The Journal of Things We Like (Lots)
Select Page
Salome Viljoen, Democratic Data: A Relational Theory for Data Governance (Nov. 11, 2020), available on SSRN.

Between 2018 and 2020, nine proposals (or discussion drafts) for comprehensive data privacy legislation were introduced in the U.S. Congress. 28 states introduced 42 comprehensive privacy bills during that time. This is on top of the European Union’s General Data Protection Regulation, which took effect in 2018, and the California Consumer Privacy Act, which took effect in 2020. Clearly, U.S. policymakers are eager to be active on privacy.

Are these privacy laws any good? Put differently, are policymakers drafting, debating, and enacting the kind of privacy laws we need to address the problems of informational capitalism? In Democratic Data: A Relational Theory for Data Governance, Salome Viljoen suggests that the answer is no.

Viljoen’s argument is simple. The information industry’s data collection practices are “primarily aimed at deriving population-level insights from data subjects” that are then applied to individuals who share those characteristics in design nudges, behavioral advertising, and political microtargeting, among others. (P. 3.) But privacy laws, both in their traditional form and in these recent proposals, “attempt to reduce legal interests in information to individualist claims subject to individualistic remedies that are structurally incapable of representing this fundamental population-level purpose of data protection.” (P. 3.)

Viljoen could not be more right, both in her diagnosis of current proposals and in their structural mismatch with the privacy, justice, and dignitary interests undermined by data-driven business models that traffic in the commodification of the human experience.

Viljoen first notes that privacy has traditionally been legally conceptualized as an individual right. The Fair Information Practice Principles (FIPPs) and a long series of federal sectoral privacy laws and state statutes grant privacy rights to consumers qua individuals. This new crop of privacy laws is no different. They guarantee rights of access, correction, deletion, and portability, among others. But all of these rights are for the individual consumer. Notice-and-choice, the framework for much of U.S. privacy law, operated the same way: Its consent paradigm centered the right to choose or consent in the individual internet user.

This also tracks the scholarly literature in privacy since 1890. Privacy has long been understood as either a negative—freedom from—or positive—freedom to—right, but almost always a right located in the individual. Modern privacy scholarship has moved away from this model, recognizing both privacy’s social value, its importance in social interaction and image management, and the connection between privacy and social trust. That terrain is well worn; its inclusion here speaks both to Viljoen’s in-depth knowledge of the literature in her field and law review editors’ adherence to a model of overlong “background” sections.

Viljoen’s contributions are not so much her descriptive claim that privacy law has traditionally conceptualized privacy in individualistic terms, but where she goes from there.

Her notion of “data governance’s sociality problem” is compelling. (P. 23.) Viljoen argues that the relationships between individuals and the the information industry can be mapped along two axes: vertical and horizontal. (Pp. 25-27.) The vertical axis is the relationship between us and data collectors. When we agree to Instagram’s terms and conditions and upload a photo of our new dog, we are creating a vertical relationship with Instagram and its parent company, Facebook. The terms of that relationship “structure[] the process whereby data subjects exchange data about themselves for the digital services the data collector provides.”

“Horizontal data relations” are those relations between and among us, data subjects all, who share relevant characteristics. Those who “match” on OKCupid are in a horizontal data relationship with each other. A gay man who “likes” pictures of Corgis is in a horizontal data relationship with those targeted for advertisements based on those latent characteristics. As is a person arrested because a facial recognition tool identified him as a suspect socially connected with the person whose voluntarily uploaded picture of the same tattoo was used to train the facial recognition AI. (P. 26.)

This leads to a critical point. The person who was arrested has a privacy interest in the collection, use, and processing of data about his tattoo. But his interest is independent of the interests of the person who actually uploaded the picture, who started this causal chain of picture, collection, processing, training AI, misidentification, and arrest. It doesn’t matter where the original picture came from. Whoever uploaded it, the victim’s privacy interest is not represented in the vertical data relationship triggered by terms and conditions, a privacy policy, or a picture upload.

Viljoen’s second important contribution flows from the first. She offers a normative diagnosis for why horizontal relationships matter for data governance law. That is, data extraction’s harms stem not only from concerns over my privacy or our visceral reaction to creepy, ubiquitous surveillance. By merely using technologies that track and extract data from us, we become unwitting accomplices in the process through which industry translates our behavior into designs, technologies, and patterns that shape and manipulate everyone else. Abetting this system is a precondition of participation in the information age.

For Viljoen, then, the information economy’s core evil is that it conscripts us all in a project of mass subordination that is (not so incidentally) making a few people very very rich.

This may be Viljoen’s central contribution, and it has already changed my understanding of privacy. Focusing on the individual elides the population-level harms Viljoen highlights. Data flows classify and categorize. Data helps industry develop models to predict and change behavior. And it is precisely this connection between data and the identification of relationships between groups of people that creates economic value. We are deeply enmeshed in perpetuating a vicious cycle that subordinates data subjects while enriching Big Tech. There is no way an individual rights-based regime that gives one person some measure of control over their data can ever address this problem.

And that is, at least in part, where current proposals for comprehensive privacy laws go awry. Although there are some differences at the margins, most proposals are binary: they guarantee individual rights of control and rely on internal compliance structures to manage data collection and use. The rights model, Viljoen shows, inadequately addresses the privacy harms of informational capitalism. So, for that matter, does the compliance model. But that conversation is for another day.

Download PDF
Cite as: Ari Waldman, The Data Economy is Political, JOTWELL (February 12, 2021) (reviewing Salome Viljoen, Democratic Data: A Relational Theory for Data Governance (Nov. 11, 2020), available on SSRN),