More Like this

1. Against Engagement

   Richards, Neil; Hartzog, Woodraw (July 2024, Boston University Law Review)

   n this Article, we focus on a key dimension of commercial surveillance by data-intensive digital platforms that is too often treated as a supporting cast member instead of a star of the show: the concept of engagement. Engagement is, simply put, a measure of time, attention, and other interactions with a service. The economic logic of engagement is simple: more engagement equals more ads watched equals more revenue. Engagement is a lucrative digital business model, but it is problematic in several ways that lurk beneath the happy sloganeering of a “free” internet Our goal in this Article is to isolate engagement as a distinct and dangerous concept that should be specifically regulated. There is a benefit to seeing past the glib justificatory rhetoric and taking a hard look at engagement-based, surveillance-advertising-funded models as potentially problematic. Unfettered engagement strategies bear significant and under-appreciated costs that are endangering our privacy, our democracy, and our culture itself. It’s time that wrongful engagement, and the asserted “free” business models it generates, started to bear the burden of those costs. 

   more » « less
2. Voting Against Autocracy

   https://doi.org/10.1353/wp.2023.a908772  

   Svolik, Milan W. (October 2023, World Politics)

   When and how do voters punish politicians for subverting democracy? To investigate the role of the public in democratic backsliding, I develop a conceptual framework that differentiates among three mechanisms: vote switching, backlash, and disengagement. The first mechanism entails defection by voters from a candidate who undermines democracy to one who does not; the latter two mechanisms entail transitions between voting and abstention. I estimate the magnitude of each mechanism by combining evidence from a series of original survey experiments, traditional surveys, and a quasi-experiment afforded by the rerun of the 2019 Istanbul mayoral election, in which the governing party, akp, attempted to overturn the result of an election that it had lost. I find that although vote switching and backlash contributed to the akp's eventual defeat the most, each of the three mechanisms served as a democratic check in some subset of the Istanbul electorate. Persuasion, mobilization, and even demobilization are all viable tools for curbing the authoritarian tendencies of elected politicians. 

   more » « less
3. Co-making against antiBlackness

   https://doi.org/10.1080/10665684.2023.2222552  

   Roby, ReAnna S.; Calabrese Barton, Angela; Tan, Edna; Greenberg, Day (July 2023, Equity & Excellence in Education)
4. Data Activism Against Feminicide

   Cruxên, Isadora; JungsDeAlmeida, Alessandra (August 2025, Queen Mary University)
5. Model Cloaking against Gradient Leakage

   https://doi.org/10.1109/ICDM58522.2023.00182  

   Wei, Wenqi; Chow, Ka-Ho; Ilhan, Fatih; Wu, Yanzhao; Liu, Ling (December 2023, IEEE International Conference on Data Mining (ICDM))

   Gradient leakage attacks are dominating privacy threats in federated learning, despite the default privacy that training data resides locally at the clients. Differential privacy has been the de facto standard for privacy protection and is deployed in federated learning to mitigate privacy risks. However, much existing literature points out that differential privacy fails to defend against gradient leakage. The paper presents ModelCloak, a principled approach based on differential privacy noise, aiming for safe-sharing client local model updates. The paper is organized into three major components. First, we introduce the gradient leakage robustness trade-off, in search of the best balance between accuracy and leakage prevention. The trade-off relation is developed based on the behavior of gradient leakage attacks throughout the federated training process. Second, we demonstrate that a proper amount of differential privacy noise can offer the best accuracy performance within the privacy requirement under a fixed differential privacy noise setting. Third, we propose dynamic differential privacy noise and show that the privacy-utility trade-off can be further optimized with dynamic model perturbation, ensuring privacy protection, competitive accuracy, and leakage attack prevention simultaneously. 

   more » « less