Fairfield & Engel 2015 – Privacy as a Public Good¶

Incomplete PDF

The PDF starts at p. 421, in the middle of the article (Duke Law Journal, vol. 65, starting at p. 385). The introduction and Part I (pp. 385–420) are missing. This card covers only Part II, Part III and the conclusion. Definitions and arguments made earlier in the article are not reflected here.

In brief

The authors argue that privacy is a public good in the strict economic sense, and that lack of privacy is a public bad that each user contributes to when disclosing information that also concerns others (p. 423) [U]. Because individual and group incentives diverge, fully informed individual choices tend to reduce overall privacy (p. 423) [U]. Drawing on public-goods experiments, they propose tools that help groups cooperate: repeat play, attention to group characteristics, marginal return, communication, sanction and framing (pp. 435–455) [U]. They conclude that the relevant privacy unit is the group, not the individual, and that tools centred on individual rights of review and deletion have proven largely ineffective (pp. 456–457) [U].

Research question and main claim¶

The question is why consumers say they want privacy but still accept services that cost significant privacy for themselves and others. The authors' answer: users believe they have no choice and no privacy anyway, because of how society as a whole acts (p. 425) [U]. They reject the explanation that consumers are lying about their preferences (p. 425) [U].
Main claim: privacy is a public good as the term is strictly defined in economics, and is subject to social dilemmas (p. 423) [U].
Policy claim: the dominant approach of individual education and empowerment has fallen short and has given rise to privacy nihilism (p. 456) [U]. The more educated and empowered individuals are, the worse the social dilemma can become for the group (p. 456) [U].
They do not claim the public-goods model excludes other models; only that individual-focused approaches have reached diminishing returns (p. 456) [U].
How the research question is stated in the introduction: Not found in the text (pages missing).

Theoretical perspective¶

They place themselves in the tradition of Coase and Ostrom, using economics to find tools groups themselves can use to sustain public goods (p. 421) [U].
They are inspired less by environmental law and more by the results of public-goods experiments (p. 421) [U].
Public-goods theory and the distinction between neoclassical predictions (all cooperation collapses) and behavioural/experimental findings (groups resist social dilemmas) run through the article (pp. 433, 456–457) [U].
Key references from economics: Ledyard on group characteristics (p. 440), Isaac & Walker on group size and marginal return (p. 441), Andreoni on framing (p. 454) [U].
From privacy law they cite Cohen on the socially situated self (p. 423, n. 185) and Solove on privacy as more than an individual harm (p. 432, n. 222; p. 446) [U].

Empirical material and context¶

No original empirical material. The authors draw on findings from experimental economics (pp. 434–455) [U].
Legal context: US case law on data breaches and standing, including Clapper v. Amnesty International (pp. 426–428) and Remijas v. Neiman Marcus (p. 428), and state breach-notification laws (p. 428) [U].
Examples from social networks (Facebook, eBay and others), targeted advertising and price discrimination (pp. 424, 430–431, 448, 452) [U].
Snowden, PRISM and the Verizon order are used as examples of how corporate data can be placed at the government's disposal (pp. 432–433) [U].
Country: mainly the United States. Europe is mentioned once, as giving more attention to privacy as a human right (p. 432) [U].

Method¶

Conceptual and normative argument: they "chiefly suggest normative approaches based on experimental results" (p. 434) [U].
They justify this by saying experimentalists avoid policy conclusions, while lawyers are trained to fit studies to cases (p. 434) [U].
They state that the literature points the way toward possible solutions; it does not lead "ineluctably" to their conclusions (p. 435) [U].
In Part II they also read US case law to show how courts deal with collective data harm (pp. 425–429) [T – summary].

Key concepts¶

Concept	How the source defines or uses it	Page	U/T
public good / public bad	Mathematically identical; only the framing differs. Lack of privacy can be seen as a public bad	423, 454	U
social dilemma	Defection is a dominant strategy for the individual, while cooperation is socially optimal	422	U
free rider	Each individual has an incentive to benefit without contributing	421–422	U
impure public good	Privacy has some private-good attributes and can be modelled as impurely public; approaches a pure public good further out in the network	442–443	U
MPCR (marginal per-capita return)	What the individual gets back from contributing; low MPCR in large networks gives less cooperation	441, 443, 449	U
conditional cooperation	Contributing if one believes others contribute; explained by inequity aversion or reciprocity	444–445	U
repeat play / shadow of the future	Cooperation can be sustained when the end of the interaction is unknown	435–437	U
second-order public good	Monitoring and sanctioning are themselves contributions to a public good	453	U
valence framing / label framing	Presenting the same situation differently, or using cues, changes cooperation	454–455	U
privacy nihilism	The claim that since consumers cannot get privacy, they must not want it	456	U
data pollution	Used about social harms from data that may need handling outside courts	429	U

How privacy and private data are understood¶

As a public good (impure) and lack of privacy as a public bad (pp. 423, 442) [U].
As collective / group: the relevant privacy unit is the group (p. 456) [U]. Disclosure affects friends and family, for example through health information (p. 424) [U].
As social and systemic harm: the law must be able to recognise harms from the collection, aggregation and exploitation of data, not only harm to specific individuals (p. 425) [U]. Increased systemic risk of identity theft is an example (p. 427) [U].
As political: lack of privacy also harms the body politic, and privacy matters for democratic processes and independent decision-making (p. 432) [U].
Privacy as a human right is mentioned, with the note that this perspective has had more attention in Europe than in the United States (p. 432) [U].
Personal data are described as something users "pay" with to access platforms (p. 423) [U].

Arguments and findings¶

The user's expected individual damage is too small to outweigh the benefit; summing everyone's damage would show the social balance is negative (p. 424) [U].
The public bad grows over time with increased storage and analysis capacity (p. 424) [U].
Courts tend to focus on specific harm to specific individuals, not on losses to social welfare (p. 425) [U]. The trend is shifting, for example in Remijas (pp. 426–428) [U].
Law has lagged science before, as with tobacco and asbestos; distributed data harms may in the same way become legally cognizable over time (p. 429) [U].
Behavioural models allow companies to extract almost all consumer surplus through price and product differentiation; this may be a social loss, not only a distributional issue (pp. 429–432) [U].
Pervasive government surveillance has a chilling effect; citizens self-censor online (p. 433) [U].
Group size alone does not prevent cooperation if the marginal return is high enough (pp. 441–442, 444) [U].
Small, tightly knit groups are the best place to leverage marginal return; tools to protect friends' privacy help such groups (p. 443) [U].
Inequity aversion may explain the strength of the nothing-to-hide fallacy: people may assume bad actors benefit more from privacy than they do (pp. 446–447) [U].
The current approach of teaching in schools and advertising that no one can be trusted online may undermine reciprocity-based cooperation (p. 448) [U].
Even small, incremental increases in the payoff from privacy-seeking behaviour (per-permission control in apps, enforcing do-not-track) can help (pp. 449–450) [U].
Pseudonymity allows reputation and communication without revealing real-world identity (p. 452) [U].
Framing privacy as a public good may give more cooperation than framing it as a public bad (pp. 454–455) [U]. If society keeps debating privacy in purely individualistic terms, individuals will be blamed when they fail (p. 456) [U].

Limitations the source itself mentions¶

Experiments necessarily leave factors out; this is what experiments must do (p. 434) [U].
The economic and experimental literatures do not lead inevitably to their conclusions (p. 435) [U].
Their list of group characteristics is not exhaustive (p. 440) [U].
Whether companies would have incentives to build the tools is left open; they do not argue who should make them (p. 440, n. 249) [U].
The motives behind conditional cooperation are not settled (p. 445) [U]. The experiments are one-shot games, while the field offers more options (p. 445) [U].
Reputation systems among friends themselves create new privacy challenges (p. 448, n. 279) [U].
They do not claim to have identified the best legal-policy solutions (p. 457) [U].

What the source does not cover¶

The state as an actor that collects and uses private data in its own administration. The state appears mainly as a recipient of corporate data (Snowden) and as a legislator and court system [T].
Norway or Nordic conditions [T].
Specific European data protection law. Europe is mentioned only in passing (p. 432) [T].
Pages 385–420 have not been read, so it is unknown what the introduction and Part I contain [T].

Possible relevance for the thesis¶

A clear example of an understanding of privacy as collective, contrasted with individual-rights approaches. Can be used as a lens for asking whether government documents frame privacy as an individual or a collective concern [T].
The point that how society talks about privacy affects behaviour (p. 455) may connect to a document-oriented approach, where the wording of documents matters [T]. The authors themselves say nothing about government documents [T].
The critique of individual rights of review and deletion (p. 457) can be compared with which rights and tools Norwegian documents emphasise [T].
The US context and economic vocabulary mean the transfer to Norway must be argued, not assumed [T].

Connections to other sources¶

Solove 2011 – Fairfield & Engel cite Solove's 2011 essay Why Privacy Matters Even If You Have Nothing to Hide (Chronicle of Higher Education) on the nothing-to-hide argument (p. 446, n. 276) [U]. Whether this is the same text as Solove 2011 in the source list must be checked; the list entry is a book [T].
Bauman et al. 2014 and Gros et al. 2017 – Fairfield & Engel use the Snowden revelations as an example (p. 432); thematic connection [T].
Bear & Mathur 2015 – both use "public good", but Fairfield & Engel use it in the strict economic sense (p. 423). Whether Bear & Mathur use it the same way must be checked on their card [T].
Fourcade 2017 – thematic: data, behavioural models and value extraction from consumers (pp. 429–431) [T].

Key quotes¶

“Indeed, we contend that privacy is a public good as that term is strictly defined in the economics literature.” (p. 423)

“One way to perceive the problem clearly is to consider lack of privacy as a public bad, to which we all contribute when we post information about ourselves that generates negative spillover effects.” (p. 423)

“Lack of privacy also harms the body politic, and therefore some remedial efforts must be political.” (p. 432)

“The relevant privacy unit is the group, rather than the individual.” (p. 456)

“Tools should not be centered on individual rights of review and deletion, which have proven largely ineffective.” (p. 457)

Open questions for Rebekka¶

The PDF lacks pp. 385–420. Do you have the full article, or do you use only Part II onwards?
Do Norwegian government documents treat privacy mainly as an individual right, or also as something collective or social? Where do you see it?
The authors argue in terms of economics and US law. Is the public-goods vocabulary useful for you, or mainly as a contrast?
Is the state in your documents a protector of privacy, a user of private data, or both? Fairfield & Engel say little about the state's own data use.
The authors say the way we talk about privacy shapes behaviour. Is this a point you want to connect to how documents enact privacy?