Against Online Abuse and Toward Sociotechnical Security & Privacy

Abstract

Technology-facilitated abuse is an escalating challenge---from toxic content on social media to image-based sexual abuse---with a significant impact on the well-being of people globally. This dissertation advocates for a sociotechnical approach to computer security and privacy to combat technology-facilitated abuse. Sociotechnical approaches investigate systems of people and technology and require studying the complexities of their interconnections. In the first part of this dissertation, I evaluated existing mitigations for online abuse, finding that they place an undue burden on individuals and have limits in addressing the systemic challenges of online abuse. In the second part, I next analyzed human-centered security and privacy research and identified opportunities for the increased consideration of social factors, such as through operationalizing sociodemographic factors or gender stereotypes. Finally, in the third part, I applied sociotechnical threat modeling to characterize two emerging forms of online abuse: interpersonal surveillance and control and synthetic nonconsensual explicit imagery (also called "deepfake nudes"). Throughout this dissertation, I also study how gender and interpersonal relationships contribute to disparate experiences of safety. Theoretically, this dissertation connects computer science inquiry with feminist science and technology studies (feminist STS) and sociology; this enables supplementing inductive descriptions of online abuse with social theories of how and why online abuse happens. Practically, this dissertation integrates people's lived experiences and social perspectives into security and privacy research, informing researchers, policymakers, and industry practitioners on more sociotechnically informed mitigations for online abuse. This dissertation underscores the transformative potential of a sociotechnical approach to security and privacy not only for mitigating online abuse but also for security, privacy, and safety at large.