Development of a Social Contract Approach to Privacy, 137 J. BUS. ETHICS 551
(2016) (“Defining privacy as a state of inaccessibility is neither practical nor desirable and, ironically, renders privacy as a
form of punishment. . . These social contracts around what, to whom, and for what purposes information flows are the
governing rules about privacy for a given community.”); Timnit Gebru et al., Datasheets for Datasets, MICROSOFT (2018)
(encouraging database creators to ask the question “Did the individuals in question consent to the collection and use of
their data?” but not requiring the answer to determine, on its own, the privacy assessment of a program).
14
ALAN WESTIN, PRIVACY AND FREEDOM 418 (1967) (“A central aspect of privacy is that individuals and organizations
determine for themselves which matters they want to keep private and which they are willing—or need—to reveal.”)
15
Id.; see also id. at 431-434 (describing exceptions to the general rule that government agencies should not share data it
has collected without consent).
9
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 11 of 51
[W]hy would they be doing this? Is it a legal requirement? Under what conditions? Where are
the—where is the material located? Who has access to it? What can it be used for?
(Turow Dep. 203-204).
Again, this is in response to whether a company can use a person’s name. When Plaintiff Cat Brooks
was asked whether she has the right to control the use of her name, she responded “that’s an absurd
question” and “it would depend on how my name was being used in conjunction with other
information.” (Brooks Dep. 70).
c. Privacy Theories and Laws Recognize Some Per Se Legitimate Data
Practices, Some Per Se Illegitimate Data Practices, and Many Practices That
Are Ambiguous and Context-Dependent
In truth, there is no simple way to reconcile the fact that privacy principles give individuals some
degree of control over their information with the fact that other private and public interests can and
should override this control in some circumstances. The concept of “mutuality” from the HEW
Report is the most honest and enduring account of privacy, despite its imprecision, because it treats
privacy as part of a social contract—as part of a set of trade-offs and cooperation that people must
engage in with each other, with industry, and with the government. 16 Prof. Turow’s surveys finding
that Americans are concerned about the collection and use of their personal information does not
negate this inescapable trade-off. (Turow Report at 3). While it may very well be that most people
prefer, in general, and all else being equal, to have control over their data, these abstract preferences
say little about where legal rights and obligations should be drawn in real world contexts where
personal interests in privacy may come into conflict with other pressing personal and public
interests.
The HEW Report explains that a respect for privacy requires data collectors to anticipate and
behave responsibly when conflicts arise between their interests and those of the data subjects. Some
data uses are in the mutual interests of the controllers and the data subjects, some are in their mutual
interests but are “not perceived as such,” and some are in direct conflict. 17 This observation was
ahead of its time and explains why privacy rules are so difficult to articulate in advance. The interests
of all stakeholders sometimes run together and sometimes against each other, and a wide range of
contextual factors determine whether the data subject’s express consent is necessary to maintain
privacy. Data subjects do not want or expect to have absolute control over their data in all
circumstances. 18 And privacy should not be regarded as a rigid “procedural mechanism divorced
from the particularities” of a data practice. 19
For clarity of discussion, data practices can be organized into three categories: (1) practices that are
per se privacy violations; (2) practices that are clearly warranted by public needs and the social
16
The work of Helen Nissenbaum and Kirsten Martin have helped add definition to the idea of privacy as an intricate
social contract where consent is just one possible route (neither necessary nor automatically sufficient) to achieving
privacy. Kirsten E. Marin, Understanding Privacy Online, 137 J. BUS. ETHICS 551 (2016) (hereinafter “Understanding Privacy
Online”); Kirsten E. Martin, Diminished or Just Different? A Factorial Vignette Study of Privacy as a Social Contract, 11 J. BUS.
ETHICS 519 (2012); Helen Nissenbaum, Privacy as Contextual Integrity, 79 WASH. L. REV. 119 (2004); HELEN NISSENBAUM,
PRIVACY IN CONTEXT (2010); Mary J. Culnan & Robert J. Bies, Consumer Privacy: Balancing Economic and Justice
Considerations, 59 J. SOCIAL ISSUES 323 (2003); J. (Hans) Van Oosterhout et al., The Internal Morality of Contracting: Advancing
the Contractualist Endeavor in Business Ethics, 31 ACADEMY OF MAGMT. REV. 521 (2006).
17
HEW Report, supra note 2 at 46.
18
Martin, Understanding Privacy Online, supra note 13 at 555 “Allowing privacy rules to vary based on the community
or relationship mirrors expectations of privacy in the world.”
19
Helen Nissenbaum, A Contextual Approach to Privacy Online, DAEDALUS, THE JOURNAL OF THE AMERICAN ACADEMY
OF ARTS & SCIENCES 32, 35 (2011).
10
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 12 of 51
contract (per se non-violations); and (3) the messy middle category, where the threat to reasonable
expectations in privacy will depend on controls and procedures that are in place, the sensitivity of
the data, and the costs and benefits of the practice.
1. Per se violations. Per se privacy violations involve observations, disclosures, or uses of
data that are nearly universally unwanted and disturbing and are also unnecessary for the welfare of
the data subject or the community. Examples of per se privacy violations include intrusive
observations or recordings of private spaces and conversations (e.g. upskirt photography or
wiretapping phone lines)20; indiscriminate data disclosure or completely unconstrained data reuse 21;
or a clearly harmful use (e.g. using personal data to create “sucker lists” of individuals who are
especially vulnerable to fraud or manipulation). 22
2. Per se non-violations. Examples of well-justified data practices (per se non-violations)
include the exceptions that are common in privacy statutes. These include disclosures or analyses that
are required by law; data practices that address a serious public health or safety risk; practices done
for the purpose of benefiting the data subject in a context where consent is impracticable; practices
that are necessary for detecting fraud; and practices that are necessary for executing a transaction or
the terms of an agreement that the data subject sought out. 23 In all of these cases, even if there is not
universal agreement, most would agree in advance to be part of a system that allowed these trade-offs
between a data subject’s control over personal information and the benefits to third parties or even to
themselves that can occur if data is accessed without their control. 24
3. Context-Dependent Privacy Violations. Between the per-se violations and per-se non-
violations is a vast middle category where a wide range of factors will have to be considered in order
to determine whether a particular practice, carried out in its real-world context, will cause a privacy
injury. These include the sorts of data practices that Prof. Turow describe as “lawful and common”
that also diminish a person’s ability to control their information, such as online tracking for the
purposes of behavioral advertising. (Turow Dep. 80-82 (discussing his own NY Times op-ed)).
For practices in this middle category, which I will refer to as the “messy middle,” a number of
factors will be relevant to a privacy assessment, and in some cases reasonable minds will disagree on
whether a data practice is appropriate. Still, most privacy experts would home in on a few factors
that cut either for or against the recognition of a privacy harm. Relevant factors include over-
collection (recording more data than is likely to be useful for legitimate primary or secondary
purposes), security and probability of misuse, the transparency of a system, the expectations and
preferences of individual data subjects, the costs of obtaining consent (both financial costs and the
costs to the utility of the data for the particular purpose), and the actual and perceived costs and
benefits to the individual, to the data controller, to third parties, and to society. Identifying a privacy
20
Nader v. General Motors Corp., 255 N.E.2d 765 (N.Y. Ct. App. 1970); The Wiretap Act, 18 U.S.C. §§2510-20; The
Video Voyeurism Prevention Act, 18 U.S.C. §1801.
21
Rest. (2d) of Torts §652D: Publicity Given to Private Life
22
Maddy Varner & Aaron Sankin, Sucker List: How Allstate’s Secret Auto Insurance Algorithm Squeezes Big Spenders, THE
MARKUP (February 25, 2020); Gema de las Heras, Scammers Target Disaster Victims. Spot Their Traps, FEDERAL TRADE
COMMISSION (August 2, 2022).
23
For example, a survey of privacy professionals conducted by the Ponemon Institute at Accenture found that most
privacy professionals agree it is acceptable to use personal information to identify and authenticate customers, to share
information with law enforcement, for fraud prevention, and for government and national security purposes. HOW
GLOBAL ORGANIZATIONS APPROACH THE CHALLENGES OF PROTECTING PERSONAL DATA 23 (2009).
24
Richard Re, Fourth Amendment Fairness, 116 MICH. L. REV. 1409 (2018) (proposing a contractualist approach to Fourth
Amendment privacy).
11
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 13 of 51
violation requires consideration of the totality of these factors, and the hallmark of a violation is
when the privacy risks cannot be justified by the benefits of a practice. 25
d. Factors to Consider in Context-Dependent Privacy Assessments
When determining whether a practice violates generally accepted principles and expectations of
privacy, the following factors will be relevant.
Actual and Perceived Benefits of a Data Practice: When it comes to the benefits of data use, privacy
expectations and preferences often operate as a sliding scale: the more benefit the data subject
derives from data use, the less invasive the practice. The economic literature on the “privacy
paradox” provides a glimpse into the tradeoffs between privacy and benefits. The privacy paradox is
the frequently replicated phenomenon where individuals report concern about their privacy but are
also willing to give up control of their personal data for small payments or perks. 26
While there are several theories that could explain the paradox, the most parsimonious explanation
is that whatever respondents say about privacy in the abstract, their concerns about privacy are
adjusted up or down depending on a large range of factors specific to each individual and each data
practice. 27 The intrinsic or deontological value of privacy in real-world contexts is fairly low for
most people, and so their context-specific concerns about a data practice can be outweighed by the
context-specific benefits. 28 My own experimental research has found that most Americans primarily
25
The FTC explains the sliding scale like this: “the level of transparency, access, and correction should be tied to the
significance of the benefit or transaction in question.” FEDERAL TRADE COMMISSION, DATA BROKERS: A CALL FOR
TRANSPARENCY AND ACCOUNTABILITY ix (2014) (hereinafter “FTC Data Brokers Report”).
26
Susan Athey et al., The Digital Privacy Paradox: Small Money, Small Costs, Small Talk, NBER WORKING PAPER NO. 23488
(2017); Kai-Lung Hui et al., The Value of Privacy Assurance: An Exploratory Field Experiment, 31 MIS QUARTERLY 19 (2007);
Jan H. Schumann et al., Targeted Online Advertising: Using Reciprocity Appeals to Increase Acceptance Among Users of Free Web
Services, 78 J. MARKETING 59 (2014). Consumers rarely alter the defaults in privacy settings (Adam S. Chilton & Omri
Ben-Shahar, Simplification of Privacy Disclosures: An Experimental Test, 45 J. LEG. STUD. 41 (2016); Eric J. Johnson et all.,
Defaults, Framing and Privacy: Why Opting In-Opting Out, 13 MARKETING LETTERS 5 (2002)) and the drafting of privacy
notices has almost no effect on behavior (Lior J. Strahilevitz & Matthew B. Kugler, Is Privacy Policy Language Irrelevant to
Consumers?, 45 J. LEG. STUD. 569 (2016); Adam S. Chilton & Omri Ben-Shahar, Simplification of Privacy Disclosures: An
Experimental Test, 45 J. LEG. STUD. 41 (2016)). In my own work, I have found that privacy-related notices are often
wasteful and do not change consumer choices. Dramatic just-in-time disclosures have the best potential to change
behavior, but they also run the risk of exaggerating a sense of threat and distorting the consumer’s evaluation of other
criteria. Jane Bambauer et al., A Bad Education, 2017 IL. L. RE V. 109 (2017). Even when consumers are paying attention
to privacy options, they rarely forego a service or benefit that they would otherwise want in order to protect their
privacy. Lior J. Strahilevitz & Matthew B. Kugler, Is Privacy Policy Language Irrelevant to Consumers?, 45 J. LEG. STUD. 569
(2016); Susan Athey et al., The Digital Privacy Paradox: Small Money, Small Costs, Small Talk, NBER WORKING PAPER NO.
23488 (2017); Alessandro Acquisti, Privacy and Human Behavior in the Age of Information, 347 SCIENCE 509 (2015) (providing
a summary of related scholarship).
27
Kirsten Martin & Helen Nissenbaum, Measuring Privacy: An Empirical Test Using Context to Expose Confounding Variables,
18 COLUM. SCI. & TECH. L. J. 176 (2017). See also Kirsten Martin & Helen Nissenbaum, 31 HARV. J. L. & TECH. 111, 130-
38 (2017) (describing some of the contextual factors that appear to matter when public data is disclosed to different
entities).
28
Long Chen et al., The Data Privacy Paradox and Digital Demand, NBER WORKING PAPER NO. 28854 (2021) (finding that
individuals who have the greatest concern for privacy also get the greatest value from data-sharing); Culnan & Bies, supra
note 16; Tamara Dinev & Paul Hart, An Extended Privacy Calculus Model for E-Commerce Transactions, 17 INFO. SYS.
RESEARCH 61 (2006); Hui et al., supra note 26; Heng Xu et al., The Role of Push-Pull Technology in Privacy Calculus: The Case of
Location-Based Services, 26 J. MGMT. INFO. SYS. 135 (2009); Susanne Barth & Mennode Jong, The Privacy Paradox—
Investigating Discrepancies Between Expressed Privacy Concerns and Actual Online Behavior—A Systematic Literature Review, 34
TELEMATICS AND INFORMATICS 1038 (2017); Kirsten Martin, Breaking the Privacy Paradox: The Value of Privacy and
Associated Duty of Firms, 30 BUS. ETHICS QUART. 65, 72 (2019) (summarizing this literature) (“Uses of information
12
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 14 of 51
value privacy for its instrumental purposes—to avoid concrete risks to themselves and others—and
will therefore perceive less of a threat to privacy when the benefits of a data practice outweigh those risks. 29
Cat Brooks’ deposition testimony illustrates the privacy paradox very well. On one hand, she states
“I believe that any personal information about me that I do not disclose myself that is collected,
stored, given, sold, whatever, to anybody else is an invasion of my privacy.” (Brooks Dep. at 74-75).
And yet, she also explains that if another person does redisclose information that they have learned
about her, the question of whether a privacy invasion has occurred “depends on who they’re
parroting it to. …. And for—not just who, but for what purpose.” (Brooks Dep. at 77-78). For
example, her Wikipedia page, which collects information about her from multiple sources and
discloses her birth name, would seem to violate the definition of privacy she offered in the abstract
since she did not create the page herself. (Brooks Dep. at 96, 99-100). But she did not object to it
and is ok with its existence. (Brooks Dep. at 96; 99-100). She also does not make use of her rights, as
a Californian, to opt out of the sale of her data through any websites other than CLEAR even
though this is a form of control she could exercise right now. (Brooks Dep. at 146; 84). Ultimately,
Brooks recognized that her philosophy about her right to control her personal information
“depends on the conditions around how that information is being repeated” and “if that
information is repeated with the potential to impact my life or the peoples whose lives are associated
with mine in a negative way.” (Brooks Dep. at 121). In other words, even in Brooks’ view, her right
to control information depends on the consequences of subsequent data disclosures and uses. This
is consistent with her explanation that she conducts cost-benefit analyses when deciding how much
she will reveal about herself and in what context. (Brooks Dep. at 77; 120; 138).
The Turow Report makes no attempt to balance, or compare, the harm from lost control over
personal data to the potential benefits either to the data subjects or to others. In deposition, Prof.
Turow explained that while he understood a court will weigh the harm to Californians against the
utility of a data practice, his report does not offer any opinion about the utility of the CLEAR
product or how the magnitude of benefits compares to the magnitude of harms. (Turow Dep. 112).
Prof. Turow did not offer an opinion about where the balance should be struck. (Turow Dep. 114-
15). In Prof. Turow’s opinion, “my mandate is to discuss the—the first order privacy harms, which
override” any analysis of benefit. Thus, his working definition of privacy harm is incomplete,
accounting for social losses alone without any accounting of social gains.
Person-Specific Privacy Preferences: Prof. Turow concludes that the CLEAR platform causes the exact
same privacy harm to all Californians. (Turow Report at 13). However, not only does he fail to
consider offsetting benefits, he does not explain why all Californians would be harmed in the same
way and to the same degree. In fact, he specifically did not consider individual differences among
class members: in deposition, he testified that he was not asked to evaluate the harm or objections
that might apply at “the individual level.” (Turow Dep. 148-150 (“The issue is not an individual class
member. The issue is the larger class of Californians.”)) But a complete evaluation of privacy harm
must take into account differing levels of sensitivity and preferences (as well as differences in the
impact of potential benefits). The risks and harm associated with a data practice is interpreted, at
least to some degree, through an individual’s experience of loss of dignity, loss of a sense of control,
and upset expectations. 30 These will vary not only by data practice, but also by person. Some people
deemed privacy violations in consumer surveys may be better judged after taking into consideration the benefits of
sharing information online.”)
29
Jane Bambauer, Privacy Tradeoffs: Who Should Make Them, and How?, working paper available at
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3905024; Jane Bambauer et al., A Bad Education, 2017 U. ILL. L.
REV. 109 (2017).
30
Daniel Solove, A Taxonomy of Privacy, 154 U. PENN. L. REV. 477, 507 (2006) describing the harms from data
aggregation. Note that Solove also describes significant economic benefits as well and does not take a position on
whether or when privacy harm occurs from aggregation and reuse.).
13
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 15 of 51
are much more privacy-sensitive than others 31, and even individuals who have the same average
privacy sensibilities may still have different expectations and dignitary losses depending on who is
accessing data or for what purpose. 32 The “messy middle” category includes data practices where
some individuals are willing to relinquish control and others are not. For example, some consumers
prefer to disclose information for the purpose of content personalization or to access free services
while others do not. 33
Actual and Perceived Benefits of Privacy Controls: Just as data practices in the middle category will be seen
as beneficial to some and less so to others, the benefits of privacy controls are similarly mixed and
individualized. While individuals who are privacy-sensitive will prefer having consent-based control
over their information, others who are privacy pragmatists could see data subject control as a cost in
some circumstances because it creates an imposition on them to manage requests and opt-ins for
services that are, in their mind, clearly worth it. 34 Some individuals might perceive consent
procedures as getting in the way of the usefulness of the data. For example, as I discuss in more
detail below, data practices that are designed to detect fraud would be undermined if the target of an
investigation had to be consulted in advance. Also, data practices that depend on a representative
sample of data in order to accomplish their goals could suffer from selection bias if only some
potential data subjects opt in. 35 Consent-based privacy controls can also frustrate competition policy
goals because the companies that are in the best position to collect and manage consents and to
combine a larger variety of types of data are the largest companies that already dominate their
markets—Google and Facebook, for example. 36
The experience in Europe shows that there are at least some disadvantages to consent-based privacy
regimes. European privacy law (the General Data Protection Regulation, or “GDPR”) caused
investment in research and development for new technology startups in the EU to falter, and
productivity of EU firms fell as compared to US counterparts. 37 Meanwhile, the internal budgets for
privacy offices at companies of all sizes (including in the U.S.) increased 29%. 38 Costs to American
companies if they have to comply with GDPR-style laws are estimated to be approximately $480 per
31
Alan Westin found that a sizable subset of individuals were privacy-unconcerned, some were pragmatists (willing to
trade privacy for benefits), and some were privacy sensitive. Alan F. Westin, Social and Political Dimensions of Privacy, 59 J.
SOCIAL ISSUES 431, 445 (2003) (describing survey work conducted in the 1990s).
32
The same person also has different privacy preferences over time. Avi Goldfarb & Catherine Tucker, Shifts in Privacy
Concerns, 102 AM. ECON. REV. 349 (2012).
33
Xu et al., supra note 28; Syagnik (Sy) Banerjee & Ruby Roy Dholakia, Mobile Advertising: Does Location Based Advertising
Work?, 3 INT. J. MOBILE MARKETING 68 (2008).
34
Outside Facebook, where sustained attention and criticism has caused a greater proportion of users to change their
privacy settings (Andrew Perrin, Americans Are Changing Their Relationship with Facebook, PEW RESEARCH CENTER (2018),
available at http://www.pewresearch.org/fact-tank/2018/09/05/americans-are-changing-their-relationship-with-
facebook/), the great majority of consumers do not bother to change default settings related to their privacy. This is true
even in Europe, where only 12.5% of Europeans opt out of data collection despite the very prominent and easy-to-use
pop-up windows providing the opportunity to do so. Guy Aridor et al., The Effect of Privacy Regulation on the Data Industry:
Empirical Evidence from GDPR, NBER WORKING PAPER NO. 26900 (2021).
35
Jane Yakowitz Bambauer, Tragedy of the Data Commons, 25 HARV. J. L. & TECH. 1, 61, 64 (2011) (published as Jane
Yakowitz).
36
James Campbell et al., Privacy Regulation and Market Structure, 24 J. ECON. & MGMT STRATEGY 47 (2015); Garrett
Johnson et al., Privacy and Market Concentration: Intended & Unintended Consequences of the GDPR (2022) (draft available at
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3477686); Christian Peukert et al., European Privacy Law and Global
Markets for Data (2020) (draft available at https://doi.org/10.3929/ethz-b-000406601); Jian Jia et al., The Short-Run Effects
of GDPR on Technology Venture Investment, NBER WORKING PAPER NO. 25248 (2018); Ran Zhuo et al., The Impact of the
General Data Protection Regulation on Internet Interconnection, NBER WORKING PAPER NO. 26481 (2020).
37
Jia, supra note 36; EUROPEAN COMMISSION, 2021 DIGITAL ECONOMY AND SOCIETY INDEX (DESI) 6, 9 (2021).
38
IAPP-EY Annual Privacy Governance Report 2021, THE INTERNATIONAL ASSOCIATION OF PRIVACY PROFESSIONALS
(IAPP) (2021).
14
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 16 of 51
data subject. 39 These costs cause the prices that consumers pay to rise—a cost that may well be
worth it to some people and in some contexts, but not to every person in every context. Thus, even
in the middle category where the benefits and risks of a data practice are mixed and subject to
differences of opinion, Prof. Turow is wrong to suggest consumer consent is a panacea to modern
data stewardship puzzles.
All of the foregoing factors must be kept in perspective when determining whether a privacy
violation has occurred with respect to practices in the middle category. In the middle category of
data practices, reasonable minds will disagree and analysis of privacy is more complex and context-
dependent.
e. A Responsible Data Steward Can Avoid Creating Privacy Harm, Even if
Individuals Do Not Have Full Control Over their Data, by Securing Data and
Limiting Its Access and Use
A company can avoid causing privacy invasions entirely if it collects or uses data only for a limited
set of clearly beneficial purposes (per se non-violations) and if it maintains the data with adequate
security and care.
Most garden variety data aggregators operate in the “messy middle” category and thus raise
challenging questions for courts and lawmakers. Garden variety data aggregators are those
companies that collect and sell personal data for a wide range of marketing purposes including
targeted advertising, customer lists, market research, and to sell people search services to the general
public. 40 CLEAR is not a garden variety data aggregator. It is a search tool that is accessible only for a
narrow set of socially valuable risk mitigation purposes. 41 CLEAR is designed and managed in a
manner that stays within the bounds of per-se non-violations of privacy because the uses it permits
confer substantial benefits on data subjects individually and collectively without creating meaningful
new risks.
Prof. Turow has taken an extreme position, claiming that any loss of any aspect of control over
personal data is a privacy harm. According to Prof. Turow, the collection of publicly available
information, without affording the data subject the opportunity to control it, would be a privacy
harm even if the data is never subsequently used or disclosed. (Turow Dep. 163-168).
Prof. Turow may be taking this absolutist position on consent and control because he has conflated
the CLEAR platform with hypothetical services that might collect every possible piece of
information about individuals and sell access to “dossiers” for unlimited use. Prof. Turow claims the
“breadth of information available in CLEAR, and the variety of uses for which Thomson Reuters
permits business and governmental entities to access information through CLEAR, cover many
aspects of Californian’s lives.” (Turow Report at 13). He states that “CLEAR [] provides customers
the ability to coalesce large numbers of data points about individuals quickly and seamlessly” (Turow
at *10) without acknowledging the strict limitations on use that are imposed. (See also Turow Dep.
204 (claiming that CLEAR uses data “willy nilly”). Prof. Turow draws parallels between this case and
39
Alan McQuinn & Daniel Castro, The Costs of an Unnecessarily Stringent Federal Data Privacy Law, INFORMATION
TECHNOLOGY AND INNOVATION FOUNDATION (ITIF) (2019).
40
FTC DATA BROKERS REPORT, supra note 25 at ii-iii (describing marketing products and people search products and
differentiating them from risk mitigation products).
41
Federal Trade Commission does not consider risk mitigation data services to be engaged in an unfair business practice
or to pose harm to consumers. Id. at iii, vii-ix. Because these types of data services are used to verify the identity or
location of a data subject during a moment of vulnerability, or to assess the risk that the data subject is committing fraud
or criminal activity, the FTC has all but endorsed these data products and have called for no legal intervention into their
business models or practices.
15
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 17 of 51
his critiques of the digital behavioral advertising industry. 42 Indeed, in his deposition testimony, he
describes Thomson Reuters’ CLEAR platform as an example of “a company that is selling audiences
to other companies.” (Turow Dep. 51 (emphasis added); see also Turow Dep. 215 (equating the use
of CLEAR to “a marketing context”)). But CLEAR does not allow its service to be used to
construct and buy audiences.
To summarize, Prof. Turow asserts that any collection or use of personal data that is done without
the data subject’s consent and control is a violation of privacy and causes privacy harm. This is not
correct. For some highly beneficial or socially valuable purposes, data can be collected and used
without consent without conflicting with any reasonable notions of privacy. And there are also some
data practices that will either offend or conform to reasonable notions of privacy depending on
context.
As the next Parts explain, the CLEAR platform is available only for a limited set of uses that are
socially valuable and do not offend contextualized notions of privacy, and Thomson Reuters
incorporates many protocols to prevent and detect misuse of the platform.
V. The Limited Set of Permissible Uses that CLEAR Allows Create Benefits, not Harm,
for Californians
Prof. Turow claims that “every Californian whose information is accessible through CLEAR has
suffered the same fundamental privacy harm of a lack of control and violation of their right to be let
alone.” (Turow Report at 13). 43 Because the Turow Report uses such an extreme definition of
privacy harm, where every unconsented collection or use is automatically harmful, his report does
not make clear how some of the other factors he highlights—the volume of data (Turow Report at
10), the “variety” of permitted uses (Turow Report at 13), or the lack of notice to data subjects
(Turow Report at 15)—relate to his ultimate conclusion that Californians are harmed. As a result, it
is not clear what type of carefully designed and thoughtfully managed data service would satisfy
Prof. Turow’s tests. It is not clear, in other words, that there is any possible version of CLEAR that
could exist that Prof. Turow would recognize as harmless or even beneficial to Californians. Even a
version of CLEAR that is restricted only to the most urgent purposes—locating an abducted child,
for example—would cause harm to every person whose data is made theoretically accessible without
their consent by Prof. Turow’s vision for privacy. (Turow Dep. 274-77).
Thomson Reuters allows the CLEAR platform only to be used for nine permissible uses, which are
the same nine permissible uses under the Gramm-Leach-Bliley Act. (TR-BROOKS013537;
42
JOSEPH TUROW, THE VOICE CATCHERS: HOW MARKETERS LISTEN IN TO EXPLOIT YOUR FEELINGS, YOUR
PRIVACY, AND YOUR WALLET (2021); JOSEPH TUROW, THE AISLES HAVE EYES: HOW RETAILERS TRACK YOUR
SHOPPING, STRIP YOUR PRIVACY, AND DEFINE YOUR POWER (2017); JOSEPH TUROW, NICHE ENVY: MARKETING
DISCRIMINATION IN THE DIGITAL AGE (2006).
43
He also argues that the CLEAR service causes economic harm because it misappropriates data subjects’ own
opportunity to sell information about themselves. (Turow Report at 5-6; Turow Dep. 227 “someone would get a
different valuation on their data than would otherwise be the case; or might sell it to another organization that would try
to get a different price and couldn’t because of Thomson Reuters.”). This is not correct. There is very little economic
incentive for individuals to gather and manage their own databases about themselves and to overcome the myriad
transaction costs that would be required for companies to contract for access to that data with each and every data
subject. These would be a classic example of insurmountable transaction costs. Ronald H. Coase, The Problem of Social
Costs, 3 J. LAW & ECON. 1 (1960) (explaining the concept of transaction costs). In theory, a company could offer money
to data subjects in order to collect and aggregate data for future sale and re-use, but this is essentially what Google,
Facebook, and other companies already do, using free services rather than cash payments, and Turow has criticized this
business model.
16
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 18 of 51
Response to Interrogatories Attachment A (hereinafter (“Attachment A”) at *2). 44 None of the
opinion surveys that Prof. Turow conducted or relied upon asked survey respondents to react to a
service that was similar to CLEAR (Turow Dep. 71 (“Not – not similar to CLEAR, no”)), and so an
analysis must rely on other resources and general knowledge in order to determine whether the
CLEAR product is consistent with a social contract that most Californians would want given full
information about risks and benefits.
One of the permissible uses is with the explicit consent of the data subject. Another is the functional
equivalent, when a fiduciary or representative of the data subject has authorized the search.
(Attachment A at *3). Since the Turow Report suggests these would be a legitimate (indeed, the only
legitimate) use of personal data since they are uses under the control of the data subject or their
representative, I will not analyze these and will focus on the other seven.
Each of the permitted purposes, separately and together, serves a compelling market or public
welfare need and can therefore justify the use of personal information without the explicit consent
of the data subjects. Thus, CLEAR limits its service to the per se non-privacy-violation category of data
practices. Some permitted uses involve practices that directly benefit the data subject in a context
where it would be impractical or impossible to get the data subject’s consent. Others involve
practices that are designed to directly protect the interests of somebody other than the data subject
(e.g. the CLEAR client or third parties), but even in these instances, the data subject often benefits
indirectly.
a. For the Direct Benefit of the Data Subject
When a wallet is returned to its owner, for example, the typical owner will not resent the good
Samaritan who looked inside the wallet for the purpose of finding the identity and address of the
owner. Similarly, some uses of the CLEAR platform are performed for the direct assistance and
benefit of the data subject.
i. Use by a Person Holding a Legal or Beneficial Interest Relating to the
Consumer
The first category of permitted use involves instances where CLEAR is used to find data subjects for
their benefit or the benefit of their family member(s).
Some of the uses that fall into this category are high stakes, as when law enforcement or social
services use CLEAR to locate individuals suffering from mental illness or displaced children (TR-
BROOKS00746, 01564). Social services have used CLEAR to identify and locate extended family
members in order to keep children out of the foster care system. (Attachment A at *4; TR-
BROOKS00766)
Others are purely financial, as when social services agencies use CLEAR to find parents or custodial
caretakers who haven’t kept up their address with child support offices so that money can be routed
to them. (TR-BROOKS01996). CLEAR was also used to find correct addresses for 60 million
Medicare and Medicaid recipients who needed new medical ID cards. (TR-BROOKS02053; TR-
BROOKS14319). This is not a trivial matter as the Social Security Administration master file was
known to have inaccuracies in 25-30% of the records. (TR-BROOKS14319). There is no single
good source of address information. For example, one in eight voter registrations are inaccurate. 45
44
Moreover, some of the data will not be retrieved unless users attest that in addition to using the system for one of the
permissible uses, they also satisfy additional prerequisites. For example, voter registration data will not be available unless
the user verifies that they are using it for an election-related or non-commercial purpose. (Attachment A at *3; TR-
BROOKS013538).
45
PEW: ONE IN EIGHT VOTER REGISTRATIONS INACCURATE; 51 MILLION CITIZENS UNREGISTERED (2012).
17
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 19 of 51
The best way to ensure that an important notice and mailing gets to as many intended recipients as
possible is to cross-reference several sources of information, which is of course what CLEAR does
well. Indeed, the Center for Medicare, Medicaid Services (CMS) selected Thomson Reuters for the
contract because in a pilot test, CLEAR was able to find and confirm the right address on a list of
5,000 names and SSNs better than other similar services. (TR-BROOKS148319).
To use a highly relevant hypothetical, how would counsel in a class action case ensure that they have
taken reasonable steps to provide notice of a settlement or judgment to all members of the class?
Finding up-to-date address information is difficult and costly, and a service like CLEAR is a practical
necessity for the proper function of consumer protection and class action work. (Attachment A at
*5).
This category alone illustrates why privacy is not exclusively concerned with the consent and control
of the data subject; there are cases where the use of data is unambiguously in the best interest of the
data subject, but receiving consent is impractical or impossible.
Prof. Turow deliberately disregards actual CLEAR use cases, treating them as nothing more than
trivial exceptions to an otherwise “malefactory” business. (Turow Dep. 258). For example, Prof.
Turow diminishes the benefits and importance of the Veterans Affairs Department using CLEAR to
locate veterans to continue benefits after they move as a “rabbit hole” that has “nothing to do with”
his “essential opinion.” (Turow Dep. 255-56). About the example of CLEAR being used by the U.S.
Department of Health and Human Services to find the correct addresses for replacement Medicare
cards, Turow said “you know you’re making up th[ese] stories of benefits for obvious reasons, but
what I’m saying is the—the cost benefit analysis is in a vacuum here.” (Turow Dep. at 267-68).
When asked about specific, contextualized use cases like these, Prof. Turow thought they were a
distraction from the abstract and unmoored issue of “whether or not CLEAR is . . . giving people
control over their data.” (Turow Dep. 256).
Because Prof. Turow has refused to consider the ways in which CLEAR is actually used, and the
way that those uses may confer benefits to Californians, his opinion cannot inform any
understanding of whether CLEAR in fact causes harm. He has not explained why the CLEAR
product should be regarded as anything other than a collection of beneficial “rabbit holes.”
ii. Administering a Transaction that a Consumer Requests or Authorizes
CLEAR data is allowed to be used when a client is verifying the identity and critical attributes of a
customer who is engaging the CLEAR client’s services. (TR-BROOKS00001168) For example,
when a business wants to secure a loan, the business will want to get through the initial requirements
of verifying the identities and good standing of its principals. (TR-BROOKS272513-20). When data
is used to establish trust that a person seeking a service is who they say they are, the data subjects
benefit by making the process faster and less cumbersome. Id. They also benefit indirectly for the
same reasons that data subjects generally benefit from protecting against fraud, which I discuss in
the next Subpart.
b. For the Protection of Others (and Often for the Indirect Benefit of the Data
Subject)
The Turow Report suggests that CLEAR causes “concrete privacy harms” any time the information
accessed through CLEAR may “create a data-picture of the individual that is not sanctioned by the
person and might well be at odds with the picture the individual has been trying to create.” (Turow
Report at 18). 46 This ignores the fact that the “picture” the data subject has been “trying to create”
46
In deposition testimony, he further offered that any time a person experiences a negative consequence as a result of
information about them being shared with a decisionmaker without their consent, it is a privacy harm and a violation of
18
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 20 of 51
may be deceitful and harmful to others who have more compelling interests in discovering the
misrepresentation than the data subject has in keeping it in place. 47 One of the greatest services that
a platform like CLEAR performs is to protect data subjects indirectly by allowing businesses to
confidently differentiate them from attempted fraudsters and criminals, as the remaining use cases
demonstrate.
i. Complying with Legal Requirements and Legal Process
This subsection addresses two permissible purposes: “Complying with federal, state, or local laws,
rules, and other applicable legal requirements”, and “complying with a properly authorized civil,
criminal, or regulatory investigation, subpoena or summons by Federal, State, or local authority.”
(Attachment A at *3).
Federal and state law often require firms or employers to perform due diligence to detect or avoid
illegal activity.
One well-known example is the anti-money laundering and Know Your Customer requirements in
the U.S. Bank Secrecy Act. Financial institutions are legally obligated to create and maintain effective
due diligence and monitoring systems. 48 Other industries that are not covered by the Bank Secrecy
Act (including the legal profession) are nevertheless under regulatory or tort-based legal obligations
to use reasonable care to guard against the criminal conduct of agents or customers. 49 CLEAR can
also be used to execute particular court orders or judgments. For example, CLEAR has been used to
discover that parents who are delinquent in child support payments have new income streams. (TR-
BROOKS001995).
CLEAR has also been used to perform due diligence on employees or contractors who are in
positions where conflicts of interest or abuses of position present significant risks. For example, one
CLEAR client discovered that a prospective vendor for a children’s hospital had an undisclosed
pedophile conviction. (TR-BROOKS0000876).
This category of permitted uses is not just allowed but required under the law, and so the fact that
CLEAR helps clients meet legal obligations is presumptively highly beneficial.
Data subjects benefit from the higher data quality and more complete records available through
CLEAR. Clients have reported that CLEAR helps reduce the rate of false positives in the course of
money laundering monitoring. (TR-BROOKS0000776). This means fewer bank customers are
suspected of engaging in money laundering and are left undisturbed from credit card freezes, bank
holds, and other consequences of follow-up investigations.
the right to be let alone, even if they are engaged in a criminal act. Turow Dep. 214. For example, he explained that it
might be “repugnant” for a non-profit to use CLEAR to investigate a donor if the non-profit did not provide notice to
the donor. (Turow Dep. at 271-72). He did not address how this could be done without undermining the very purpose
of the search—to detect money laundering, bribery, or other malfeasance.
47
See, e.g., NISSENBAUM, supra note 11 at 178 (describing the employment context. “It is clear why presentation of self
would be important for applicants, but just as clear why companies might resist applicants’ claims trumping their own.”)
48
See OFFICE OF THE COMPTROLLER OF THE CURRENCY, MONEY LAUNDERING, available at
https://www.occ.treas.gov/topics/supervision-and-examination/bank-operations/financial-crime/money-
laundering/index-money-laundering.html.
49
ABA Task Force on Gatekeeper Regulation and the Profession discussed a new ABA Model Rule that would impose
“client due diligence” on the legal profession to avoid indirectly enabling money laundering or illegal tax shelters.
AMERICAN BAR ASSOCIATION STANDING COMMITTEE ON ETHICS AND PROFESSIONAL RESPONSIBILITY, FORMAL
OPINION 491: OBLIGATIONS UNDER RULE 1.2(D) TO AVOID COUNSELING OR ASSISTING IN A CRIME OR FRAUD IN
NON-LITIGATION SETTINGS (2020).
19
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 21 of 51
ii. Protecting Against or Preventing Fraud
If privacy were understood to require absolute control by a data subject, it would be very difficult to
uncover fraud and false impersonations. 50 If a company has poor quality information about business
associates or clients such that it cannot screen out fraudsters, the company will have to assume a
certain amount of loss due to fraud and charge a high enough interest rate to all customers in order
to cover those costs. The inability to screen out the fraudster will cause pure waste—a money
hemorrhage—which will affect the lender and other borrowers alike. In the worst-case scenario, lack
of information about the quality of a borrower can lead to a pernicious cycle—a “market for
lemons”—where those who are willing to pay the higher interest rates are disproportionately more
likely to be fraudsters. This can cause various markets to fail entirely. 51
Thus, in order for certain activities to take place at all or to flourish, companies need to develop a
way to investigate and gauge the veracity of the information that a prospective client or business
partner has supplied. In other words, they need to access personal information about the
prospective client or partner from a source that is not under the client or partner’s control.
CLEAR is used for this very purpose. CLEAR clients have used CLEAR to uncover fraud when
trying to find a defaulted borrower (TR-BROOKS00001168) or when verifying bank account
applicants. (TR-BROOKS00001688) They have used the service to investigate whether borrowers
are using multiple identities. (TR-BROOKS00001992).
The CLEAR clients who improve their detection of fraud are serving their own interests, of course,
but they also serve the interests of their non-fraudulent customers and data subjects who will receive
better terms if fraudsters are found and weeded out. A banking system with well-functioning
verification protocols ensures that customers are able to avoid both street crime (by keeping their
assets in a bank instead of holding them in cash) as well as identity theft. 52 Data subjects also benefit
from fraud prevention when identity thieves are thwarted from illegally securing loans or initiating
contracts in the data subject’s name.
CLEAR is also regularly used by government agencies to prevent fraud-on-the-public. For example,
when the government engages with outside vendors to perform specific tasks, a significant portion is
lost to fraud. During the pandemic, vendor fraud has been a particularly pronounced problem,
costing taxpayers as much as $1 billion by some estimates. 53 CLEAR allows government clients to
find evidence in the personal histories of vendor owners or key employees that the vendor may be
engaged in fraud—for example, if they own multiple cars or properties in different states out of
proportion to the income expected from their line of work, or if they have a history of unpaid debts
or successive bankruptcies. (TR-BROOKS146614). CLEAR clients have also used CLEAR to
screen out fraud and waste in their Medicare/Medicaid systems. (TR-BROOKS00002103).
When these searches are performed, the taxpayers are the beneficiaries because they avoid the waste
that they’d have to eventually pay for when tax evasion or fraud is uncorrected. But it’s important to
50
Richard Posner, The Right of Privacy, 12 GEORGIA L. REV. 393, 401, 403 (“To the extent that people conceal personal
information in order to mislead, the economic case for according legal protection to such information is no better than
that for permitting fraud in the sale of goods.” And “there is a prima facie case for assigning the property right [to
personal data] away from the individual where secrecy would reduce the social product by misleading the people with
whom he deals.”).
51
See George A. Akerlof, The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism, 84 Q. J. ECON. 488 (1970).
52
Nicholas Economides & Przemyslaw Jeziorski, 36 MARKETING SCIENCE 815 (2017) (studying mobile devices that can
verify identity in Tanzania. Mobile payments networks allow for easy deposits and withdrawals and reduce the risk of
robbery).
53
J. DAVID MCSWANE, PANDEMIC, INC: CHASING THE CAPITALISTS AND THIEVES WHO GOT RICH WHILE WE GOT
SICK (2022).
20
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 22 of 51
recognize that data subjects also benefit directly from fraud investigations performed with the help
of CLEAR because those investigations can clear up suspicion that would fester or result in more
intrusive actions in the absence of CLEAR. For example, if a new bank account is flagged for
possible identity theft, CLEAR data can be and has been used to resolve the ambiguity in favor of
the account-holder and ensure that their account isn’t frozen or reported to law enforcement. (TR-
BROOKS00001689) When CLEAR has been used to investigate tax fraud, it has occasionally led to
information that shows the target of the investigation had unclaimed tax benefits that they didn’t
realize they were entitled to, thereby reducing the number of revenue service letters, investigations,
and collection attempts. (TR-BROOKS00001872) It has also allowed CLEAR clients to provide
information to potential vendors to let them know that they need licenses or other requirements so
that they can avoid unintentional non-compliance. (TR-BROOKS00001872). And all of this makes
tax and government benefits compliance work more efficiently, which in turn benefits the public
(taxpayers).
Thus, the prevention of fraud is another context where unconsented access to the sort of personal
data contained in CLEAR is well-justified. Fraud prevention is often thought to be at the core of the
interests that must override interests in control over personal data. 54
iii. Use by a Law Enforcement Agency on a Matter Related to Public
Safety
Crime takes a large toll on U.S. residents, imposing as much as $1 trillion in losses per year. 55 Crime
flourishes in low information environments where the perpetrator’s identity or activity cannot be
observed. 56 A good deal of crime control is preventative, using whatever resources are available to
reduce the opportunity for a criminal to commit an unlawful act without detection and
identification. One such resource is, of course, police patrol. 57 Others are technological—artificial
lighting 58, alarm systems 59, surveillance cameras 60, and mobile phones. 61 All of these cause dramatic
reductions in crime, and this reinforces the important role of discovery of a criminal’s preparatory
activities and attempts.
Recently, private firms and police have begun to use more data-intensive means of illuminating (and
therefore deterring) criminal activity. Digital identification using DNA databases and crime detection
54
For example, the GDPR permits the processing of personal data without consent if it is necessary for the purposes of
preventing fraud. EU GDPR Recital 47. The FTC also recognized that transparency can get in the way of the accuracy
of an independent risk mitigation data system. FTC Data Brokers Report, supra note 25 at vii-ix.
55
Aaron Chalfin, The Economic Costs of Crime, in THE ENCYCLOPEDIA OF CRIME & PUNISHMENT (2015).
56
They are, in other words, crimes of opportunity that exploit low information. Philip J. Cook & John MacDonald,
Public Safety Through Private Action: An Economic Assessment of BIDS, 121 ECON. J. 445 (2011); Eric Piza et al., CCTV
Surveillance for Crime Prevention: A 40-Year Systematic Review with Meta-Analysis, 18 CRIM. & PUB. POLICY 134 (2019).
57
Sherman and Weisburd 1995; Evans and Owens 2006; Mohler et al., Randomized Controlled Field Trials of Predictive
Policing, 110 J. AM. STAT. ASSOC. (2015).
58
Aaron Chalfin et al., Reducing Crime Through Environmental Design: Evidence from a Randomized Experiment of Street Lighting in
New York City, NBER WORKING PAPER NO.25798 (2019)
59
Philip J. Cook & John MacDonald, The Role of Private Action in Controlling Crime, in CONTROLLING CRIME: STRATEGIES
AND TRADEOFFS (Philip J. Cook et al., eds.) (2011).
60
Mikael Priks, The Effects of Surveillance Cameras on Crime: Evidence from the Stockholm Subway, 125 Econ. J. F289 (2015);
Jillian B. Carr & Jennifer L. Doleac, Keep the Kids Inside? Juvenile Curfews and Urban Gun Violence, 100 Rev. Econ. & Stat.
609 (2018).
61
Jonathan Klick et al., Mobile Phones and Crime Deterrence: An Underappreciated Link, in RESEARCH HANDBOOK ON THE
ECONOMICS OF CRIMINAL LAW (Alon Harel & Keith N. Hylton, eds. 2012).
21
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 23 of 51
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 24 of 51
litigation defense. It also includes executing contractual rights, for example, to recover or repossess
collateral and other items if a customer has breached an agreement that establishes the firm’s right to
recovery. CLEAR clients have also used CLEAR to protect their customers and employees during
the COVID-19 pandemic by finding the best contact information for the purpose of contact tracing.
These cases have a lot of conceptual overlap to uses related to fraud (e.g. where a customer has
promised to return collateral in the event of a breach but then skips town) and to uses related to the
direct benefit of the data subject (e.g. where a customer is notified that they may have been exposed
to a virus or other public health risk so that they can take appropriate precautions.) Thus, the
reasoning laid out in the subsections above apply to this category as well.
In summary, the set of limited uses for which the CLEAR platform is made available to clients is
narrowly tailored to the practical interests of society. Prof. Turow believes that these beneficial
examples are exceptions—the rare beneficial use for an otherwise corrupt and exploitative company.
He compared Thomson Reuters to mobsters and Tammany Hall gangsters who should be regarded
as bad even though they built the Brooklyn Bridge. (Turow Dep. 273-74). His Report and deposition
testimony repeatedly suggest that the CLEAR platform collects excessive amounts of data and
makes them available for marketing and nearly any other commercial purpose that CLEAR clients
wish. (Turow Dep. 51 (claiming CLEAR is selling “audiences” to its clients); Turow Dep. 257
(referring to Thomson Reuters as a “malefactor”); Turow Report at 13 (claiming that CLEAR is
accessible for a “variety of uses”)). Given this mistaken belief, Prof. Turow concludes that the
beneficial use cases still cannot justify the platform as a whole. (Turow Dep. 274). But his
understanding of exceptional beneficial use cases against a backdrop of routine privacy-invading
uses is badly mistaken. Based on the evidence referenced above, the beneficial use cases are the
routine use cases. Indeed, they are the only permitted uses.
c. The Individual and Collective Benefits for Californians Are Highly
Dependent on the Particulars of the Data Subject, the Purpose of the Search,
and the Quality of the Data Relative to Other Available Options
Prof. Turow claims that “while the specific content of individual reports may vary, every Californian
whose information is accessible through CLEAR has suffered the same fundamental privacy harm
of a lack of control and a violation of their right to be let alone.” (Turow Report at 13). Prof. Turow
also believes that CLEAR customers would have an option to get the information they need to
complete a socially beneficial project without resorting to the use of a data search product like
CLEAR. (Turow Dep. 262 (“Why couldn’t they go to another place to find those data that is more
legitimate?”); Turow Dep. 267 (“maybe they could find out a better way to get those addresses”).
The implication is that Prof. Turow believes there either is or could be a way to access the right data,
at the right time, with the consent and control of the data subject, in order to accomplish the socially
valuable goals described above.
This is not correct. Some permissible uses of CLEAR performed for the direct benefit of the data
subject have to be done without advanced authorization, such as locating the data subject to deliver a
payment that is owed to them. (See discussion infra, Part V(c)(i)). The premise of the search is that
the data subject cannot be found. For other permissible purposes, data subject control would
undermine the goal of detecting fraud, crime, or other wrongdoing. There is no “better way.” Other
options are either equally lacking in consent, more noisy and error-prone, or both (e.g. interviewing
neighbors or finding other data aggregators like the Google search bar).
The effects on individual data subjects will be mixed: not every data subject will be made better off,
at least not directly. But as long as the purpose of running a search is legitimate (e.g. to prevent
crime or fraud), and as long as on balance the information made available through a platform like
CLEAR is more complete or more accurate than the information sources that would be used in its
absence, there will be more “winners” than “losers”, and society will be well-served.
23
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 25 of 51
To understand why this is so, consider the role of information when consumers make decisions
about which restaurants to patronize. When consumers have greater access to information about
businesses, they are more likely to have a good experience and, thus, will eat out more often. But not
every restaurant will see its patronage increase: information from Yelp reviews and other
independent review systems tends to help smaller or less well-known market participants more, and
chain restaurants therefore lose their relative share of the restaurant market. 65
The same mixed effects occur when businesses judge their potential clients and contractors, too. 66
Imagine how a person who is tasked with vetting new consultants to make sure they are not
fraudsters or embezzlers would approach their job. The vettor would inevitably have some
information—the results of a Google search, or a public records search, for example. This is the
baseline we must judge CLEAR against. We must consider the impact of CLEAR on data subjects
as well as society as compared to a decision that is made with lower-quality information. 67 That is,
what we must not do is fall subject to the Nirvana fallacy where the use of CLEAR is compared to a
situation where the company does not access any information without the data subject’s consent and
still, miraculously, winds up correctly identifying which prospective consultants are going to engage
in criminal activity. Prof. Turow engages in this fallacy when he said “CLEAR is not the be all and
end all of – of the ways to find out [information for socially beneficial purposes]. It just happens to
be used because it exists.” (Turow Dep. 275).
When comparing the effect of the use of CLEAR on a lower-quality source of information, the
marginal effect of CLEAR will cause the vettor’s opinion of the data subject to either improve,
diminish, or stay about the same. Let’s consider how this can affect data subjects and the decision-
makers as a whole by plotting the effects of using CLEAR against a lower-quality alternative source.
It is possible that some data subjects can wrongly be perceived as more risky after a CLEAR search
than they would have been perceived after a search of a different kind, but as long as the overall
accuracy and completeness of the data is better in the CLEAR platform, these instances will be
greatly outnumbered by the instances where a decision-maker makes better judgments. That is, in
most cases, CLEAR will help decision-makers have confidence about innocent clients who would
otherwise look more suspicious with lower quality data. And CLEAR would also help decision-
makers have greater suspicion about the data subjects who are not innocent—who are likely to
engage in fraud or crime.
A visual representation of CLEAR’s impact in a fraud detection system might look like this:
65
Luca (2011). The same is true for hotels. Hollenbeck (2018)
66
A range of social institutions have allowed market transactions to take place even when individuals do not know each
other well enough to have longstanding relationships of trust. Word of mouth, gossip, and status signals help create
enough of an incentive for trade partners to cooperate, and enough of a disincentive against deception. Benjamin Klein
& Keith B. Leffler, The Role of Market Forces in Assuring Contractual Performance, 89 J. POL. ECON. (1981); Hongbin Cai et al.,
Seller Reputation: From Word-of-Mouth to Centralized Feedback, 34 INT’L J. IND. ORG. 51 (2014); Carl Shapiro, Premiums for
High Quality Products as Returns to Reputations, 98 Q. J. ECON. 659 (1983).
67
The Turow Report questions the quality of the data that the CLEAR service makes available. Turow Report at 14
n.40. However, Prof. Turow does not claim that the quality of CLEAR data is inferior to the quality of the data that
would be used by CLEAR clients in the absence of the CLEAR service. My analysis in this Part assumes that data
available on CLEAR is more complete and leads to more accurate assessments of the data subjects. This is a justified
assumption based on the empirical literature, cited earlier in this report, that finds access to more and more varied data
leads to better “matching,” to use the economics term. CLEAR’s business model depends on adding value by improving
accuracy—both by raising red flags that would go unnoticed and by removing unwarranted suspicion—and has selected
its vendors and designed its user interface to do precisely that. I directly address CLEAR’s protocols for maintaining data
quality in Part VI.
24
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 26 of 51
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 27 of 51
whose more complete record of personal details suggest that they pose a greater risk than the vettor
would have assumed with lower quality information, but who in fact do not pose any such risk. This
is genuine system error (assuming the set in group F is not empty). But the losses to this group are
counterbalanced by the gains among data subjects in group D who are correctly perceived to be less
risky when a more complete and accurate set of information is used. So, although some individuals
will be misclassified, fewer will be than would occur under conditions of lower quality data. 69 Thus,
even if Group F is not empty, as long as harm to Group C is substantially smaller than the benefits
to Group F, the social contract approach to privacy would suggest that most people would agree in
advance that use of better, more complete information should be used in these circumstances. And
of course all data subjects and the general public stand to benefit from the increased detection of
fraudsters (group C).
Those who are not well-steeped in the empirical information economics literature might doubt the
size of these arrows. They might believe that more complete information leads to more distrust
overall, and a net loss of matching and opportunities. This isn’t so. The experience with credit
reports shows that lending becomes more common, and at cheaper interest rates, when there is
more information, even though the increase in information includes more data related to late
payments, defaulting, and other seemingly negative attributes. 70 Nor do the type of risk scoring
systems that can be displayed in CLEAR with the user’s input and direction (TR-BROOKS 001314-
21; 272516) lead decision-makers to have heightened suspicion; a 2018 study of the impact of public
safety risk assessment scores in the pre-trial detention decisions in the state of Kentucky found that
after a new law went into effect that required judges to look at an automated risk score, the
percentage of arrestees who were released without bail increased across the board, including for
individuals with relatively high risk scores. 71
The end result is that, system-wide, the impact of CLEAR is positive because it increases accuracy
and efficiency. These cost-savings are ultimately shared by consumers (who don’t have to pay extra
to cover the costs of system-wide fraud) and the public at large (who are better protected from
fraud, money-laundering, theft, and other crimes). For individual data subjects, the benefits or costs
will depend on the particulars of each search context (what the nature of the assessment or vetting
is, what data sources are available, what the next best alternative to CLEAR would be if the client
had to use something else). But even though the impact on individual data subjects varies by person
and search, the effect is positive on net. 72 For at least some purposes, people will generally accept
and even expect that information will be used to capture these social benefits.
69
Avi Goldfarb & Catherine Tucker, Digital Economics, NBER WORKING PAPER NO. 23684 at 11-13 (2017).
70
Prior to the advent of the credit report, some individuals could gain the trust of others by having a family member,
neighbor, or organization vouch for them, or by having a good reputation. These social vouching systems leave much to
be desired-- they used noisy signals of trustworthiness such as whether a creditor’s daughter was chaste. DAVID
GRAEBER, DEBT: THE FIRST 5,000 YEARS 328 (2011). Perhaps because these mechanisms were so noisy and nepotistic,
credit markets for large sums, suitable for investments, were very tight and reserved almost exclusively for business
rather than personal use. Lewis Tappan created the first large scale systematic collection of records for the purposes of
assessing credit-worthiness in 1841 after losing much of his wealth a few years earlier in an economic depression caused
by the over-extension of credit. (Tappan’s large collection of gossip and reputation ledgers were the first information
assets of what would eventually become Dun & Bradstreet). These efforts, which eventually led to the credit reporting
industry, allowed individuals without letters of introduction and other trappings of the upper class to access large loans
and build up personal wealth. Id.
71
Megan Stevenson, Assessing Risk Assessment in Action, 103 MINN. L. REV. 303, 351, 354. Note that the study found that
detainees with high risk scores were, at the same time, somewhat less likely to be released within three days of booking,
suggesting that scores led to individuals in this group being treated with more leniency in some cases, and less in others.
Detainees with low and moderate risk scores were 9% and 7% (respectively) more likely to be released within three days
of booking. Id. at 354.
72
The private sector has spawned and sustained an industry of information intermediaries precisely because the data
collection and organizing services that they offer allow clients to more accurately (and, thus, more fairly and with less
26
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 28 of 51
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 29 of 51
In Rasheed Shabazz’s case, Plaintiffs’ complaint alleges that CLEAR poses risk by retrieving court
records related to a divorce and unpaid child support against a different person who shares
Shabazz’s birthname (Reginald James). (Complaint at 11). While it’s true that CLEAR provides these
court records upon a search of the plaintiff’s birthname and date of birth, the CLEAR report
provides an “Averse Records” risk indicator score of 0—a signal to users that they should not
assume these records actually pertain to the James Reginald who was born in . (Video, TR-
BROOKS010275 at 13:00). In order to even see these records, the user has to click on “Averse
Records” despite the fact that the platform indicates there is a “0.00” risk. (Id.). To know whether
Rasheed Shabazz is likely to be made worse off as a result of the retrieval of these records, we need
to consider what information would be consulted in the absence of CLEAR. Manual public records
or court docket searches could very well result in the same retrieval of records, and without the “0”
adverse records risk score signaling a mismatch. Things could be even worse if a decision-maker
who isn’t able to access CLEAR decides instead to run a Google search on “Rasheed Shabazz.”
They may discover a rape conviction of somebody named
and become concerned that they cannot rule out, without
more information, that this might be the same “Rasheed Shabazz”. Users of CLEAR, on the other
hand, will not even encounter the records for (TR-BROOKS008719-8740;
Video, TR-BROOKS010275).
Cat Brooks’ and Rasheed Shabazz’s CLEAR reports provide information that is more positive and
reassuring than a naïve Google search, public records search, or more limited search for information
would suggest. CLEAR reports show factors like name changes in context, and they do not provide
erroneous information about arrests (Brooks) or convictions (Shabazz) that did not take place but
are publicly reported in Google searches. Although both named plaintiffs have multiple names
associated with their social security numbers—something that tends to heighten concern and lead to
more scrutiny when it comes up in verification and background checks— the fact that both have
changed names is a fact that is likely to be uncovered by entities who are vetting and verifying each
plaintiff whether they use CLEAR or not. By providing a more complete record, CLEAR will tend
to reduce suspicion. Thus, as far as I can tell, neither named plaintiffs’ data has been accessed
through CLEAR in a manner that caused concrete harm to them, and there is reason to believe that
CLEAR can help, and possibly already has helped, the named plaintiffs by steering users of the
CLEAR platform toward an action that is less circumspect and more beneficial.
Thus, while the particular benefits and costs of the CLEAR platform depend on many factors that
will vary search by search, economic theory and the extant empirical evidence suggest CLEAR can
bestow benefits on data subjects and California residents more generally. And data subject control is
neither necessary nor desirable to serve the interests of Californians.
The foregoing analyses show that CLEAR only permits access to personal data when the benefits of
access are particularly strong. CLEAR facilitates some of the most important economic and public
safety functions that people expect and rely on.
VI. CLEAR Has Strong Protections Against Risks of Misuse or Mistaken Associations
The Turow Report opined that there was a lack of protections in place to ensure the reliability
of information made available through CLEAR and to ensure that clients were not using the data
outside the range of permissible purposes. (Turow Report at 5 (referring to “protections in place (or
lack thereof)” of each of these)). Prof. Turow’s analysis of CLEAR’s safeguards is based on his
belief that CLEAR’s protocols, even if highly effective at preventing and detecting access or use of
data that violates the terms of service, would still not reduce the “essential misuse of CLEAR, which
fact against her. Mr. Shabazz has also publicly disclosed his birthname and his decision for changing his name. (TR-
BROOKS008701).
28
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 30 of 51
is that CLEAR, itself, takes data . . . and doesn’t give people control over those data.” (Turow Dep.
305).
To the contrary, because CLEAR has reasonable protocols for credentialing, training, and
monitoring clients that use the system, CLEAR does not create a significant risk of misuse by clients
or unintended data disclosures. And since the permitted uses are highly beneficial, the CLEAR
product as a whole avoids creating significant risks of privacy harm.
a. CLEAR Minimizes the Risk of Misuse Using Credentialing, Training, User
Interface Design, Query Logs, and Audits
The Turow Report states “CLEAR customers must certify that they have a ‘permissible use’ for
CLEAR when they log in. I have seen no evidence that Thomson Reuters investigates that a
customer’s use is, in fact, permissible.” (Turow Report at 13-14). This may give the impression that
it would be easy to use CLEAR for unpermitted purposes and that this may happen with some
frequency. I do not agree; the record provides ample evidence that Thomson Reuters has good
systems for preventing, preempting, and detecting misuse. 75
First, CLEAR has many precautions in place at the front end, when a client begins the credentialing
process. To become a subscriber, a client must first go through a credentialing process in which a
compliance team at Thomson Reuters reviews the application and considers whether the client’s
typical use cases are permissible. (TR-BROOKS013638-40; TR-BROOKS013534-35). They also vet
each client to ensure that newer businesses, or businesses with little public presence or reputation,
do not present a high risk for engaging in abuse. (TR-BROOKS295350). I have also reviewed
records demonstrating that Thomson Reuters salespersons escalate some new clients to the
compliance team if, for example, the type of business of the prospective client is a new type of
customer for CLEAR, if the business is relatively new, or if the business address is in a different
state from where the principals live. (Id.)
This is not a formality. A 2020 compliance report shows that the compliance team did follow-up
investigation requiring discussions with managers and owners of prospective clients and also
conducted independent investigations. (Id.) Prospective clients have been denied because they
intended to use CLEAR for wage garnishment (not a permitted use), because they intended to use
the service to return found or unclaimed money (not a permitted use even though presumptively
helpful to data subjects), because the intended uses were FCRA-related searches related to
employment or lending, because the business was not properly licensed, and because Thomson
Reuters could not validate the business as legitimate, among other reasons. (Id.) Thomson Reuters
declined over in potential contracts during the year on account of concerns that some of
the prospective clients posed a risk of misuse. (Id.; TRBROOKS305116). This represents a
significant portion of potential new business—over of potential contracts—that were declined
in order to maintain TR’s standards for keeping risk to a minimum. (TRBROOKS305116).
Clients must be recredentialed, and employees with access to CLEAR must be re-trained, every two
years at the latest (Attachment A at *5). In addition to agreeing to use CLEAR only for a permissible
purpose, clients must also affirmatively certify that they will not use CLEAR for any FCRA purpose,
or any purpose other than those explicitly permitted. (Attachment A at *5).
75
Turow also claims that a customer may run a “person search” that allows the customer to view “information that
pertains to other individuals unrelated to the target of their investigation (and, accordingly, unrelated to the customer’s
selected permissible use.)” (Turow Report at 14). I disagree that the limited details retrieved related to a data subject’s
neighbors, associates, and family are “unrelated to the target of their investigation” because many of the intended
purposes for CLEAR involve situations were talking to family members, neighbors, or associates can resolve a problem.
For example, if law enforcement are trying to find a missing or kidnapped person, contacting the person’s neighbors and
business associates will certainly be useful and relevant for the investigation.
29
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 31 of 51
CLEAR clients are contractually obligated to refrain from using, copying, or distributing data
accessed through CLEAR in any way other than those described in the licensing agreement. (TR-
BROOKS000482; TR-BROOKS001369) Sharing passwords is prohibited, and subscribers must
ensure that they have taken reasonable precautions to avoid unauthorized access to the CLEAR
system. (TR-BROOKS000482). Specifically, clients must use adequate malware protection software
(TR-BROOKS000482). And they must immediately report to Thomson Reuters if they believe an
employee has misused the platform in violation of the terms of use. (TR-BROOKS013538).
Once a subscriber has access to CLEAR, Thomson Reuters uses reasonable protocols to ensure that
clients have the incentive to comply with their contractual terms and to detect and report instances
of misuse. The format of the query system re-enforces the rules and obligations: except for limited
exceptions where other security is in place, a client first has to get through a two-factor
authentication process to access the query screen. (TR-BROOKS001374). Before entering search
terms, the user must select a permissible purpose (Id.) Outside a limited batch retrieval process, users
can run only one search at a time. Clients are forbidden from using scripts or other automated
searches. (TR-BROOKS001369).
The data that is ultimately retrieved for a search will depend on the permissible purpose. For
example, records covered by the Driver’s Privacy Protection Act (“DPPA”) can be used only for
purposes of fraud prevention, debt recovery, in connection with a legal proceeding or insurance
claim, or for use by a government agency and will not display unless the user selects one of these
purposes. Voter registration records can only be used for election-related or non-commercial
purposes. (Attachment A at *3)
Users understand that queries are logged so that the search and the login credentials of the person
who ran the search can be reviewed retrospectively. (Fox Dep. at 155, 160; TR-BROOKS001378).
The user manual makes this clear, and the settings menu on the CLEAR dashboard also makes this
obvious by allowing users to see their own “History.” (TR-BROOKS001378). Thus, all users have
the information and the motivation to refrain from unpermitted searches.
Thomson Reuters has also created systems to make it likely to detect non-compliance. The licensing
agreements with CLEAR’s data vendors give vendors the right to audit CLEAR’s systems, including
the query logs. (Fox Dep. at 176). The and some states’ Departments of Motor
Vehicles have conducted such audits. (Id.) The licensing agreement explains that Thomson Reuters
has the right to use the logs to audit or monitor the use of their system in order to find violations of
the agreement. (TRBROOKS000482).
Thomson Reuters also conducts its own audits. In 2020, Thomson Reuters initiated a public records
audit program that selected thirty customers each month to respond to a survey designed to elicit
whether CLEAR data were being used in compliance with the agreement terms.
(TRBROOKS304709-20). The goal is “to ensure that our investigative products are used by
LEGITIMATE ENTITIES with a PERMITTED USE CASE with a demonstrable PERMISSIBLE
USE under the applicable regulations.”) (TR-BROOKS 305116). The program ensured that half of
the audit surveys were sent to CLEAR users who may be at higher risk of improper access—i.e.
users who were in collections or repo industries or who worked in settings with less managerial
oversight (e.g. realtors or clients who run a home business). (TR-BROOKS 304709-20). Clients are
obligated under the terms of their licensing agreement to comply with the audit, and those who do
not respond to the audit survey after a second reminder email are informed that their access to
CLEAR will be suspended. (TR-BROOKS304719-20). Thomson Reuters also actively monitors its
query logs for some of the most common misuses of databases: snooping for the details of
celebrities, politicians and other highly visible individuals. (Fox Dep. at 162-63). Thomson Reuters
can and does terminate access upon discovery of a suspected breach of security or breach of the
agreement terms. (TR-BROOKS 00483).
30
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 32 of 51
Finally, clients are required to immediately report to Thomson Reuters if they discover that any
employee has misused CLEAR data. (Attachment A at *1) Administrators of the clients have access
to all queries run by their employees through a service called “QuickView.” (Fox Dep. 163-64; TR-
BROOKS001372). And clients must assist Thomson Reuters in investigations of privacy breaches.
(TR-BROOKS 000483).
The United States District Court for the Southern District of New York and the Second Circuit
Court of Appeals was similarly convinced that Thomson Reuters has adequate protections in place
to prevent misuse, finding that the protections were well designed to prevent this misuse of CLEAR
for (unpermitted) purposes related to credit reporting. Kidd v. Thomson Reuters Corp., 299 F.Supp.3d
400, 407 (S.D. NY. 2017); Kidd v. Thomson Reuters Corp., 925 F.3d 99, 101-02, 107-08 (2021). In
addition to noting the vetting processes I’ve described here, the District Court also explained that
Thomson Reuters opens investigations when it learns of alleged misuse. Between 2012 and 2017,
Thomson Reuters opened forty-six investigations which resulted in ten terminations. Id. at 402.
Based on all of these measures and protections, Turow’s opinion that the CLEAR product has a
“lack” of “protections in place” is unfounded. (Turow Report at 5). To the contrary, CLEAR has
multiple overlapping protections that help prevent or detect misuse.
b. CLEAR Minimizes the Risk of Overcollection and Mistaken Associations
Using Data Quality Procedures, User Interface Design, and Training
TR also reduces risks of unintended harm by limiting data collection and by ensuring that users
understand the limits of the CLEAR platform.
TR only collects data that is likely to be useful and reliable. First, it acquires new data only when
there is a demonstrated need. The impetus for collecting new data from a new or existing vendor
typically starts with client requests. Thomson Reuters staff then conduct an internal analysis to see if
existing data access could meet the needs of their clients. If it does not, then Thomson Reuters
would seek a new data source. (Fox Dep. at 25). The type of information contained in CLEAR
reports are well-matched to their functional purpose and do not contain gratuitous detail. It does not
provide content of communications or detailed geolocation data. The Complaint describes CLEAR
as providing a “360 degree view” of the data subjects, but the specific list of information—"address,
cell phone number, and information about relatives, neighbors and associates” (Complaint at 2)—
actually offer a quite limited “view”, and provide only the sort of data that might be essential for,
e.g., locating the plaintiffs in an emergency. Plaintiffs insinuate that CLEAR provides information
about abortion, desecrating a flag, and homosexual acts (Complaint at 8), but as far as I can tell these
refer only to criminal records. They are items in a long list of specific criminal convictions that may
appear in a data subject’s criminal record. (Plaintiffs’ video of CLEAR search-TR-BROOKS010275;
TR-BROOKS001363 et seq. (user manual)). Even when CLEAR contracts with private sources of
data such as banks that have access to sensitive data, these sources only provide basic account and
header information. Much of the information comes from public records. As a whole, the sensitivity
of the data is similar to, and arguably lower than, the sensitivity of information contained in credit
reports.
Factors considered when acquiring data include not only the quantity or completeness of a data
source (to ensure it will have efficacy to meet the needs of clients), but also the quality of the data.
Accuracy is a significant factor for assessing quality. There are two levels of accuracy to assess: one is
whether the vendors’ records accurately reflect the information contained in their raw sources, and
the other is whether those raw sources accurately reflect facts about the data subjects. (Fox Dep. 25).
CLEAR uses the reputation of the vendor, TR’s experience with the vendor, and the relevance of
the type of data in the vendor’s possession to assess whether to contract with the vendor for new
data. (Fox Dep. at 29-31). One factor in this assessment is the relationship the vendor has to the
31
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 33 of 51
original source of records. (Fox Dep. 157). Thomson Reuters steers away from vendors that hoover
as much data as possible without exercising quality control. (Fox Dep. 30). And conversely these
higher-reputation vendors implicitly find that CLEAR is a responsible data shepherd as none have
invoked privacy as a reason to refuse to license data to Thomson Reuters. (Fox Dep. 121).
Because Thomson Reuters keeps a minimum standard of reliability for both of these layers, it creates
a natural limit to the amount of data that will be collected even when Thomson Reuters is aware that
clients have noticed a gap in records. By maintaining data quality standards as well as standards for
data vendors, Thomson Reuters would not and could not hoover all available data regardless of
accuracy. In fact, CLEAR sometimes improved the accuracy of public records vendors by
triangulating between records and correcting typos and other errors. (Fox Dep. 28). Thus, CLEAR
data is likely to be more accurate than raw public records, which would often become the source of
information for CLEAR’s clients in the absence of services like CLEAR. Thomson Reuters also uses
dynamic procedures to maintain data quality. When CLEAR has discovered an error in an arrest or
criminal record, this prompts Thomson Reuters to “take a close look at our data and see if we can
make any improvements. . . If we see [] a number of them coming out of a particular court, that’s
when we will probably have a conversation with a vendor.” (Fox Dep. at 158). Thomson Reuters
has a procedure that enables clients or data subjects to report discovered errors, and following
investigation, Thomson Reuters will either delete an inaccurate record or sever the connection
between the record and the data subject. (Response to Interrogatories Attachment A at *7). Thus,
there are stop gaps in place if the prospective efforts to avoid errors fail.
These data quality procedures ensure that CLEAR does not over-collect data by including
information that is either flawed, irrelevant, or gratuitous given the limited purposes that the
CLEAR platform allows. 76
CLEAR’s value proposition is not just that it offers more or different data than, say, a public records or
Google search, but that it includes a higher quality set of data that is organized in a manner that
helps users make sense of it. Thus, Thomson Reuters has an obvious business incentive to reduce
error and clutter as much as it can. Thomson Reuters has lost customers on the basis of mis-pinned
information—that is, information that matches part of a search query but actually pertains to
another person. (Fox Dep. 94 (describing what “mis-pinning” is)). In one case, Thomson Reuters
and its former client did not agree on the right way to balance the risk of over-producing records
that are not related to the search target versus the risk of not producing records that actually do
relate to the target. 77 These are difficult business decisions upon which reasonable minds will differ,
and the tension between overinclusion and underinclusion errors is usually best addressed by
collecting more data, not less.
To reduce the risk of harm from error, Thomson Reuters has many protocols to help prevent clients
from misinterpreting CLEAR reports. During the training process, clients are taught that they must
treat CLEAR reports as a retrieval service and come to their own conclusions about the assessment
or meaning. (Fox Dep. 131). Thomson Reuters’ training explains that the data is unverified and that
there will be times when common names or other search parameters are shared by two or more
individuals. Clients are instructed to view the data “critically” and to consider whether the underlying
records are describing the target of their search. (Fox Dep. 171).
76
Thus, based on the evidence I have seen, I disagree with Prof. Turow’s conclusion that Thomson Reuters is
“stockpiling unnecessary information.” Turow Report at 15.
77
As one CLEAR employee put it, “This is one of those items that is a balancing act of being loose enough to bring
back non exact matches but not too loose to where we bring in irrelevant records. We are always evaluating those
processes” TR-BROOKS057007. It should be noted that this former client went to a different service, presumably a
competitor. It did not stop running due diligence searches. Id.
32
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 34 of 51
A user will have to understand that a full name and a state of residence is more likely to pull up
records on one unique person if the name is uncommon, and is at greater risk of pulling multiple
individuals’ data if the name is popular. (Fox Dep. 131-135). The subscriber license uses bold and
capital letters to highlight the portion of the agreement that reminds clients that information
provided may not be suitably accurate or up-to-date for their purposes (TR-BROOKS00483).
The user interface design also guards against overconfidence. CLEAR has a screen in the application
that indicates how current the records that the client is viewing are, and how frequently they are
updated. (Fox Dep. 59). The search results page has a filter button that allows users to filter the
results in various ways (TR-BROOKS001368), and this serves as a visual reminder that the results
can be overinclusive.
As with the assessment of benefits, risks of misuse and misinterpretation should be assessed in
comparison to the sources of information that clients will use in the absence of CLEAR. Prof.
Turow offers no reason to believe that the misinterpretation or mis-association would be worse with
CLEAR than without it. In CLEAR’s absence, clients may instead turn directly to CLEAR’s data
sources or to competitors. Public records are erroneous, and states explicitly deny that they make
any warranties about the accuracy or completeness of the records. (TR-BROOKS000620-21).
For the foregoing reasons, Thomson Reuters adequately manages the risk of misuse, error, and
misinterpretation of the data that the CLEAR platform makes available. Given that it permits the
service to be used only for per-se non-violative purposes, it is unlikely that CLEAR has caused harm
to any Californian. It’s effect on Californians as a class is positive.
VII. Conclusion
To summarize my analyses above:
• Prof. Turow’s theory of privacy harm is that one will occur each and every time a company
collects, shares, or uses information about a person without their consent and control. In fact,
while data subject control is an important factor, privacy is nearly universally understood as
a right and interest that must be analyzed in context.
• Prof. Turow believes his evaluation of privacy does not need to consider the specific
purposes for which CLEAR data is collected or used, nor the specific benefits and harms that may
result from those uses, because all Californians suffer from a loss of control over their data. In fact,
there are several socially valuable and individually beneficial uses of personal data that are
routinely recognized as legitimate and privacy-respecting even though they occur without
consent. These include using information to find the data subject in order to help them, preventing
and detecting fraud, and carrying out active law enforcement and community caretaking activities.
• Prof. Turow believes any beneficial impact certain uses of CLEAR might provide are
outweighed and outnumbered by “malefactory” conduct. In fact, the only purposes for which the
CLEAR platform may be used are those that directly or indirectly benefit data subjects.
• Prof. Turow doubts that the CLEAR platform has adequate security to ensure that the
product isn’t accessed or used outside the limited set of purposes that Thomson Reuters allows. In
fact, Thomson Reuters uses layers of security and care including a credentialing process,
two-factor authentication, contractual obligations of care and reporting, routine auditing,
and redress and removal of clients who breach their agreements.
• Prof. Turow believes each and every Californian has been harmed by the CLEAR platform.
In fact, even the named Plaintiffs have benefited from their data’s inclusion in the CLEAR
33
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 35 of 51
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 36 of 51
Exhibit A:
35
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 37 of 51
Jane Bambauer
520.626.6004 | janebambauer@email.arizona.edu
ACADEMIC POSITIONS
University of Arizona James E. Rogers College of Law, Tucson, AZ.
The Dorothy H. and Lewis Rosenstiel Distinguished Professor of Law (2022 – present)
• Distinguished Legal Scholar (selected by peer committee) (2021-present)
• Deputy Director for Societal Impact, NSF Center for Quantum Networks (2020-present)
• Faculty Director, QuantLaw (2013-present)
• Faculty Co-Director, TechLaw (2019-present)
• Co-founder and Executive Editor, JOURNAL OF FREE SPEECH LAW (2021- present)
Professor of Law (2017-2022)
Associate Professor of Law (2012 - 2017)
Brooklyn Law School, Brooklyn, NY.
Visiting Assistant Professor of Law (2010 - 2012)
EDUCATION
Yale Law School, J.D., 2006.
• Journal of Law and Feminism, Symposium Director and Editor
• Legal Assistance Clinic, Landlord/Tenant Intern at the New Haven Legal Assistance Association
• Legislative Advocacy Clinic, Tax Reform Advocate
Yale University, B.S. in Mathematics, with Distinction, 2002.
• Deforster (2002), Stanley (2001), and Barge-Runck (2000) prizes for mathematics
ARTICLES
Reconsidering Tort Duty When AI Makes Everything Foreseeable (work-in-progress).
Filtered Dragnets and the Anti-Authoritarian Fourth Amendment (work-in-progress).
Reckless Associations, 36 HARVARD JOURNAL OF LAW & TECHNOLOGY __ (forthcoming 2023) (with Saura
Masconale and Simone Sepe).
When a Small Change Makes a Big Difference, 55 U.C. DAVIS LAW REVIEW 2337 (2022) (with Tal Zarsky
& Jonathan Mayer).
From Damage Caps to Decarceration: Extending Tort Law Safeguards to Criminal Sentencing, 102 BOSTON
UNIVERSITY LAW REVIEW 1667 (2021) (with Andrea Roth).
The Quiet Resignation: Why Do So Many Female Lawyers Abandon Their Careers?, 10 U.C. IRVINE LAW
REVIEW 799 (2020) (with Tauhidur Rahman).
The Algorithm Game, 94 NOTRE DAME L. REV. 1 (2018) (with Tal Zarsky).
Snake Oil Speech, 93 WASHINGTON LAW REVIEW 73 (2018).
Information Libertarianism, 105 CALIFORNIA LAW REVIEW 335 (2017) (with Derek Bambauer).
36
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 38 of 51
Other People’s Papers, 94 TEXAS LAW REVIEW 205 (2015).
Outrageous and Irrational, 100 MINNESOTA LAW REVIEW 281 (2015) (with Toni Massaro).
Hassle, 113 MICHIGAN LAW REVIEW 461 (2015).
Is Data Speech?, 66 STANFORD LAW REVIEW 57 (2014).
The New Intrusion, 88 NOTRE DAME LAW REVIEW 205 (2012).
The Secret of My Success: How Status, Prestige, and School Performance Shape Legal Careers, 9 JOURNAL OF
EMPIRICAL LEGAL STUDIES 893 (2012) (peer-reviewed) (with Richard Sander).
Tragedy of the Data Commons, 25 HARVARD JOURNAL OF LAW & TECHNOLOGY 1 (2011) (published as
Jane Yakowitz).
Marooned: An Empirical Investigation of Law School Graduates Who Fail the Bar Exam,
60 JOURNAL OF LEGAL EDUCATION 3 (2010) (peer-reviewed) (published as Jane Yakowitz).
BOOKS
BUBLICK, BAMBAUER & ARELLANO, DOBBS ADVANCED TORTS: ECONOMIC AND DIGNITARY TORTS (2d
Ed.) (2022).
ESSAYS, CHAPTERS, TECHNICAL PAPERS, AND OTHER WORKS
Sober and Self-Guided Newsgathering, invited chapter in MEDIA AS A GOVERNANCE INSTITUTION (Gus
Hurwitz et al., eds., Cambridge University Press, forthcoming 2023).
Conceptions of Fairness in Artificial Intelligence, invited chapter in THE PHILOSOPHICAL FOUNDATIONS OF
IT LAW (David Mangan, ed., Oxford University Press, forthcoming 2022).
Reinventing Cambridge Analytica One Good Intention at a Time, LAWFARE (2022).
Platforms: The First Amendment Misfits, 97 INDIANA LAW JOURNAL 1047 (2022) (with J. Rollins & V.
Yesue).
Reckless Associations: A New Tort for a New Information Ecosystem, LAWFARE (2022).
Facial Recognition as a Least Bad Option, HOOVER INSTITUTE REPORT (2021).
Choosing Freedom, INQUEST (August 19, 2021) (with S. Mayson, A. Roth, and M. Stevenson).
COVID Apps Are Terrible. They Didn’t Have to Be. LAWFARE.COM REPORT (2020) (with Brian Ray).
Plausible Deniability, 2020 PRIVACY IN STATISTICAL DATABASES 91 (2020) (peer-reviewed) (co-authored
with David Sidi).
Privacy Versus Research in Big Data, in THE CAMBRIDGE HANDBOOK OF CONSUMER PRIVACY (Evan
Selinger et al., eds., Cambridge University Press, 2018).
Sensorship, in RON COLLINS & DAVID SKOVER, ROBOTICA: SPEECH RIGHTS & THE NEW TECHNOLOGIES
(Cambridge University Press, 2018).
37
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 39 of 51
The Empirical First Amendment, 78 OHIO ST. L. J. 947 (2018) (symposium issue).
Dr. Robot, 51 U.C. DAVIS L. REV. 383 (2017) (symposium issue).
Identifying and Countering Fake News, University of Arizona Privacy & Free Expression White Paper
(2017) (with M. Verstraete & D. Bambauer) (to be published in ARIZ. J. EMERGING TECH.).
Biometric Privacy Laws: How a Little-Known Illinois Law Made Facebook Illegal, George Mason University
School of Law Program on Economics & Privacy White Paper (2017).
A Bad Education, 2017 U. ILL. L. REV. 109 (2017) (with J. Loe & D. Winkelman).
Glass Half Empty, 64 UCLA L. REV. DISC. 434 (2016) (invited contribution).
Are Privacy Disclosures Informational or Ideological?, 65 DEPAUL L. REV. 503 (2017) (Clifford symposium
issue).
The Relationships Between Speech and Conduct, 49 U.C. DAVIS L. REV. 1941 (2016) (invited response to
Information Fiduciaries and the First Amendment by Jack Balkin).
Assessing Data Intrusion Threats, SCIENCE (letter) (with D. Barth-Jones et al.) (2015).
All Life Is an Experiment. (Sometimes It’s a Controlled Experiment.), 47 LOY. CHI. L. REV. 487 (2015)
(symposium issue).
Freedom From Thought, 65 EMORY L. J. 219 (2015) (symposium issue).
Cyber-Exploitation and Distributed Enforcement, B.U. L. REV. ANNEX (November 6, 2015) (with D.
Bambauer) (invited response to HATE CRIMES IN CYBERSPACE by Danielle Citron).
Collection Anxiety, 99 CORNELL L. REV. ONLINE 195 (2014) (invited response to Digital Innocence by
Joshua Fairfield and Erik Luna).
Fool’s Gold: An Illustrated Critique of Differential Privacy, 16 VAND. J. ENT. & TECH. L. 701 (2014) (with K.
Muralidhar and R. Sarathy).
Vanished: Analyzing Censorship with Google Transparency Data, 18 VA. J. L. & TECH. 137 (2013) (with D.
Bambauer).
Defending the Dog, 91 OR. L. REV. 1203 (2013) (symposium issue).
How the War on Drugs Distorts Privacy Law, 64 STAN. L. REV. ONLINE 131 (2012).
SELECTED PRESENTATIONS, COMMITTEES, AND AWARDS
Board Member, TPRC Conference on Communications, Information, and Internet Policy, American
University Washington College of Law, Washington, D.C., 2017-present.
Program Committee Member, UNESCO Privacy in Statistical Databases, 2016-present.
Debate: Dobbs and the Potential Implications for Data Privacy, The Federalist Society Regulatory
Transparency Project (2022).
38
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 40 of 51
Regular guest, The Cyberlaw Podcast, Steptoe & Johnson, Washington, D.C. (2020-present).
Reporter, The Collection and Use of Personally Identifiable Data, Uniform Law Commission, 2020-
2021.
Debate: Private Control Over Public Discussion, The Federalist Society National Lawyers Convention,
Washington, D.C., 2021 (arguing against Eugene Volokh, Randy Barnett, and Adam Candeub in favor
of Facebook’s and Twitter’s ability to restrict and remove content on their platforms).
Distinguished Legal Scholar, awarded to one University of Arizona law faculty member each year for
exceptional achievement in research (three year term, 2021-2024).
WeHealth, Trading Privacy for Effectiveness in Exposure Notifications: A Debate (with Peter Singer et al.),
2021.
Debate: Today’s Distrust in Big Tech Is Healthy, Silicon Flatirons Topics in Law & Technology Event
(leading the “opposed” team, which won the Oxford-style debate), Boulder, CO., 2021.
Cats, Cars, and Nazis, Freedom of Expression Scholars Conference Works-in-Progress, Yale Law School,
New Haven, CT., 2020.
Privacy & Pandemics: Responsible Uses of Technology & Health Data, Future of Privacy Forum,
September 2020.
Privacy and the Pandemic (invited panelist), Electronic Privacy Information Center, June 2020.
How the Pandemic Will Affect Washington’s Privacy Policy Debate (invited panelist), Technology
Policy Institute, April 2020.
Symposium on Algorithmic Criminal Justice (invited panelist), UCLA College of Law, Los Angeles, CA,
January 2020.
Director, Program on Economics & Privacy, George Mason University Antonin Scalia Law School,
Arlington, VA, 2018-2019.
Panelist, Commercial Speech and the First Amendment, Yale Law School’s Abrams Institute for Freedom of
Speech, New York, NY, 2019.
Measuring Grossly Excessive Punishment, Criminal Law Works-in-Progress Roundtable, Yale Law School,
New Haven, CT., 2019 (invited author).
A Differentially Private Census Bureau?, Population Association of America Annual Conference, Austin,
TX, 2019 (invited panelist).
Testimony, GDPR and CCPA: Opt-Ins, Consumer Control, and the Impact on Competition and Innovation,
Senate Judiciary Committee, Washington, D.C., 2019.
Testimony, The Impact of Privacy Regulations on Competition and Innovation, Federal Trade Commission
Hearings on Competition and Consumer Protection in the 21st Century, Washington, D.C., 2018.
Testimony, Free Speech and Data Privacy, Federal Trade Commission Hearings on Competition and
Consumer Protection in the 21st Century, Washington, D.C., 2018.
39
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 41 of 51
How Speech Libertarianism Serves Progressives, invited response to Michael Seidman, Can Free Speech Be
Progressive?, 118 COL. L. REV. 2219 (2018), Virtual Roundtable hosted by FIRST AMENDMENT WATCH,
2018.
Machine Influencers and Decision Makers, University of Arizona College of Science Public Lecture Series,
Tucson, AZ, 2018 (rebroadcast on AZPM).
Female Labor Participation and the Limits of Law, Governance and the Law, University of Arizona and
World Bank Development Economics Workshop, Tucson, AZ, 2017.
Guest blogger, Volokh Conspiracy, THE WASHINGTON POST, August-September 2017.
The Algorithm Game, Privacy Law Scholars Conference, UC Berkeley Law School, Berkeley, CA, 2017.
The First Amendment’s Application to Rapidly Developing Information Technologies, The William O. Douglas
Annual Lecture, Gonzaga University School of Law, Spokane, WA, 2017.
Google Fellow, Privacy Law Salon, Miami, FL, 2017.
Who Owns Your Data?, Presentation to the Conference of Chief Justices, Scottsdale, AZ, 2017.
Information Libertarianism, Free Expression Scholars Conference, Yale Law School, New Haven, CT,
2016. (plenary session).
Other People’s Papers, University of Haifa Law Department Cyberforum, Haifa, Israel, 2016.
Bad Education, University of Haifa Law Department, Haifa, Israel, 2016.
Privacy Versus Research in the Age of Big Data, University of Utah LABS colloquium, Salt Lake City, UT,
2016.
2015-16 Privacy Fellow, George Mason University Law & Economics Center Research Roundtable
(supporting experimental research on privacy).
Criminal Justice Roundtable, invited commentary on The New Surveillance Discretion: Automated
Suspicion, Big Data and Policing by Elizabeth Joh, Vanderbilt University School of Law, Nashville, TN,
2015.
Harry W. Stonecipher Award for top scholarship in communication and free speech law, Association for
Education in Journalism and Mass Communication Law & Policy Division, 2015.
Program Chair, TPRC Conference on Communications, Information, and Internet Policy, George
Mason University School of Law, Arlington, VA, 2015.
Truth in the Age of Uncertainty, University of Maryland Law School, Baltimore, MD, 2015.
Hassle, Author talk, University of Michigan Law School, Ann Arbor, MI, 2015.
Data, the Police, and the Whole Constitution, University of Texas Faculty Colloquium, Austin, TX, 2015.
Freedom From Knowledge, The 2015 Randolph W. Thrower Symposium, Emory University School of
Law, Atlanta, GA, 2015.
40
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 42 of 51
The Era of Big Data: Balancing Privacy Rights, Law Enforcement Needs, and Corporate Interests, National
Asian-Pacific American Bar Association Annual Conference, Scottsdale, AZ, 2014.
Public Health in the Shadow of the First Amendment, Yale Law School, New Haven, CT, 2014.
Program Vice Chair, TPRC Conference on Communication, Information and Internet Policy, George
Mason University School of Law, Arlington, VA, 2014.
Surveillance in the Technological Age: The Case of the NSA, Keynote Address at the International Criminal
Law Conference, Babes-Bolyai University, Cluf-Napoca, Romania, 2014.
Is Data Speech?, Yale Law School Information Society Project, New Haven, CT, 2014.
Google Fellow, Privacy Law Salon (invitational event for scholars and industry), Miami, FL, 2014.
When Worlds Collide: Technology, Privacy, and Security, Appellate Judges Education Institute, San Diego,
CA, 2013.
The De-Identified Data Commons, United Nations Economic Commission for Europe (Eurostat Work
Session), Ottawa, Canada, 2013.
Into the Breach: Introduction to the 2012-2013 National Health Law Moot Court Problem, 34 J. LEG. MED.
337 (2013).
Problem Author, National Health Law Moot Court, Southern Illinois University, Carbondale, IL, 2012
(fictitious civil rights case based on a data breach at a public health clinic).
Seeding the Data Commons: Legal Safe Harbors for Research Data, Think Tank on the Identifiability of
Genomic Data and Biospecimens, National Institute of Health / National Cancer Institute, Rockville,
MD, 2012.
Invited Debater, Is Privacy Overhyped?, The Open Society Institute, New York, NY, 2012.
Invited Panelist, Balancing Privacy Interests and Innovation, Law and Information Society Symposium,
Fordham Law School, New York, NY, 2012.
Invited Panelist, Bullying Bans and the First Amendment, Seton Hall Law School, Newark, NJ, 2012.
Google Fellow, Privacy Law Salon (invitational event for scholars and industry), Miami, FL, 2012.
Invited Speaker, The Benefits and Risks of De-Identification, The Future of Privacy Forum, Washington,
D.C., 2011.
Tragedy of the Data Commons, The Privacy Working Group Speaker Series, Information Technology and
Innovation Foundation, Washington, DC, 2011.
The Illusory Privacy Problem in Sorrell v. IMS Health, Technology Policy Institute, 2011 (technical
paper) (with Daniel Barth-Jones).
Invited Interviewee, Tragedy of the Data Commons, Surprisingly Free Podcast Series, Mercatus Center,
George Mason University, 2011.
Tragedy of the Data Commons, Google Public Policy Speaker Series, Mountain View, CA, 2011.
41
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 43 of 51
Moderator and Poster Presentation, The Secret of My Success: How Status, Prestige, and School Performance
Shape Legal Careers, Conference on Empirical Legal Studies, Yale Law School, New Haven, CT, 2010.
The Secret of My Success: How Status, Prestige, and School Performance Shape Legal Careers, Law & Society
Association Annual Meeting, Chicago, IL, 2010.
What Are the Effects of Failing the Bar? An Empirical Study, Law & Society Association Annual Meeting,
Denver, CO, 2009.
OP-EDS AND PRESS COVERAGE
Jane Bambauer, Bills Meant to Check Big Tech’s Power Could Lead to More Disinformation, WASHINGTON
POST (June 6, 2022) (with Anupam Chander).
Jane Bambauer, Letting Police Access Google Location Data Can Help Solve Crimes, WASHINGTON POST
(March 28, 2022) (note: I did not choose the headline).
Timothy B. Lee, Why the 2020 Census Has 9 Fake People in a Single House, FULL STACK ECONOMICS
(March 7, 2022).
Michael Finnegan, FBI Says Fortune Seized in Beverly Hills Raid Was Criminals’ Loot. Owners Say: Where’s
the Proof?, L.A. TIMES (Sept. 19, 2021).
Chris Roberts, Marijuana Legalization Is Retiring Police Dogs. Why That’s Good—And Why All Drug K9
Unites Should Go, FORBES (May 30, 2021).
George F. Will, A Cheerleader’s Salty Language Gives the Supreme Court a Chance to Bolster the First
Amendment, WASH. POST (April 26, 2021) (describing amicus brief in Mahonoy Area School Dist. v. B.L.).
Kashmir Hill, Your Face Is Not Your Own, N.Y. TIMES (March 18, 2021).
Alisa Alering, Are You Ready for the Quantum Internet?, SCIENCE NODE (September 21, 2020).
Alexei Barrionuevo, Coronavirus Tracing Apps Are Being Tested Around the World, But Will Concerts Get on
Board?, BILLBOARD.COM (May 28, 2020).
George Will, A Cheerleader’s Salty Language Gives the Supreme Court a Chance to Bolster the First
Amendment, WASHINGTON POST (April 27, 2021).
Jenny McGrath, Inside the Groundbreaking Plan to Truly Anonymize Your 2020 Census Data, DIGITAL
TRENDS (February 28, 2020).
Natasha Singer, The Government Protects Our Food and Cars. Why Not Our Data?, N. Y. TIMES (2019).
Danielle Abril, Facebook’s $5 Billion Privacy Settlement Argued Consumers Weren’t Harmed. Experts Think
the Damage Was Incalculable, FORTUNE (2019).
Timmy Broderick, Privacy, Accuracy, and the Looming 2020 Census, CHRISTIAN SCIENCE MONITOR (2019).
Campus Free Speech, ARIZONA 360 (April 19, 2019) (interview broadcast on Arizona Public Media).
James Langford, #GoodLuck: Odds Stacked Against Devin Nunes in His Lawsuit Against Twitter,
WASHINGTON EXAMINER (March 20, 2019) (reproduced in The Hill).
42
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 44 of 51
Andrew Howard, UA Professor Cautions Senators to Go Slow in Bid to Protect Personal Data, CRONKITE
NEWS ARIZONA PBS (March 12, 2019).
Justin Sherman, How to Regulate the Internet Without Becoming a Dictator, FOREIGN POLICY (quoting Is
Data Speech?).
Ariel Shapiro, Trump’s SCOTUS Nominee Thinks That ISPs Have First Amendment Rights, Which Could
Spell Bad News for Your Privacy, CNBC.COM (2018).
Tim Steller, Cochise Officials Pursue Blogger for Publishing Transcript, ARIZONA DAILY STAR (2017).
Ronald K. Collins, 11 First Amendment Experts Comment on Legality of NYT Release of Trump’s Tax
Returns, CONCURRING OPINIONS (2016).
Jacob Gershman, Free Speech Case About Cash vs. Plastic Store Prices Gests a Supreme Court Review, WALL
ST. J. (2016) (quoting from amici brief for Expressions Hair Design v. Schneiderman).
Stephanie Fairyington, Do You Have a Right to Privacy in Your Marriage?, TIME (2016).
Christopher Slobogin, The Definition of Suspicion in an Era of Modern Policing, JOTWELL (2016)
(reviewing my article, Hassle).
Fernanda Santos, Border Patrol Accused of Profiling and Abuse, N.Y. TIMES (2015).
Connor Harris, Matters of Mismatch: The Debate over Affirmative Action’s Effectiveness, HARVARD
POLITICAL REVIEW (2015) (describing findings from The Secret of My Success).
Kathleen Hale, Glory Johnson Says She Was Victim in Her Altercation with Brittney Griner,
SPORTSILLUSTRATED.COM (2015).
Tim Cushing, The Government’s Antipathy Towards Transparency Has Made FOIA Lawsuits the Default
Process, TECHDIRT (2014).
Driver-Less Cars About a Decade Away, UA Engineer Says, ARIZONA ILLUSTRATED (2014) (television
coverage).
Katherine Jacobsen, Google Challenges Secret Court Over Gag Order, THE CHRISTIAN SCIENCE MONITOR
(2013).
The Bright Side of Telephone Metadata, THE HUFFINGTON POST (2013).
Jerry Brito, What Europe’s ‘Right to Be Forgotten’ Has in Common with SOPA, TIME.COM (2012).
More Bad Ideas From the E.U., FORBES.COM (2012).
Leon Neyfakh, Our Data, Ourselves, THE BOSTON GLOBE (2011) (describing Tragedy of the Data Commons
in depth).
L. Gordon Crovitz, The 0.00002% Privacy Solution, THE WALL STREET JOURNAL (2011) (quoting
Tragedy of the Data Commons).
Ashby Jones, New Study: Forget the Rankings, Just Bring Home Straight A’s, THE WALL STREET JOURNAL
LAW BLOG (2010) (describing findings from The Secret of My Success).
43
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 45 of 51
Terry Baynes, Don’t Sweat Your School’s Status, THE AMERICAN LAWYER (2010) (describing findings
from The Secret of My Success).
Karen Sloan, Those Who Never Pass Bar Exam Initially Lag Far Behind Those Who Do, THE NATIONAL
LAW JOURNAL (2010) (describing findings from Marooned).
Karen Dybis, What Happens When You Fail the Bar Exam?, NATIONAL JURIST (2010) (describing
findings from Marooned).
SELECTED AMICUS BRIEFS AND PRO BONO WORK
ACLU v. Clearview, No.2020-CH-04353 (IL), Brief of Eugene Volokh, Jane Bambauer, and First
Amendment Clinic at Duke Law School in support of defendant Clearview’s motion to dismiss arguing
that the Illinois Biometric Privacy Act as applied to Clearview is an unconstitutional restriction of
speech.
Alabama v. U.S. Department of Commerce, No. 3:21-cv-211-RAH (M.D. Alab.), Brief of Jane
Bambauer as Amicus Curiae in support of Alabama (arguing that the use of Differential Privacy to
protect privacy in Census Data will needlessly degrade the quality of data used for redistricting and for
the enforcement of Voting Rights Act claims).
Hassell v. Bird, No. S235968 (CA), Brief of Amici Curiae First Amendment and Internet Law Scholars
in support of appellant, Yelp, Inc. (arguing that the First Amendment and Section 230 of the
Communications Decency Act prohibit a court from ordering a non-party Internet intermediary to
remove content).
ACLU Foundation of Arizona v. Department of Homeland Security, No. 14-2052 (D. Ariz.), co-
plaintiff with Derek Bambauer and ACLU of Arizona in Freedom of Information Act request for
documents related to Border Patrol operations in Arizona north of the Mexican border.
Expressions Hair Design v. Schneiderman, No. 15-1391 (U.S. Supreme Court), Brief of Amici Curiae
First Amendment Scholars in support of petitioner’s writ of certiori (arguing that credit card surcharge
bans, as applied, are unconstitutional restrictions on commercial speech).
Sorrell v. IMS Health Inc., 564 U.S. 552 (2011), Brief of Dr. Khaled El Emam and Jane Yakowitz as
Amici Curiae in support of IMS Health (arguing that HIPAA-compliant pharmacy data pose little risk
of re-identification to the patients).
EXPERT WITNESS TESTIMONY
Sander v. The Regents of the University of California, Case No. RG18-928614 (Sup. Ct. Cal. 2021)
(deposition testimony).
44
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 46 of 51
Exhibit B:
45
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 47 of 51
Attachment B: Documents & Files Considered
Description / Beg Bates No. End Bates No.
2020-12-03 (Dkt 1) Class Action Complaint n/a
2021-09-10 (Dkt 60) Defendant's Answer and Affirmative n/a
Defenses to Plaintiffs' Class Action Complaint
2022-04-04 Thomson Reuters Responses to Plaintiffs' n/a
First Set of Interrogatories (with Attachment A)
2022-06-01 Expert Report of Professor Joseph Turow n/a
2022-05-16 Steve Fox Deposition Transcript n/a
2022-08-10 Cat Brooks Deposition Transcript n/a
2022-08-18 Rasheed Shabazz Deposition Transcript n/a
2022-08-26 Joseph Turow Deposition Transcript n/a
PLAINTIFFS_005538 PLAINTIFFS_005547
PLAINTIFFS_005548 PLAINTIFFS_005580
PLAINTIFFS_005581 PLAINTIFFS_005607
PLAINTIFFS_005608 PLAINTIFFS_005617
PLAINTIFFS_005618 PLAINTIFFS_005652
PLAINTIFFS_005653 PLAINTIFFS_005661
PLAINTIFFS_005662 PLAINTIFFS_005703
PLAINTIFFS_010275 PLAINTIFFS_010275
PLAINTIFFS_011040 PLAINTIFFS_011125
PLAINTIFFS_011126 PLAINTIFFS_011253
PLAINTIFFS_011254 PLAINTIFFS_011274
PLAINTIFFS_011275 PLAINTIFFS_011312
PLAINTIFFS_011314 PLAINTIFFS_011314
TR-BROOKS000011 TR-BROOKS000028
TR-BROOKS000056 TR-BROOKS000059
TR-BROOKS000075 TR-BROOKS000076
TR-BROOKS000082 TR-BROOKS000086
TR-BROOKS000482 TR-BROOKS000484
TR-BROOKS000610 TR-BROOKS000624
TR-BROOKS000746 TR-BROOKS000749
TR-BROOKS000761 TR-BROOKS000765
TR-BROOKS000766 TR-BROOKS000770
TR-BROOKS000776 TR-BROOKS000780
TR-BROOKS000806 TR-BROOKS000812
TR-BROOKS000841 TR-BROOKS000844
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 48 of 51
Attachment B: Documents & Files Considered
Description / Beg Bates No. End Bates No.
TR-BROOKS000860 TR-BROOKS000864
TR-BROOKS000876 TR-BROOKS000879
TR-BROOKS000896 TR-BROOKS000897
TR-BROOKS001097 TR-BROOKS001099
TR-BROOKS001168 TR-BROOKS001169
TR-BROOKS001261 TR-BROOKS001308
TR-BROOKS001312 TR-BROOKS001327
TR-BROOKS001363 TR-BROOKS001482
TR-BROOKS001564 TR-BROOKS001565
TR-BROOKS001688 TR-BROOKS001691
TR-BROOKS001692 TR-BROOKS001695
TR-BROOKS001816 TR-BROOKS001819
TR-BROOKS001855 TR-BROOKS001858
TR-BROOKS001872 TR-BROOKS001873
TR-BROOKS001985 TR-BROOKS001987
TR-BROOKS001988 TR-BROOKS001990
TR-BROOKS001991 TR-BROOKS001994
TR-BROOKS001995 TR-BROOKS001996
TR-BROOKS001999 TR-BROOKS002002
TR-BROOKS002003 TR-BROOKS002005
TR-BROOKS002051 TR-BROOKS002053
TR-BROOKS002086 TR-BROOKS002093
TR-BROOKS002101 TR-BROOKS002104
TR-BROOKS002239 TR-BROOKS002242
TR-BROOKS002251 TR-BROOKS002252
TR-BROOKS002253 TR-BROOKS002254
TR-BROOKS002292 TR-BROOKS002294
TR-BROOKS003259 TR-BROOKS003261
TR-BROOKS003512 TR-BROOKS003527
TR-BROOKS003528 TR-BROOKS003529
TR-BROOKS003530 TR-BROOKS003532
TR-BROOKS003533 TR-BROOKS003535
TR-BROOKS003536 TR-BROOKS003542
TR-BROOKS003543 TR-BROOKS003543
TR-BROOKS003544 TR-BROOKS003544
TR-BROOKS003545 TR-BROOKS005219
TR-BROOKS005220 TR-BROOKS008254
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 49 of 51
Attachment B: Documents & Files Considered
Description / Beg Bates No. End Bates No.
TR-BROOKS008255 TR-BROOKS008276
TR-BROOKS008277 TR-BROOKS008277
TR-BROOKS008278 TR-BROOKS008278
TR-BROOKS008279 TR-BROOKS008279
TR-BROOKS008280 TR-BROOKS008308
TR-BROOKS008309 TR-BROOKS008318
TR-BROOKS008319 TR-BROOKS008319
TR-BROOKS008320 TR-BROOKS008320
TR-BROOKS008321 TR-BROOKS008321
TR-BROOKS008322 TR-BROOKS008323
TR-BROOKS008324 TR-BROOKS008324
TR-BROOKS008325 TR-BROOKS008325
TR-BROOKS008326 TR-BROOKS008341
TR-BROOKS008342 TR-BROOKS008343
TR-BROOKS008344 TR-BROOKS008350
TR-BROOKS008351 TR-BROOKS008352
TR-BROOKS008353 TR-BROOKS008718
TR-BROOKS008719 TR-BROOKS008740
TR-BROOKS008741 TR-BROOKS008741
TR-BROOKS008742 TR-BROOKS008742
TR-BROOKS008743 TR-BROOKS008743
TR-BROOKS008744 TR-BROOKS008744
TR-BROOKS008745 TR-BROOKS008745
TR-BROOKS008746 TR-BROOKS008746
TR-BROOKS008747 TR-BROOKS008758
TR-BROOKS008759 TR-BROOKS008759
TR-BROOKS008812 TR-BROOKS008812
TR-BROOKS008813 TR-BROOKS008813
TR-BROOKS008814 TR-BROOKS008814
TR-BROOKS008815 TR-BROOKS008815
TR-BROOKS008816 TR-BROOKS008826
TR-BROOKS008827 TR-BROOKS008829
TR-BROOKS008830 TR-BROOKS008830
TR-BROOKS010211 TR-BROOKS010218
TR-BROOKS010219 TR-BROOKS010221
TR-BROOKS010222 TR-BROOKS010222
TR-BROOKS010223 TR-BROOKS010223
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 50 of 51
Attachment B: Documents & Files Considered
Description / Beg Bates No. End Bates No.
TR-BROOKS010224 TR-BROOKS010233
TR-BROOKS010234 TR-BROOKS010234
TR-BROOKS010235 TR-BROOKS010235
TR-BROOKS013534 TR-BROOKS013535
TR-BROOKS013536 TR-BROOKS013538
TR-BROOKS013638 TR-BROOKS013640
TR-BROOKS013720 TR-BROOKS013722
TR-BROOKS018313 TR-BROOKS018313
TR-BROOKS018645 TR-BROOKS018645
TR-BROOKS030530 TR-BROOKS030535
TR-BROOKS035826 TR-BROOKS035829
TR-BROOKS035870 TR-BROOKS035871
TR-BROOKS038817 TR-BROOKS038835
TR-BROOKS039301 TR-BROOKS039307
TR-BROOKS041418 TR-BROOKS041426
TR-BROOKS041734 TR-BROOKS041735
TR-BROOKS044782 TR-BROOKS044782
TR-BROOKS046629 TR-BROOKS046629
TR-BROOKS046643 TR-BROOKS046644
TR-BROOKS046645 TR-BROOKS046646
TR-BROOKS046655 TR-BROOKS046655
TR-BROOKS049485 TR-BROOKS049485
TR-BROOKS049907 TR-BROOKS049907
TR-BROOKS055702 TR-BROOKS055703
TR-BROOKS055720 TR-BROOKS055721
TR-BROOKS055730 TR-BROOKS055732
TR-BROOKS055733 TR-BROOKS055734
TR-BROOKS055737 TR-BROOKS055738
TR-BROOKS055754 TR-BROOKS055755
TR-BROOKS055786 TR-BROOKS055787
TR-BROOKS055803 TR-BROOKS055804
TR-BROOKS055844 TR-BROOKS055845
TR-BROOKS055887 TR-BROOKS055887
TR-BROOKS055888 TR-BROOKS055888
TR-BROOKS057002 TR-BROOKS057018
TR-BROOKS060012 TR-BROOKS060013
TR-BROOKS070585 TR-BROOKS070592
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
Case 3:21-cv-01418-EMC Document 186-3 Filed 03/28/23 Page 51 of 51
Attachment B: Documents & Files Considered
Description / Beg Bates No. End Bates No.
TR-BROOKS127726 TR-BROOKS127847
TR-BROOKS146612 TR-BROOKS146615
TR-BROOKS148319 TR-BROOKS148322
TR-BROOKS165057 TR-BROOKS165058
TR-BROOKS272512 TR-BROOKS272512
TR-BROOKS272513 TR-BROOKS272520
TR-BROOKS295350 TR-BROOKS295350
TR-BROOKS304709 TR-BROOKS304720
TR-BROOKS305113 TR-BROOKS305117
In addition to the cited articles and materials, I considered:
1.
https://www.google.com/search?q=cat+brooks&oq=cat+brooks&aqs=chrome..69i57j46i20i131i263i433i512j35i39j0i
20i263i512j0i512l6.4135j0j7&sourceid=chrome&ie=UTF-8#ip=1
2.
https://www.google.com/search?q=rasheed+shabazz&sxsrf=ALiCzsaJMNorTZlshznsr9YgR5aBS4eV4Q%3A16625
86881434&ei=ARAZY66WGv-
fkPIPqJyNuAk&ved=0ahUKEwjumufj0oP6AhX_D0QIHShOA5cQ4dUDCA8&uact=5&oq=rasheed+shabazz&gs_lcp
=Cgdnd3Mtd2l6EAMyBAgjECcyBAgAEB4yBAgAEB4yBggAEB4QBTIFCAAQhgMyBQgAEIYDOgQIABBDOgQILhB
DOhEILhCABBCxAxCDARDHARDRAzoFCC4QkQI6BQgAEJECOgcIABCxAxBDOggIABCABBCxAzoKCC4QxwE
QrwEQQzoKCAAQsQMQgwEQQzoLCAAQgAQQsQMQiwM6BwgAEEMQiwM6CAgAEIAEEIsDOgoILhCxAxBDEIs
DOg4IABCABBCxAxCDARCLAzoHCC4QsQMQQzoFCAAQgAQ6CAguEIAEELEDOgcIABCABBAKOhAILhCABB
CHAhDHARCvARAUOgsILhCABBCxAxDUAjoFCC4QgAQ6BggAEB4QBzoHCC4QsQMQDToKCC4QsQMQ1AIQ
DToECC4QDToECAAQDToHCCMQsAIQJ0oECEEYAEoECEYYAFAAWOMxYK80aABwAHgAgAHKAYgBlBGSA
QYyLjEzLjGYAQCgAQG4AQLAAQE&sclient=gws-wiz
3. https://oaklandvoices.us/author/rasheed/
4. https://caselaw.findlaw.com/ga-court-of-appeals/1174544.html
5. William L. Prosser, Privacy, 48 CAL. L. REV. 383, 396 (1960).
6. Joseph E. Stiglitz, The Contributions of the Economics of Information to Twentieth Century Economics,
115 Q. J. ECON. 1441 (2000).
7. Jin-Hyuk Kim & Liad Wagman, Screening Incentives and Privacy Protection in Financial Markets:
A Theoretical and Empirical Analysis, 46 RAND J. OF ECON. 1 (2015).
8. Wendy Edelberg, Risk-Based Pricing of Interest Rates for Consumer Loans, 53 J. MONETARY ECON.
2283 (2006).
9. Inmaculada Hernandez & Yuting Zhang, Using Predictive Analytics and Big Data to Optimize
Pharmaceutical Outcomes, 74 AM. J. HEALTH-SYST PHARM. 1494 (2017).
10. Daniel J. Solove & Danielle Keats Citron, Privacy Harms, work in progress available at
https://scholarship.law.gwu.edu/faculty_publications/1534/.
11. CHRISTOPHER SLOBOGIN, PRIVACY AT RISK (2007).
HIGHLY CONFIDENTIAL - ATTORNEYS' EYES ONLY
File and source
- File
- gov.uscourts.cand.374304.186.3.pdf
- Size
- 1,570,369 bytes
- SHA-256
- d47abfbefdbd2007bdbd259e9c0aa59e59de3a930d86fbdbde7bff0e6d56335c
- Original
- PACER (login required)