Exploring algorithmic realism in the data economy

Abstract

My doctoral research develops a deeper understanding of the promises of algorithmic interventions for the data economy inspired by algorithmic realism: an algorithmic framework cognizant of political, porous, and contextual aspects of the social world. I design, evaluate, and deploy algorithmic interventions aimed as anticipatory and mitigation measures against ethical issues of different domains of the data economy using three case studies. In each case study, the contingencies and fluidity of the data economy are accounted for and embraced in the designs. Specifically, through the development and evaluation of a human-centric labeling framework for machine learning, anticipatory models of privacy preferences for the smart home, and a technology probe on transparency of profiling in online behavioral advertising, I show how algorithmic interventions can promote ethical practices, balance conflicting forces, and promote user trust in the data economy. Findings illuminate a path of ethics, opportunities for increased user participation amidst power imbalances, and mutual benefits of such interventions in light of the prevailing forces of the data economy. However, findings also reveal a number of challenges such interventions may face, mainly around feasibility, countering economic forces, and mismatched or conflicting expectations between users and service providers of the data economy. I discuss such challenges and offer future research directions around feasibility, algorithmic authority, conflicting forces, mismatched expectations, and shared accountability in highly decentralized data economy systems.