Privacy Research Group

The Privacy Research Group is a weekly meeting of students, professors, and industry professionals who are passionate about exploring, protecting, and understanding privacy in the digital age.

Joining PRG:

Because we deal with early-stage work in progress, attendance at meetings of the Privacy Research Group is generally limited to researchers and students who can commit to ongoing participation in the group. To discuss joining the group, please contact Tom McBrien. If you are interested in these topics, but cannot commit to ongoing participation in PRG, you may wish to join the PRG-All mailing list.
PRG Student Fellows—Student members of PRG have the opportunity to become Student Fellows. Student Fellows help bring the exciting developments and ideas of the Research Group to the outside world. The primary Student Fellow responsibility is to maintain an active web presence through the ILI student blog, reporting on current events and developments in the privacy field and bringing the world of privacy research to a broader audience. Fellows also have the opportunity to help promote and execute exciting events and colloquia, and even present to the Privacy Research Group. Student Fellow responsibilities are a manageable and enjoyable addition to the regular meeting attendance required of all PRG members. The Student Fellow position is the first step for NYU students into the world of privacy research. Interested students should email Student Fellow Coordinator Tom McBrien with a brief (1-2 paragraph) statement of interest or for more information.

PRG Calendar:


Fall 2019 [12:45-2:00pm, Furman Hall, 245 Sullivan Street, Room 120]

December 4: Albert Fox-Cahn
November 20: Jina Huh-Yoo
November 13: Mark Verstraete & Tal Zarsky
November 6:
October 30: Tomer Kenneth
October 23:
October 16: Salome Viljoen
October 9: Katja Langenbucher
October 2: Michal Shur-Ofry
September 25:Mark Verstraete
September 18: Gabe Nicholas & Michael Weinberg — Data, To Go: Privacy and Competition in Data Portability

     ABSTRACT: Today’s major social media platforms face serious scrutiny around privacy and anti-competitive behavior. Data portability, the principle that users should be able to move their data from one service to another, has been hailed as a way to improve competition without compromising privacy. As the logic goes, portability offers users ownership over their data by allowing them to download it, and at the same time, offers competitors opportunities to innovate, by letting them build and grow new products with uploaded incumbent data. This presentation will argue that, in the context of social network data, these dual goals of privacy and competitiveness from data portability are incompatible. When incumbents choose which data to make available in a portability regime, decisions that benefit the privacy of uploaders and third-party users harm the utility to competitors, and visa versa. There are at least three ways competitors might use social network data to create new products: to seed new profiles on a competing platform, to offer insights and novel applications through machine learning, or to recreate features from the incumbent platform and allow users to migrate over. This presentation will consider the privacy/competition trade-off for each of these three scenarios by looking at real data from Facebook’s portability platform, Download Your Information. In all three, the data Facebook makes available is likely insufficient to bring about meaningful competition. This could be improved by allowing users to export their social graphs, adding globally unique identifiers, or increasing the contextual data made available. However, these changes would compromise the privacy of uploading users and their connections, even ones who did not upload their own data to the new platform. Regulators who consider incorporating data portability should be specific with their goals, and in the case of social networks, choose between encouraging private data ownership and competition.

September 11: Ari Waldman — Privacy, Discourse, and Power

     ABSTRACT: This project is about the discourses of privacy and privacy law. It constructs the landscape of privacy discourse, where it has been, where it is going, and who it empowers along the way. The dominant discourse of privacy today, often called "notice and consent", is explicitly neoliberal. This regime has been roundly criticized by privacy scholars as a failure. And yet, for all its faults, notice-and-consent always made sense from a sociological or phenomenological perspective. That is, it was inadequate yet scrutable; because of the latter, we determined the former. Neoliberal privacy law is ineffective, but it was always accessible and open for interrogation from the ground up. That inadequate, yet relatable discourse, however, is now losing ground to the inscrutable, unaccountable discourse of technology designers. I argue that the same neoliberal social, political, and legal forces, superpowered by more advanced technology and a more powerful technologist profession, are shifting privacy law discourse from accessible concepts like choice to inaccessible computer code, from something regulators could interrogate to the “black box”  language of technology. The discourse of privacy law and, thus, power over its translation into practice, resides in the design team, where engineers, supervised by other engineers, make consequential choices about how, if at all, to interpret the requirements of privacy law and integrate them into the code of technologies they create. Based on primary source research, this project argues that the code-based discourse of engineers is gaining hegemonic power in privacy law, thereby defining privacy law and what it means in practice, stacking the deck against robust privacy protections, and undermining the promise of privacy laws already passed.

Spring 2019

May 1:

April 24: Sheila Marie Cruz-Rodriguez — Contractual Approach to Privacy Protection in Urban Data Collection
     ABSTRACT: Data collection and data analysis methods are involved in a continuous iterative process, as a result, the tools we use to safeguard certain social interests may lag when compared against technology’s rapid pace. This article is written under the assumption that a sensor program is to be deployed in a jurisdiction[1] that has recognized a right to locational privacy, making the collection of certain urban data unlawful absent contractual agreement between the parties or a specific regulation exempting such collection from a consent requirement. This analysis only includes data collection by private parties for commercial purposes, issues arising from government data collection practices are outside the scope of this article. The article considers the effectiveness of individual and community methods for solving the issues arising from urban data collection. In the first section, I evaluate an individual method by examining whether the contractual approach to urban data collection effectively safeguards consumer privacy interests. This is achieved by examining the notice and consent elements of contracts. Next, I analyze the challenges facing each element of the contractual approach along with its exceptions as recognized in the relevant literature (de-identification and anonymization) in order to conclude that in my view they are inadequate approaches to protect consumer privacy. The second section considers the community method under three main approaches. First, the implementation of a consent board who would be a party to the data collection contract and will analyze the relevant terms in order to either supply or withhold consent for the collection of urban data. This approach would be a community solution that aims to solve several issues arising from the problems related to notice and consent. Next, a Civic Data Trust that would manage the requirements for collection and access to the data itself. The Data Trust would be a local entity that would have control over the urban data. Lastly, a private non-profit entity authorized by the legislative power, to develop and enforce rules regarding urban data collection, use, and retention practices. Finally, the article discusses whether the challenges identified in the previous sections have been addressed by the alternatives discussed and identify areas of further research.

[1] As an example, In R. v. Jarvis, 2019 SCC 10, the Supreme Court of Canada ruled on February 14, 2019 that individuals are entitled to a reasonable expectation of privacy in public spaces. This decision has set as a precedent that a person’s reasonable expectation of privacy can no longer be purely based on one’s location, but instead in a “totality of circumstances”.

April 17: Andrew Selbst — Negligence and AI's Human Users
     ABSTRACT: Negligence law is often asked to adapt to new technologies. So it is with artificial intelligence (AI). But AI is different. Drawing on examples in medicine, financial advice, data security, and driving in semi-autonomous vehicles, this Article argues that AI poses serious challenges for negligence law. By inserting a layer of inscrutable, unintuitive, and statistically-derived code in between a human decisionmaker and the consequences of that decision, AI disrupts our typical understanding of responsibility for choices gone wrong. The Article argues that AI’s unique nature introduces four complications into negligence: 1) unforeseeability of specific errors that AI will make; 2) capacity limitations when humans interact with AI; 3) introducing AI-specific software vulnerabilities into decisions not previously mediated by software; and 4) distributional concerns based on AI’s statistical nature and potential for bias.Tort scholars have mostly overlooked these challenges. This is understandable because they have been focused on autonomous robots, especially autonomous vehicles, which can easily kill, maim, or injure people. But this focus has neglected to consider the full range of what AI is. Outside of robots, AI technologies are not autonomous. Rather, they are primarily decision-assistance tools that aim to improve on the inefficiency, arbitrariness, and bias of human decisions. By focusing on a technology that eliminates users, tort scholars have concerned themselves with product liability and innovation, and as a result, have missed the implications for negligence law, the governing regime when harm comes from users of AI. The Article also situates these observations in broader themes of negligence law: the relationship between bounded rationality and foreseeability, the need to update reasonableness conceptions based on new technology, and the difficulties of merging statistical facts with individual determinations, such as fault. This analysis suggests that though there might be a way to create systems of regulatory support to allow negligence law to operate as intended, an approach to oversight that it not based in individual fault is likely to be a more fruitful approach.

April 10: Sun Ping — Beyond Security: What Kind of Data Protection Law Should China Make?
      ABSTRACT: In September 2018, the draft of Personal Information Protection Law was firstly listed as one of the Class I items in the 5-year Legislative Plan of 13th Standing Committee of the National People’s Congress (NPCSC), which means China will finally make an official draft of DPL in near future. Actually, as early as 2003, with the rapid development on E-government, Chinese government has begun to try to make a DPL, but 5 years later, the attempt was failed without any result, except for an expert’s proposal draft with European style. It is from around 2010 that security in cyberspace gradually became the dominant concern in public and political opinions which provided the new incentive for Chinese DPL legislation. From then on, a whole new legislative mode, security-driven mode, appears through the articles and sections related to data protection scattered in criminal law, civil law and administrative law. However, the legislation under the security-driven mode only covers limited private sector, targets merely on data breach and illegal obtainment, lacks non-partial enforcing mechanism, and hardly balances different values. Meanwhile, with incredible technology development and application, the initiatives of personal data collecting and using in public and private sectors have become more and more ambitious without comprehensive legal frame of data protection. More importantly, because China is still struggling in transition, the current data protection laws have no roots in constitutional values, and hence, no compatible and profound theories. The incoming draft of DPA is an opportunity for China to re-comprehend the practice (Datalization, Centralization and Preconditionalization), structure their own theory (Marxism and informational person(human)), draw on experiences and lessons from Europe and the U.S., and mitigate the risk of international misunderstanding and distrust.

April 3: Moran Yemini — Missing in "State Action": Toward a Pluralist Conception of the First Amendment
     ABSTRACT: Online speech intermediaries, particularly social platforms, have an enormous impact on internet users' freedom of expression. They determine the speech rules for most of the content generated and information exchanged today, and routinely interfere with users' speech, while enjoying practically unchecked power to block, filter, censor, manipulate, and surveil. Accordingly, our current system of free expression lacks one of the main requirements of a just system - the notion that no form of power is immune from the question of legitimacy. Scholarly responses to this situation tend to assign decreased weight to constitutional norms as means to impose duties on online intermediaries and promote internet users' speech, while focusing, instead, on other means, such as non-legal norms, legislative and administrative regulation, and technological design. This Article will swim against this current, arguing that a speech promoting environment cannot be sustained without an effective constitutional check on online intermediaries' exercise of powerUnfortunately, existing First Amendment doctrine poses high barriers for structural reform in the existing power relations between online intermediaries and their end users: (1) the "state action" doctrine prevents users from raising speech-related claims against online intermediaries; and (2) an expansive interpretation of what constitutes "speech" serves as a Lochnerian vehicle for intermediaries to claim immunity from government regulation. This Article will discuss these doctrinal barriers, as well as possible modifications to existing doctrine, which could create an environment more supportive of users' speech. However, more importantly, the main contribution of this Article to existing scholarship centers on the argument that a deeper reassessment of traditional doctrinal assumptions is required in order for the First Amendment to fulfill its speech-protecting role in the digital age. The underlying premise of traditional thinking about speech-related constitutional conflicts conceptualizes such conflicts as necessarily bipolar, speaker-government, equations. Accordingly, courts and scholars ordinarily focus on asking whether "the state" is present on one side of the equation or whether "a speaker" exists on the other. This way of thinking about speech-related conflicts suffers from grave limitations when trying to cope with the realities of networks comprised of multiple speakers and multiple censors/regulators (with potential overlaps between these categories). The bipolar conception of the First Amendment is simply incompatible with the type of conflicts that pluralist networks generate. Consequently, if the First Amendment is to have a significant speech-protective meaning in the digital ecosystem, a more sophisticated analysis than the reigning bipolar conception of the First Amendment is necessary. This Article will propose such an alternative analysis, which shall be denominated a pluralist conception of the First Amendment.

March 27: Nick Vincent — Privacy and the Human Microbiome
     ABSTRACT: By even the most modest estimates, an individual human being may harbor over 30 trillion bacterial cells within and on her body. These bacterial communities thrive on our skin and hair, in our mouths, and throughout our digestive tracts, and they have the ability to carry and communicate far more information about us than we may at first realize. The composition of these communities is affected by what we eat, where we live, medications we have taken, and even the status of our overall health. This talk will address the burgeoning interest in using bacterial communities and the human microbiome in the spheres of forensic science and, more pertinent to our goals, privacy, before diving into an open discussion on the promises, drawbacks, and challenges of this technology. Is the thought of identifying the presence of a murderer at a crime scene, or using targeted advertising based off the bacterial communities in and around us the stuff of science fiction? Or is it just around the corner?

March 13: Nick Mendez — Will You Be Seeing Me in Court? Risk of Future Harm, and Article III Standing After a Data Breach
     ABSTRACT: We examine the circuit split that has developed in the U.S. federal court system regarding an individual's ability to sue in federal court after their data has been compromised through a data breach or leak but has not been exploited in any other way. Following the U.S. Supreme Court's decision in Spokeo v. Robins, a plaintiff must show concrete and particularized harm to maintain standing to sue in federal court. Under Clapper, this may include imminent harm, including a substantial risk that the harm will occur. Four circuit courts (1st; 2d; 4th; 8th) have held that the mere exposure of data is insufficient to sustain an assertion that there is a substantial risk of harm, meaning therefore that the data breach victim has not experienced an injury in fact, and lacks standing to sue. Five circuits (DC; 3d; 6th; 7th; 9th) have held the opposite: that data in the hands of a hacker creates a substantial risk of harm that is sufficient to constitute an injury in fact, conferring standing on a data breach victim. Currently before the Court is a petition for certiorari on the 9th Cir. case In re Zappos that would potentially resolve the issue. We weigh the costs and benefits of each theory, and assess its applicability to other areas of privacy law. Attached as background reading are the cert petition and brief in opposition from Zappos.

March 6: Jake Goldenfein — Through the Handoff Lens: Are Autonomous Vehicles No-Win for Users
     ABSTRACT: There is a great deal of hype around the potential social benefits of autonomous vehicles. Rather than simply challenge this rhetoric as not reflecting reality, or as promoting certain political agendas, we try to expose how the transport models described by tech companies, car manufacturers, and researchers each generate different political and ethical consequences for users. The paper explores three archetypes of autonomous vehicles - fully driverless cars, advanced driver assist systems, and connected cars. Within each archetype we describe how the components and actors of driving systems might be redistributed and transformed, how those transformations are embedded in the human-machine interfaces of the vehicle, and how that interface is largely determinative of the political and value propositions of autonomous vehicles for human ‘users’ – particularly with respect to privacy, autonomy, and responsibility. To that end, this paper introduces the analytical lens of ‘handoff’ for understanding the ramifications of the different configurations of actors and components associated with different models of autonomous vehicle futures. ‘Handoff’ is an approach to tracking societal values in sociotechnical systems. It exposes what is at stake in transitions of control between different components and actors in a system, i.e. human, regulatory, mechanical or computational. The handoff analytic tracks the reconfigurations and reorientations of actors and components when a system transitions for the sake of ‘functional equivalence’ (for instance through automation). The claim is that, in handing off certain functions to mechanical or computational systems like giving control of a vehicle to an automated system, the identity and operation of actor-components in a system change, and those changes have ethical and political consequences. Thinking through the handoff lens allows us to track those consequences and understand what is at stake for human values in the visions, claims and rhetoric around autonomous vehicles.

February 27: Cathy Dwyer — Applying the Contextual Integrity Framework to Cambride Analytica
     ABSTRACT: A continuing challenge for the Digital Privacy research community has been the identification and articulation of norms regarding information flow. The Contextual Integrity Framework (CI) gives us a tool to discover these norms. An application of CI to 2,011 news articles about the Cambridge Analytica revelations of 2018 found evidence that mini-contexts could be useful in identifying norms regarding information flow that are fast-evolving and rarely articulated. These mini-contexts emerge in circumstances where norms and transmission principles are clearly expressed. I define a mini-context as a narrowly specified situation where information flow is clearly articulated and described. I call them mini-contexts, because they have a much reduced scope of relevance than broader contexts, such as those associated with the flow of medical or educational information. The political consulting firm Cambridge Analytica carried out extensive collection of personal data from Facebook, and used that data to develop predictive models of individuals in order to target political advertising during the 2016 ‘Brexit’ vote, and the US presidential election that same year. The result of this revelation was an international uproar, and spurred a global discussion about the use of personal information. From the Cambridge Analytica case, I will discuss two mini-contexts. The first is the “I Accept” mini-context. A recurring topic in the discussion around Cambridge Analytica is the degree to which individual users are themselves responsible for giving up their privacy. This context describes what information can be shared when a person clicks ‘I Accept,’ without reading the policy or terms. The second mini-context centers around the evaluation of personality types using data from online behavior. In this mini-context, there is evidence that the application of personality evaluations without explicit consent is a violation of appropriate information flows.


February 20: Ignacio Cofone & Katherine Strandburg — Strategic Games and Algorithmic Transparency
      ABSTRACT: We challenge a commonly held belief in the industry, government, and academia: that algorithmic decision-making processes are better kept opaque or secret because otherwise people may “game the system,” leading to inaccurate or unfair results. We first show that the situations in which people can game the system, even with all information about the decision-process, are narrow, and we suggest how to identify such situations. This depends on the proxies used: how easy they are to fake, and whether such “faking” changes the underlying feature that is measured. We then develop normative considerations to determine, from the sub-set of situations where decision-subjects can effectively game, when does this gaming justify opacity, and when should transparency be mandated irrespective of gaming. This should depend on the social costs of false positives and false negatives and the accuracy of proxies. Proxies with very high or very low false positives and false negatives should be disclosed, and whether proxies with high false positives and low false negatives or the converse should be disclosed depends on the relative social costs of each. In such way, we show that the situations in which algorithmic secrecy is justified are much narrower than is normally assumed, and we thereby hope to advance the discussion on algorithmic transparency.

February 13: Yan Shvartshnaider — Going Against the (Appropriate) Flow: A Contextual Integrity Approach to Privacy Policy Analysis
     ABSTRACT: We present a method for annotating and analyzing privacy policies that is theoretically grounded in the framework of contextual integrity (CI). Unlike previous annotation techniques, CI helps formalize and detect issues with descriptions of data collection and exchange. We demonstrate this method with a case study comparing Facebook’s privacy policy before and after the Cambridge Analytica scandal. Surprisingly, our analysis shows that despite providing additional details on information handling practices, the updated policy suffers from a substantial increase in ambiguous statements due to missing or excessive contextual details, such as what information is being transferred, from whom, by whom, to whom, and under what conditions. We also demonstrate that our approach can scale using crowdworking to produce CI annotations of large privacy policy corpora. We perform a proof-of-concept experiment with 99 crowdworkers annotating 48 excerpts of privacy policies from 17 companies. The resulting high precision annotations indicate that CI annotation and analysis could be widely applied to privacy policies in future research.

January 30: Sabine Gless — Predictive Policing: In Defense of 'True Positives'
     ABSTRACT: Predictive policing has triggered a heated debate around the issue of false positives. Biased machine training can wrongly classify individuals as high risk simply as a result of belonging to a particular ethnic group and many agree such persons should not have to shoulder the burden of over-policing due to an inherent stochastic problem. This provocation however makes a case for the ‘true positives’. It claims that those who are caught red-handed, as a consequence of biased police profiling, offer the best opportunity to address the issue of biased profiling as they have a high incentive to raise the problem of discrimination during criminal proceedings. While the line of argument starts with a purely pragmatic consideration, it can be grounded on a more general reasoning of undesirability of discriminatory stops and searches as inherently unfair and a threat to social peace. To create an efficient legal tool against discriminatory law-enforcement, the defence should be entitled to contest a conviction based on biased predictive policing, with a specific exclusionary rule protecting ‘true positives’ against the use of tainted evidence.

Fall 2018

December 5: Discussion of current issues

November 28: Ashley Gorham — Algorithmic Interpellation
     ABSTRACT: The use of algorithmic logic for purposes as different as military strikes and targeted advertisements alone ought to alarm us. And yet, despite their rapidly increasing presence in our lives, our understanding of where and how algorithms are used, as well as their material effects, remains at a minimum. To be fair, algorithms are a technical and therefore unsurprisingly intimidating topic, and just what an algorithm is is not immediately obvious to many, if not most, people. Even among those who think they have a sense of what an algorithm is, it is still hard to define. As Tarleton Gillespie (2016) notes, as social scientists, “[w]e find ourselves more ready to proclaim the impact of algorithms than to say what they are” (18). With this in mind, and in light of the pervasiveness of algorithms in contemporary society, we have set to clarify the operations of algorithms through the use of Althusser’s theory of ideology, and in particular his concept of interpellation. It is our main contention that algorithms operate as mechanisms of capitalist interpellation and that a proper understanding of algorithms must appreciate this aspect of their workings. The argument will proceed as follows: first, we will offer a brief, and, admittedly incomplete, overview of the ways in which other scholars have conceptualized algorithms.  Second, we will examine Althusser’s theory of ideology, and, as his theory is a complicated one, we will discuss it in some detail. Finally, we will apply Althusser’s theory to the operations of algorithms, considering how an algorithm is well understood as a mechanism that “gives us a name.”

November 14: Mark Verstraete — Data Inalienabilities
     ABSTRACT: This paper explores the theoretical links between personal information and alienability. More specifically, I present a conceptual framework for thinking about limitations on the alienability of personal data. To that end, I argue that restrictions on the alienability of personal data are justifiable based on both analogies to other objects that are subject of limitations on transfer and the unique nature of personal data. One set of alienability limitations are present in Intellectual Property and constrain the alienability of creative works. For instance, Copyright’s Termination Transfer Right gives authors an inalienable option to regain rights in their creative works after a set period of years. Similarly, the doctrine of moral rights allows authors to retain some control over their work even after sale—preventing purchasers from destroying or altering the work. A second suite of alienability restrictions governs entitlements that are intimately bound up in the body and personhood. Third, and finally, personal data is unique. Unlike many other artifacts that are transferred or sold, personal data cannot be fully severed from people about whom the data refers. By contrast, traditional commodities like cars or furniture do not relate back to previous owners in the way that personal data does. Data subjects, on the other hand, have a continuing interest in the use of data about them that cannot be fully extinguished by transfer or sale.

November 7: Jonathan Mayer — Estimating Incidental Collection in Foreign Intelligence Surveillance
     ABSTRACT: Section 702 of the Foreign Intelligence Surveillance Act (FISA) authorizes the Intelligence Community to acquire communications within the United States when targeting non-U.S. persons outside the United States. Because of the increasingly global nature of communications, Section 702 intercepts foreseeably involve communications where a U.S. person is a party. This property of Section 702, dubbed "incidental collection," has been a subject of controversy for over a decade because it involves acquisition of a U.S. person's communications without a probable-cause warrant. Lawmakers on both sides of the aisle have called on the Intelligence Community to estimate the scale of incidental collection, in order to better understand how Section 702 operates and weigh Fourth Amendment considerations. Senior national security officials in the Obama and Trump administrations have acknowledged the value of estimating incidental collection, and the Intelligence Community has assessed possible methodologies and called for input from outside experts. In this session, I will present preliminary results from a working group on estimating incidental collection under Section 702. Princeton CITP convened leaders in national security, surveillance law, and privacy-preserving computation for a daylong session this summer in order to explore the problem and consider new methodologies. The group's scope was narrowly focused on estimation; it did not address broader policy or legal considerations for Section 702. I will explain points of consensus on data sources, statistics, and rejected methodologies, and I will present a possible path forward that leverages the latest privacy-preserving computation techniques to estimate incidental collection.

October 31: Sebastian Benthall — Trade, Trust, and Cyberwar
     ABSTRACT: In recent years, several nations have passed new policies restricting the use of information technology of foreign origin. These cybersecurity trade polices are legitimized by narratives around national security and privacy while also having "protectionist" economic implications. This talk frames cybersecurity trade policies in the broader theoretical context of war and trade. It then examines how cyberwar is different from other forms of trade conflict, and what implications this has for the potential for broad economic and political alignment on cybersecurity.

October 24: Yafit Lev-Aretz — Privacy and the Human Element
     ABSTRACT: The right to privacy has been traditionally discussed in terms of human observation and the formation of subsequent opinion or judgment.  Starting with Warren and Brandeis' "right to be let alone,"  and continuing with the privacy torts', the early days of privacy in the legal sphere placed crucial emphasis on human presence.  Often made arguments such as "I've got nothing to hide" on the one hand, and "you are being watched" on the other hand,  go to the heart of the human element which became an intuitive component around which the right to privacy has been structured, evolved, and interpreted over the years.  Nowadays, however, most information flows do not involve a human in the loop, and while we are pretty uncomfortable with human observation and subsequent judgment, algorithmic observation and judgment do not provoke similar discomfort.  This discrepancy can account for the privacy paradox, which refers to the difference between stated positions on information collection and widespread participation in it.  It can also explain the significant expansion of the privacy bundle in the past decade, to include concerns such as discrimination, profiling, unjust enrichment, and online manipulation.  In my work, I point to the failure of privacy as a policy goal and build on the work of Priscila Regan and Dan Solove to explain this failure in, beyond the use of wrong metaphors and the individual focus, the mismatch between the strong human presence in privacy intuitions and the modern surveillance culture that growingly capitalizes on diverse means of humanless tracking.  Consequently, I call for a conceptual shift that keeps privacy within the boundaries of the human element and discusses all other informational risks under a parallel paradigm of legal protection.

October 17: Julia Powles — AI: The Stories We Weave; The Questions We Leave
     ABSTRACT: It has become almost automatic. While public conversation about artificial intelligence readily diverts into problems of the long future (the rise of the machines) and ingrained past (systemic inequality, now perpetuated and reinforced in data-driven systems), a small cadre of tech companies amasses unprecedented power on a planetary scale. This talk is an exploration and invitation. It interrogates the debates we have, and those we need, about AI, algorithms, rights, regulation, and the future. It examines what we talk about, why we talk about it, what we should ask and solve instead, and what is required to spur a richer, more imaginative, more innovative conversation about the world we wish to create.

October 10: Andy Gersick — Can We Have Honesty, Civility, and Privacy Online? Implications from Evolutionary Theories of Animal and Human Communication
     ABSTRACT: Early internet optimism centered on two unique affordances of online interaction that were expected to empower disenfranchised and diasporic groups: the mutability of online identities and the erasure of physical distance. The ability to interact from the safety of a distant and sometimes hidden vantage has remained a core feature of online social life, codified in the rules of social-media sites and considered in discussion of legal privacy rights. But it is now far from clear that moving our social lives online has “empowered” the disenfranchised, on balance. In fact, the disembodied and dispersed nature of online communities has increasingly appeared to fuel phenomena like trolling, cyberbullying and the deliberate spread of misinformation. Science on the evolution of communication has a lot to say about how social animals evaluate the trustworthiness of potential mates and rivals, allies and enemies. Most of that work shows that bluffing, false advertisement and other forms of deceptive signaling are only held in check when signal-receivers get the chance to evaluate the honesty of signal-producers through direct and repeated contact. It’s a finding that holds true across the animal kingdom, and it has direct implications for our current socio-political discourse. The antagonistic trolls and propagandistic sock-puppets that have invaded our politics are using deceptive strategies that are as old as the history of communication. What’s new, in human social evolution, is our vulnerability to those strategies within a virtual environment. I will discuss elements of evolutionary theory that seem relevant to online communication and internet privacy, and I hope to have a dialogue with attendees about (a) how those theories intersect with core elements of internet privacy law, and (b) whether we have to alter our basic expectations about online privacy if we want social-media interactions that favor cooperation over conflict.

October 3: Eli Siems — The Case for a Disparate Impact Regime Covering All Machine-Learning Decisions
     ABSTRACT: The potential for Machine Learning (ML) tools to produce discriminatory models is now well documented. The urgency of this problem is compounded both by the rapid spread of these tools into socially significant decision structures and by the unique obstacles ML tools pose to the correction of bias. These unique challenges fit into two categories: (1) the uniquely obfuscatory nature of correlational modeling and the threat of proxy variables standing in for impermissible considerations, and (2) the overriding tendency of ML tools to “freeze” historical disparities in place, and to replicate and even exacerbate them. Currently, two ML tools with identical biases stemming from identical issues will be reviewed differently depending on the context in which they are utilized. Under Title VII, for example, statistical evidence of discrimination would be sufficient to initiate a claim, but the same claim under the Constitution would be dismissed at the pleading stage without additional evidence of intent to discriminate. This paper attempts to work within the (profoundly flawed) strictures of existing Constitutional and statutory law to propose the adoption of a unified, cross-contextual regime that would allow a plaintiff challenging the decisions of an ML tool to utilize statistical evidence of discrimination to carry a claim beyond the initial pleading stage, empowering plaintiffs to demand a record of a tool’s design and the data upon which it trained. In support of extending a disparate impact regime to all instances of ML discrimination, I carefully analyze the Supreme Court’s treatment of statistical evidence of discrimination under both the Fourteenth Amendment and under statutory Civil Rights law. While the Supreme Court has repeatedly disavowed the application of disparate-impact style claims to Fourteenth Amendment Equal Protection, I argue that, for myriad reasons, its stated logic in doing so does not hold when the decision-maker in question is an ML tool. By analyzing Equal Protection holdings from the fields of government employment, death penalty sentencing, policing, and risk assessment as well as holdings under Title VII of the Civil Rights Act, the Fair Housing Act, and the Voting Rights Act, I identify the contextual qualities that have factored into the Court’s decisions to allow or disallow disparate impact evidence. I then argue that the court’s own reasoning in barring the use of such evidence in contexts like death penalty sentencing and policing decisions cannot apply to ML decisions, regardless of context.

September 26: Ari Waldman — Privacy's False Promise
     ABSTRACT: Privacy law—a combination of statutes, constitutional norms, regulatory orders, and court decisions—has never seemed stronger. The European Union’s General Data Protection Regulation (GDPR) and California’s Consumer Privacy Act (CalCPA) work in parallel with the Federal Trade Commission’s broad regulatory arsenal to put limits on the collection, use, and manipulation of personal information. The United States Supreme Court has reclaimed the Fourth Amendment’s historical commitment to curtail pervasive police surveillance by requiring warrants for cell-site location data. And the EU Court of Justice has challenged the cross-border transfer of European citizens’ data, signaling that American companies need to do far more to protect personal information. This seems remarkably comprehensive. But the law’s veneer of protection is hiding the fact that it is built on a house of cards. Privacy law is failing to deliver its promised protections in part because the responsibility for fulfilling legal obligations is being outsourced to layers of compliance professionals who see privacy law through a corporate, rather than substantive lens. This Article provides a comprehensive picture of this outsourcing market and argues that the industry’s players are having an outsized and constraining impact on the construction of privacy law in practice. Based on original primary source research into the ecosystem of privacy professionals, lawyers, and the third-party vendors on which they increasingly rely, I argue that because of a multilayered process outsourcing corporate privacy duties—one in which privacy leads outsource privacy compliance responsibilities to their colleagues, their lawyers, and an army of third-party vendors—privacy law is in the middle of a process of legal endogeneity: mere symbols of compliance are replacing real progress on protecting the privacy of consumers.

September 19: Marijn Sax — Targeting Your Health or Your Wallet? Health Apps and Manipulative Commercial Practices
     ABSTRACT: Most popular health apps (e.g. MyFitnessPal, Headspace, Fitbit) are not just helpful tools aimed at improving the user's health; they are also commercial services that use the idea of health to monetize their user base. In order to do so, popular health apps rely on (1) advanced analytical tools to 'optimize' monetization, and (2) propagate a rather particular health discourse aimed at making users understand their own health in a way that serves the commercial interests of health apps. Given the fact that health is very important to people and given the fact that health apps often try to mask their commercial intentions by appealing to the user's health, I argue that commercial health app practices are potentially manipulative. I offer a conception of manipulation to help explain how health app users could be manipulated by health apps. To address manipulation in health apps, it would be wise to not only focus on questions of informational privacy and data protection law, but also consider decisional privacy and unfair commercial practice law.

September 12: Mason Marks — Algorithmic Disability Discrimination
     ABSTRACT: In the Information Age, we continuously shed a trail of digital traces that are collected and analyzed by corporations, data brokers, and government agencies. Using artificial intelligence tools such as machine learning, they convert these traces into sensitive medical information and sort us into health and disability-related categories. I have previously described this process as mining for emergent medical data (EMD) because the health information inferred from digital traces often arises unexpectedly (and is greater than the sum of its parts). EMD is employed in epidemiological research, advertising, and a growing scoring industry that aims to sort and rank us. This paper describes how EMD-based profiling, targeted advertising, and scoring affects the health and autonomy of people with disabilities while circumventing existing health and anti-discrimination laws. Because many organizations that collect EMD are not covered entities under the Health Information Portability and Accountability Act (HIPAA), EMD-mining circumvents HIPAA's Privacy Rule. Moreover, because the algorithms involved are often inscrutable (or maintained as trade secrets), violations of anti-discrimination laws can be difficult to detect. The paper argues that the next generation of privacy and anti-discrimination laws must acknowledge that in the Information Age, health data does not originate solely within traditional medical contexts. Instead, it can be pieced together by artificial intelligence from the digital traces we scatter throughout real and virtual worlds.

Spring 2018

May 2: Ira Rubinstein Article 25 of the GDPR and Product Design: A Critical View [with Nathan Good and Guilermo Monge, Good Research]
     ABSTRACT: The General Data Protection Regulation (GDPR) seeks to protect the privacy and security of EU citizens in a world that is vastly different from that of the 1995 Data Protection Directive. This is largely due to the rise of the Internet and digital technology, which together define how we communicate, access the world of ideas, and make ourselves into social creatures. The GDPR seeks to modernize European data protection law by establishing data protection as a fundamental right. It requires data controllers to respect the rights of individuals including new rights of erasure and data portability and to comply with new obligations including accountability, a risk-based approach, impact assessments and data protection by design and default (DPDD). Ideally, this new DPDD obligations will change business norms by bringing data protection to the forefront of product design. Although the, GDPR strives to remain sufficiently broad and flexible to allow for creative solutions, it also adopts a belt and suspenders approach to regulation, imposing multiple, overlapping obligations on data controllers. What, then, is the specific task of the DPDD provision? It requires organizations to implement privacy-enhancing measures at the earliest stage of design and to select techniques that by default are the most protective of individuals' privacy and data protection. More specifically, Article 25 requires that "controllers shall ... implement appropriate technical and organisational measures ... in an effective  manner... in order to meet the requirements of this Regulation and protect the rights of data subjects" and to ensure that "by default, only personal data which are necessary for each specific purpose of the processing are processed."This begs several questions, however. For example, do organizations achieve these goals by implementing specific measures over and above those they might otherwise put into effect to meet their obligations under the remainder of the Regulation? Are certain "technical and organizational measures" (like pseudonymisation and data minimisation) required or merely recommended? Are there specific design and engineering techniques that organizations should follow to satisfy their DPDD obligations? And how do organizations know when their efforts satisfy Article 25 requirements, especially when they have already complied with other obligations? In this paper, we examine what technology companies are doing currently to satisfy their obligations under Article 25 in the course of establishing overall GDPR compliance programs. We expect to find that companies with limited privacy resources are confining their efforts to a compliance-based approach, resulting in a patchwork of privacy practices rather than adoption of a privacy-based model of product design. And we predict that in a rush to achieve compliance, these companies will fail to implement the methods and practices that comprise privacy by design as that term is understood, not by regulators, but by engineers and designers as described in our earlier work (Rubinstein & Good, Privacy by Design: A Counterfactual Analysis of Google and Facebook Privacy Incidents). In other words, many firms will treat the DPDD obligation as just a checkbox requirement.We will investigate these claims via case studies. However, we do not rely on surveys of a representative sampling of regulated firms or snowball sampling of industry practitioners whose work exemplifies the methods and practices that engineers and designers rely on to achieve specific privacy (and security) goals. Rather, we will analyze what two groups of vendors are offering their customers to help them operationalize the GDPR generally and Article 25 in particular. We will look at both privacy technology vendors (a new niche market of firms selling into the private-sector market of firms needing help with GDPR compliance) and cloud infrastructure vendors (like Microsoft) who are marketing their platform to large multinationals and SMEs as GDPR-ready. (If necessary, we may supplement this approach with telephone interviews but mostly for purposes of follow up questions rather than as a source of primary knowledge.) Finally, we report on the incentives and motivations behind these practical solutions and discuss how supervisory authorities might develop policies to encourage firms to adopt appropriate solutions and develop the necessary expertise to achieve them. In sum, we provide an analysis of Article 25 with the goal of helping EU regulators bridge the gap between the ideals and practice of data protection by design and default.

April 25: Elana Zeide — The Future Human Futures Market
     ABSTRACT: This paper considers the emerging market in student futures as a cautionary tale. Income sharing arrangements involve the explicit and literal commodification of “human capital” by for-profit third parties who broker income sharing agreements between private investors and students with promising predictive data profiles. This paper considers the problematic legal and ethical aspects of the predictive technologies driving these markets and draws a parallel to the role schools and third-party career platforms play in sorting, scoring, and predicting student futures as part of a formal education. These matching systems not only mete out opportunity but preempt access to opportunity (see Kerr & Earle).
Many coding “bootcamps” take an untraditional approach to student financing. Some after a money-back employment “guarantee.” Others use “human capital” contracts. Instead of requiring students to pay tuition up front to take out onerous loans based on uncertain career paths, schools claim a portion of a graduate’s wages upon gainful employment. A two-year software engineering program in San Francisco, for example, asks for no money upfront but then takes 17% of students’ internship earnings during the program and 17% of salaries for three years after finding a job. Other schools, advocates, and policymakers push for similar private education funding arrangements, including bills introduced in the U.S. Senate and House of Representatives in 2017. They promote “income share agreements” as more equitable and efficient for students than the traditional student loan system, where debt may be disproportionate to post-graduation wages. These arrangements raise numerous constitutional, legal, and ethical questions. Do students have to accept the first offer they receive? How can they be enforced? How might this arrangement shift who can obtain a post-secondary credential? Are they a simply a modern version of indentured servitude? A less discussed but key component of the developing “futures” market is the role of opportunity brokers: third parties who design, implement, and “take the complexity out of” income sharing agreements. These for-profit companies match interested investors with promising “opportunities” based on proprietary predictive analytics that project future income. Some go beyond commodifying student futures to securitizing them: as one commentator writes “human capital - the present value of individuals’ future earnings - may soon become an important investable asset class, following in the footsteps of home mortgage debt.” Schools are themselves opportunity brokers, credential-creators, and career matchmakers that end up determining whose futures we support - individually, institutionally, or as a society. Scholars and popular entertainment offer chilling accounts of the dystopian aspects of a scored society, governed by anticipatory and proprietary data-models likely to reinforce existing patterns of privilege and inequity. Ubiquitous surveillance systems that chill free expression, promote performativity and create circumstances ripe for social control and engineering. Except we already have such a system in place: the formal education system. American schools not only provide whatever one considers “an education,” but also sort, score, and predict student potential. The tools they use to do so - textbooks, SATs, and standards like the Common Core - are subject to intense public scrutiny. Schools increasingly rely on for-profit vendors to provide the platforms and tools that deliver, assess, and document student progress. These include “personalized learning systems” that continuously monitor student progress and adapt instruction at scale - what some have called the “mass customization” of education. They use predictive analytics classify students, infer characteristics, and predict optimal learning pathways. Higher education institutions also use predictive platforms to make recruiting and admissions decisions, award financial aid, and detect students at risk of dropping out. Social media platforms and people analytics firms increasingly mediate and automate candidate-employer matching. This system might similarly not just deny but preempt access to opportunity without accompanying due process provisions. And it is likely to do so in ways that reinforce today's inequities - creating a new segregation of education.

April 18: Taylor Black — Performing Performative Privacy: Applying Post-Structural Performance Theory for Issues of Surveillance Aesthetics
     ABSTRACT: In 2017, Scott Skinner-Thompson published “Performative Privacy,” based in part on work with this group. In the article, he “identifies a new dimension of public privacy” and argues for a reading of certain public acts of anti-surveillance resistance as performative, and therefore to be legally understood as expressions of speech and protected as such. In this talk, I extend the framework of performative privacy from the perspective of performance studies, and discuss some new applications of critical theory and performance theory in contemporary issues of surveillance. As a discipline performance studies, particularly its critique of speech as act and its intervention in the use of liveness in action, offers an opportunity to meaningfully trouble the distinction between efficacy and expression underlying the question of performative privacy. To test these limits and demonstrate the possible applications of performance theory, I follow the performative privacy framework in two directions. First, we’ll examine privacy’s impact on performance and aesthetics in the rise of the post-Snowden “surveillance art” movement. Then, I incorporate Clare Birchall’s emerging research on “shareveillance” to explore the question of efficacy in surveillance resistance and the resulting impacts of performance entering into privacy discourse.

April 11: John Nay Natural Language Processing and Machine Learning for Law and Policy Texts
     ABSTRACT: Almost all law is expressed in natural language; therefore, natural language processing (NLP) is a key component of efforts to automatically structure, explore and predict law and policy at scale. NLP converts unstructured text into a formal, structured representation that computers can analyze.  First, we provide a brief overview of the different types of law and policy texts and the different types of machine learning methods to process those texts. We introduce the core idea of representing words, sentences and documents as numbers. Then we describe NLP and machine learning tools for leveraging the text data to accomplish tasks. We describe methods for automatically summarizing content (sentiment analyses, text summaries, topic models), extracting content (entities, attributes and relations), retrieving information and documents, predicting outcomes related to text, and answering questions.

April 4: Sebastian Benthall — Games and Rules of Information Flow
     ABSTRACT: Attempts to characterize the nature of privacy must acknowledge the complexity of concept. They tend to be either particularist (acknowledging many, unrelated, particular meanings) or contextualist (describing how the same concept manifests itself differently across social contexts. Both these approaches are insufficient for making policy and technical design decisions about technical infrastructure that spans many different contexts. A new model is needed, one that is compatible with these theories but which characterizes privacy considerations in terms of the reality of information flow, not our social expectations of it. I build a model of information flow from the theories of Fred Dretske, Judea Pearl, and Helen Nissenbaum that is compatible with both intuitive causal reasoning and contemporary machine learning methods. This model clarifies that information flow is a combination of causal flow and nomic association, where the associations of information depend on the causal structure of which the flow is a part. This model also affords a game theoretic and mechanism design extensions using the Multi-Agent Influence Diagram framework.
I employ this model to illustrate several different economic contexts involving personal information, as well as what happens when these contexts collapse. The model allows for a robust formulation of the difference between a tactical and a strategic information flow, which roughly correspond to the differences between the impact of a sudden data breach and the chilling effects of ongoing surveillance.

March 28: Yann Shvartzshanider and Noah Apthorpe Discovering Smart Home IoT Privacy Norms using Contextual Integrity
     ABSTRACT: The proliferation of Internet of Things (IoT) devices for consumer “smart” homes raises concerns about user privacy. We present a survey method based on the Contextual Integrity (CI) privacy framework that can quickly and efficiently discover privacy norms at scale. We apply the method to discover privacy norms in the smart home context, surveying 1,731 American adults on Amazon Mechanical Turk. For $2,800 and in less than six hours, we measured the acceptability of 3,840 information flows representing a combinatorial space of smart home devices sending consumer information to first and third-party recipients under various conditions. Our results provide actionable recommendations for IoT device manufacturers, including design best practices and instructions for adopting our method for further research.

March 21: Cancelled

March 7: Cancelled

February 28: Thomas Streinz TPP’s Implications for Global Privacy and Data Protection Law
     ABSTRACT: On 8 March, the remaining eleven parties of the original Trans-Pacific Partnership (TPP)–Australia, Brunei, Canada, Chile, Japan, Malaysia, Mexico, New Zealand, Peru, Singapore, and Vietnam–will meet in Santiago, Chile to revive the TPP via the awkwardly (and arguably misleadingly) labelled Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP). This is a surprising development for two reasons: 1) After President Trump withdrew the US from the original TPP in January 2017, most observers believed the agreement was dead for good. 2) The TPP11 parties preserved the vast majority of the provisions of the original TPP (with notable exceptions mainly in the investment and IP chapters) despite the fact that the agreement mainly followed US models of (so called) free trade agreements (FTAs) and was in fact promoted as “Made in America” by the Office of the United States Trade Representative (USTR) during the Obama administration which was particularly proud of a new set on rules that it branded as the "Digital2Dozen." The chapter on “electronic commerce” which contains most but not all provisions with relevance for internet law and regulation got incorporated into CPTPP without any modifications and is bound to become the template for future trade agreements (including the ongoing renegotiations of NAFTA) without EU participation. In my presentation for PRG, I will focus on TPP’s (weak) provision on “personal information protection” (Article 14.8) and its innovative rules for free data flows (Article 14.11) and against data localization requirements (Article 14.13). I will explain and we should discuss why the EU views these rules as problematic from a privacy perspective. In its recent agreement with Japan, which is also a TPP party, this can got kicked down the road, but on 31 January 2018 the European Commission announced that it would endorse provisions for data flows and data protection in EU trade agreements. The crucial difference to the US model as incorporated in TPP is that the EU will likely require compliance with the General Data Protection Regulation (GDPR) as a condition for free data flows—complementing the existing adequacy assessment procedures and leveraging its trade muscle to promote the GDPR as the global standard.

February 21: Ben Morris, Rebecca Sobel, and Nick Vincent — Direct-to-Consumer Sequencing Kits: Are Users Losing More Than They Gain?
     ABSTRACT: Direct-to-consumer genetic sequencing, provided by companies like 23andMe, Ancestry, and Helix, has opened a myriad of scientific and legal issues ranging from the statistical interpretation of results to access, regulation, and user privacy. Interestingly, the most recent efforts have attempted to tie together direct-to-consumer testing with the blockchain and cryptocurrencies, but consumer protection and privacy concerns remain. In this presentation, we will provide a history of the direct-to-consumer genetic sequencing market and how we have arrived at the current market. We will also highlight some of the legal and regulatory issues surrounding the activities of these companies in relation to FDA requirements, the Genetic Information Nondiscrimination Act of 2008 (GINA), and the Health Insurance Portability and Accountability Act of 1996 (HIPAA). Finally, we will use current examples from the emerging market of direct-to-consumer gut microbiome sequencing kits as a study for how privacy policies of these companies are evolving in a developing market and what concerns customers could (and perhaps should) have when using these kits.

February 14: Eli Siems — Trade Secrets in Criminal Proceedings: The Battle over Source Code Discovery
     ABSTRACT: In Intellectual Property law, a trade secret is information which “derives economic value from not being generally known . . . to other persons who can obtain economic value from its disclosure or use.” (Uniform Trade Secrets Act). Unlike patent, trade secret law will cease to protect against the use of information once that information becomes generally known. A trade secret, once disclosed, is the proverbial cat out of the bag. For this reason, courts have developed an evidentiary privilege protecting trade secrets from disclosure in trial unless a party shows that such disclosure is actually necessary to a just resolution. This privilege has developed over decades of civil litigation. Recently, a confluence of factors has led to an increase in assertions of the trade secret privilege in criminal trials. State police departments and prosecutors have begun contracting with private software developers for the use of algorithmic tools that generate either forensic proof to be used at trial, “risk assessment” to be used at sentencing, or data for policing. Criminal defendants have sought access to the source code for such programs only to be met with claims that the information sought is privileged as a trade secret. In addressing what a criminal defendant must show to overcome the privilege, some courts have directly applied the standard from civil common law, while others have imported key elements of that standard. Assuming that a defendant must always make some showing to justify the disclosure of “trade secret” source code in her criminal trial, her effective defense will require an understanding of the nature of her burden—must she show that the code is “necessary” to her defense (a replica of the civil standard), that the code is simply “material and relevant” (in line with basic criminal discovery standards), or something in between? This talk will draw from a spate of cases in which defendants sought the source code from “probabilistic genotyping” programs in order to define the contours of these standards as they have recently been applied. Centrally, it will identify the factors that have led courts to find that criminal defendants have failed to carry the burden of establishing either relevance or necessity of the source code. It will reveal that judges have relied on the same validation studies properly considered at the admissibility stage (where the court must determine the reliability of expert/scientific evidence) to determine that a defense review of the source code is either irrelevant or unnecessary. The idea that validation studies can defeat a defendant’s claim that source code is relevant or necessary to her defense fails to account for two key considerations—first, that a defendant may seek to challenge something other than the reliability of the software, and second, that validation of these tools may not be providing the type of assurance legally sufficient to defeat a defendant’s discovery requests. In addition to critiquing judicial reasoning, this talk will address deficiencies in defense pleadings and potential adaptations that may lead to more successful discovery motions in the future.

February 7: Madeline Bryd and Philip Simon Is Facebook Violating U.S. Discrimination Laws by Allowing Advertisers to Target Users?
     ABSTRACT: In 2016, ProPublica published an article revealing the startlingly easy method Facebook’s advertising program provided to exclude protected classes from seeing employment, housing, and credit advertisements. The article raised numerous questions about potential liability and what other mechanisms advertisers could use to discriminate via Facebook’s platform. This presentation will address whether Facebook can be held liable for advertising discrimination based on the discriminatory uses of its platform by advertisers; the current state of U.S. discrimination laws with respect to targeted online advertising in general, and; whether online platforms can escape liability through the Communications Decency Act (CDA) § 230. Our analysis of potential discriminatory uses will focus on research done by Krishna Gummadi and his team that explores Facebook’s advertising features (to be presented at FAT* ‘18, February 2018). Their paper identifies three ways in which advertisers can target users: PII-based targeting, attribute-based targeting, and look-alike audience targeting. Each targeting tool will be analyzed in the context of employment, housing, and credit discrimination laws to address whether these features can be illegally used by advertisers. Finally, we will address possible ways in which Facebook can be held liable for these illegal uses, despite any protection against liability that it may enjoy under CDA-230.

January 31: Madelyn Sanfilippo Sociotechnical Polycentricity: Privacy in Nested Sociotechnical Networks 
Knowledge access is both constrained and supported by social, political, human, economic, and technological factors, making formal and informal governance of knowledge a set of complex sociotechnical constructs. Political theory surrounding polycentric governance has long structured inquiry into contexts in which public service provision is nested or competing. This talk will define and discuss applications of polycentric governance theory to sociotechnical knowledge services, in order to support empirically grounded policy-making. Polycentricity is often defined in terms of many nested or overlapping contexts or jurisdictions, which may compete with or compliment one another, yet is also fundamentally about the many centers of decision-making within those contexts or jurisdictions. Sociotechnical polycentricity pertains not only to the complex exogenous policy environment, but also to endogenous decisions of firms or actors, which themselves overlap with this external environment. Extensive literature demonstrates how polycentricity illuminates complexity and supports policy recommendations or improvements, based off of failures, complexity, or conflicts in cases; this talk will explore polycentric frames applied to questions around sociotechnical governance, including various examples centered on knowledge access and privacy.

January 24: Jason Schultz and Julia Powles Discussion about the NYC Algorithmic Accountability Bill 
     ABSTRACT: The New York City Council recently passed one of the first laws in the United States to address “algorithmic accountability.” The bill, NY 1696 proposed by council member James Vacca, creates a task force to explore how the city can best open up public agency’s computerized decision-making tools to public scrutiny. This effort raises many technical, legal, and political questions about how algorithmic systems fit into the broader notions of responsible and responsive government. Julia Powles and Jason Schultz have each been involved in the debate over the bill and will lead a discussion of its contents, its context, and its next steps. Julia Powles' recent New Yorker piece for some more background.

Fall 2017

November 29: Kathryn Morris and Eli Siems Discussion of Carpenter v. United States
November 15:Leon Yin Anatomy and Interpretability of Neural Networks
November 8: Ben Zevenbergen Contextual Integrity for Password Research Ethics?
November 1: Joe Bonneau An Overview of Smart Contracts
October 25: Sebastian Benthall Modeling Social Welfare Effects of Privacy Policies
October 18: Sue Glueck Future-Proofing the Law
October 11: John Nay — Algorithmic Decision-Making Explanations: A Taxonomy and Case Study
October 4:Finn Bruton — 'The Best Surveillance System we Could Imagine': Payment Networks and Digital Cash
September 27: Julia Powles Promises, Polarities & Capture: A Data and AI Case Study
September 20: Madelyn Rose Sanfilippo AND Yafit Lev-Aretz — Breaking News: How Push Notifications Alter the Fourth Estate
September 13: Ignacio Cofone — Anti-Discriminatory Privacy

Spring 2017

April 26: Ben Zevenbergen Contextual Integrity as a Framework for Internet Research Ethics
April 19: Beate Roessler Manipulation
April 12: Amanda Levendowski Conflict Modeling
April 5: Madelyn Sanfilippo Privacy as Commons: A Conceptual Overview and Case Study in Progress
March 29: Hugo Zylberberg Reframing the fake news debate: influence operations, targeting-and-convincing infrastructure and exploitation of personal data
March 22: Caroline Alewaerts, Eli Siems and Nate Tisa will lead discussion of three topics flagged during our current events roundups: smart toys, the recently leaked documents about CIA surveillance techniques, and the issues raised by the government’s attempt to obtain recordings from an Amazon Echo in a criminal trial. 
March 8: Ira Rubinstein Privacy Localism
March 1: Luise Papcke Project on (Collaborative) Filtering and Social Sorting
February 22: Yafit Lev-Aretz and Grace Ha (in collaboration with Katherine Strandburg) Privacy and Innovation     
February 15: Argyri Panezi Academic Institutions as Innovators but also Data Collectors - Ethical and Other Normative Considerations
February 8: Katherine Strandburg Decisionmaking, Machine Learning and the Value of Explanation
February 1: Argyro Karanasiou A Study into the Layers of Automated Decision Making: Emergent Normative and Legal Aspects of Deep Learning
January 25: Scott Skinner-Thompson Equal Protection Privacy

Fall 2016

December 7: Tobias Matzner The Subject of Privacy
November 30: Yafit Lev-Aretz Data Philanthropy
November 16: Helen Nissenbaum Must Privacy Give Way to Use Regulation?
November 9: Bilyana Petkova Domesticating the "Foreign" in Making Transatlantic Data Privacy Law
November 2: Scott Skinner-Thompson Recording as Heckling
October 26: Yan Shvartzhnaider Learning Privacy Expectations by Crowdsourcing Contextual Informational Norms
October 19: Madelyn Sanfilippo Privacy and Institutionalization in Data Science Scholarship
October 12: Paula Kift The Incredible Bulk: Metadata, Foreign Intelligence Collection, and the Limits of Domestic Surveillance Reform

October 5: Craig Konnoth Health Information Equity
September 28: Jessica Feldman the Amidst Project
September 21: Nathan Newman UnMarginalizing Workers: How Big Data Drives Lower Wages and How Reframing Labor Law Can Restore Information Equality in the Workplace
September 14: Kiel Brennan-Marquez Plausible Cause

Spring 2016

April 27: Yan Schvartzschnaider Privacy and loT AND Rebecca Weinstein - Net Neutrality's Impact on FCC Regulation of Privacy Practices
April 20: Joris van Hoboken Privacy in Service-Oriented Architectures: A New Paradigm? [with Seda Gurses]

April 13: Florencia Marotta-Wurgler Who's Afraid of the FTC? Enforcement Actions and the Content of Privacy Policies (with Daniel Svirsky)

April 6: Ira Rubinstein Big Data and Privacy: The State of Play

March 30: Clay Venetis Where is the Cost-Benefit Analysis in Federal Privacy Regulation?

March 23: Diasuke Igeta An Outline of Japanese Privacy Protection and its Problems

                  Johannes Eichenhofer Internet Privacy as Trust Protection

March 9: Alex Lipton Standing for Consumer Privacy Harms

March 2: Scott Skinner-Thompson Pop Culture Wars: Marriage, Abortion, and the Screen to Creed Pipeline [with Professor Sylvia Law]

February 24: Daniel Susser Against the Collection/Use Distinction

February 17: Eliana Pfeffer Data Chill: A First Amendment Hangover

February 10: Yafit Lev-Aretz Data Philanthropy

February 3: Kiel Brennan-Marquez Feedback Loops: A Theory of Big Data Culture

January 27: Leonid Grinberg But Who BLocks the Blockers? The Technical Side of the Ad-Blocking Arms Race

Fall 2015

December 2: Leonid Grinberg But Who BLocks the Blockers? The Technical Side of the Ad-Blocking Arms Race AND Kiel Brennan-Marquez - Spokeo and the Future of Privacy Harms
November 18: Angèle Christin - Algorithms, Expertise, and Discretion: Comparing Journalism and Criminal Justice
November 11: Joris van Hoboken Privacy, Data Sovereignty and Crypto
November 4: Solon Barocas and Karen Levy Understanding Privacy as a Means of Economic Redistribution
October 28: Finn Brunton Of Fembots and Men: Privacy Insights from the Ashley Madison Hack

October 21: Paula Kift Human Dignity and Bare Life - Privacy and Surveillance of Refugees at the Borders of Europe
October 14: Yafit Lev-Aretz and co-author, Nizan Geslevich Packin Between Loans and Friends: On Soical Credit and the Right to be Unpopular
October 7: Daniel Susser What's the Point of Notice?
September 30: Helen Nissenbaum and Kirsten Martin Confounding Variables Confounding Measures of Privacy
September 23: Jos Berens and Emmanuel Letouzé Group Privacy in a Digital Era
September 16: Scott Skinner-Thompson Performative Privacy

September 9: Kiel Brennan-Marquez Vigilantes and Good Samaritan

Spring 2015

April 29: Sofia Grafanaki Autonomy Challenges in the Age of Big Data
                 David Krone Compliance, Privacy and Cyber Security Information Sharing
                 Edwin Mok Trial and Error: The Privacy Dimensions of Clinical Trial Data Sharing
                 Dan Rudofsky Modern State Action Doctrine in the Age of Big Data

April 22: Helen Nissenbaum Respect for Context' as a Benchmark for Privacy: What it is and Isn't
April 15: Joris van Hoboken From Collection to Use Regulation? A Comparative Perspective
April 8: Bilyana Petkova
 Privacy and Federated Law-Making in the EU and the US: Defying the Status Quo?
April 1: Paula Kift — Metadata: An Ontological and Normative Analysis

March 25: Alex Lipton — Privacy Protections for the Secondary User of Consumer-Watching Technologies

March 11: Rebecca Weinstein (Cancelled)
March 4: Karen Levy & Alice Marwick — Unequal Harms: Socioeconomic Status, Race, and Gender in Privacy Research

February 25 : Luke Stark — NannyScam: The Normalization of Consumer-as-Surveillorm

February 18: Brian Choi A Prospect Theory of Privacy

February 11: Aimee Thomson — Cellular Dragnet: Active Cell Site Simulators and the Fourth Amendment

February 4: Ira Rubinstein — Anonymity and Risk

January 28: Scott Skinner-Thomson Outing Privacy


Fall 2014

December 3: Katherine Strandburg — Discussion of Privacy News [which can include recent court decisions, new technologies or significant industry practices]

November 19: Alice Marwick — Scandal or Sex Crime? Ethical and Privacy Implications of the Celebrity Nude Photo Leaks

November 12: Elana Zeide — Student Data and Educational Ideals: examining the current student privacy landscape and how emerging information practice and reforms implicate long-standing social and legal traditions surrounding education in America. The Proverbial Permanent Record [PDF]

November 5: Seda Guerses — Let's first get things done! On division of labor and practices of delegation in times of mediated politics and politicized technologies
October 29:Luke Stark — Discussion on whether “notice” can continue to play a viable role in protecting privacy in mediated communications and transactions given the increasing complexity of the data ecology and economy.
Kristen Martin — Transaction costs, privacy, and trust: The laudable goals and ultimate failure of notice and choice to respect privacy online

Ryan Calo — Against Notice Skepticism in Privacy (and Elsewhere)

Lorrie Faith Cranor — Necessary but Not Sufficient: Standardized Mechanisms for Privacy Notice and Choice
October 22: Matthew Callahan — Warrant Canaries and Law Enforcement Responses
October 15: Karen Levy — Networked Resistance to Electronic Surveillance
October 8: Joris van Hoboken —  The Right to be Forgotten Judgement in Europe: Taking Stock and Looking Ahead

October 1: Giancarlo Lee — Automatic Anonymization of Medical Documents
September 24: Christopher Sprigman — MSFT "Extraterritorial Warrants" Issue 

September 17: Sebastian Zimmeck — Privee: An Architecture for Automatically Analyzing Web Privacy Policies [with Steven M. Bellovin]
September 10: Organizational meeting

Spring 2014

April 30: Seda Guerses — Privacy is Security is a prerequisite for Privacy is not Security is a delegation relationship
April 23: Milbank Tweed Forum Speaker — Brad Smith: The Future of Privacy
April 16: Solon Barocas — How Data Mining Discriminates - a collaborative project with Andrew Selbst, 2012-13 ILI Fellow
March 12: Scott Bulua & Amanda Levendowski — Challenges in Combatting Revenge Porn

March 5: Claudia Diaz — In PETs we trust: tensions between Privacy Enhancing Technologies and information privacy law: The presentation is drawn from a paper, "Hero or Villain: The Data Controller in Privacy Law and Technologies” with Seda Guerses and Omer Tene.

February 26: Doc Searls Privacy and Business

February 19: Report from the Obfuscation Symposium, including brief tool demos and individual impressions

February 12: Ira Rubinstein The Ethics of Cryptanalysis — Code Breaking, Exploitation, Subversion and Hacking
February 5: Felix Wu — The Commercial Difference which grows out of a piece just published in the Chicago Forum called The Constitutionality of Consumer Privacy Regulation

January 29: Organizational meeting

Fall 2013

December 4: Akiva Miller — Are access and correction tools, opt-out buttons, and privacy dashboards the right solutions to consumer data privacy? & Malte Ziewitz What does transparency conceal?
November 20: Nathan Newman — Can Government Mandate Union Access to Employer Property? On Corporate Control of Information Flows in the Workplace

November 6: Karen Levy — Beating the Box: Digital Enforcement and Resistance
October 23: Brian Choi — The Third-Party Doctrine and the Required-Records Doctrine: Informational Reciprocals, Asymmetries, and Tributaries
October 16: Seda Güerses — Privacy is Don't Ask, Confidentiality is Don't Tell
October 9: Katherine Strandburg — Freedom of Association Constraints on Metadata Surveillance
October 2: Joris van Hoboken — A Right to be Forgotten
September 25: Luke Stark — The Emotional Context of Information Privacy
September 18: Discussion — NSA/Pew Survey
September 11: Organizational Meeting

Spring 2013

May 1: Akiva Miller — What Do We Worry About When We Worry About Price Discrimination
April 24: Hannah Block-Wheba and Matt Zimmerman — National Security Letters [NSL's]

April 17: Heather Patterson — Contextual Expectations of Privacy in User-Generated Mobile Health Data: The Fitbit Story
April 10: Katherine Strandburg — ECPA Reform; Catherine Crump: Cotterman Case; Paula Helm: Anonymity in AA

April 3: Ira Rubinstein — Voter Privacy: A Modest Proposal
March 27: Privacy News Hot Topics — US v. Cotterman, Drones' Hearings, Google Settlement, Employee Health Information Vulnerabilities, and a Report from Differential Privacy Day

March 6: Mariana Thibes — Privacy at Stake, Challenging Issues in the Brazillian Context
March 13: Nathan Newman — The Economics of Information in Behavioral Advertising Markets
February 27: Katherine Strandburg — Free Fall: The Online Market's Consumer Preference Disconnect
February 20: Brad Smith — Privacy at Microsoft
February 13: Joe Bonneau  — What will it mean for privacy as user authentication moves beyond passwo
February 6: Helen Nissenbaum — The (Privacy) Trouble with MOOCs
January 30: Welcome meeting and discussion on current privacy news

Fall 2012

December 5: Martin French — Preparing for the Zombie Apocalypse: The Privacy Implications of (Contemporary Developments in) Public Health Intelligence
November 7: Sophie Hood — New Media Technology and the Courts: Judicial Videoconferencing
November 14: Travis Hall — Cracks in the Foundation: India's Biometrics Programs and the Power of the Exception

November 28: Scott Bulua and Catherine Crump — A framework for understanding and regulating domestic drone surveillance

November 21: Lital Helman — Corporate Responsibility of Social Networking Platforms
October 24: Matt Tierney and Ian Spiro — Cryptogram: Photo Privacy in Social Media
October 17: Frederik Zuiderveen Borgesius — Behavioural Targeting. How to regulate?

October 10: Discussion of 'Model Law'

October 3: Agatha Cole — The Role of IP address Data in Counter-Terrorism Operations & Criminal Law Enforcement Investigations: Looking towards the European framework as a model for U.S. Data Retention Policy
September 26: Karen Levy — Privacy, Professionalism, and Techno-Legal Regulation of U.S. Truckers
September 19: Nathan Newman — Cost of Lost Privacy: Google, Antitrust and Control of User Data