CAREER: Auditing Online Choice Architecture for Antitrust Effects

This project will develop a science of online choice architecture (OCA) audits, inspired by recent and impactful independent algorithm audits, and conduct OCA audits to inform antitrust law and public policy. Legislators and regulators worldwide are scrutinizing "self-preferencing," where a business uses dominance in a market to give its products and services an advantage in an adjacent market, often by nudging user choices in ways that can diverge from user preferences. OCA audits will use randomized controlled trials that vary user interface elements in real-world settings, pinpointing a range of effects with causal inference and external validity. Carrying out OCA audits of user interfaces will provide much-needed empirical evidence for antitrust policymakers, and the project will also develop tools and methods for collecting additional evidence. The research will bridge and contribute to the human-computer interaction and antitrust literatures by achieving three main research goals: (1) Develop new tools and methods for auditing online choice architecture, improving an existing toolkit for building browser-based studies to enable audits of web user interfaces and creating a new toolkit for implementing audits of operating system user interfaces. (2) Evaluate methods for recruiting participant panels into OCA audits, accounting for ethics, representativeness, efficacy, and cost. (3) Conduct OCA audits to quantitatively estimate the effects of self-preferencing user interfaces in the search, e-commerce, and browser markets. The search engine audits will study choice ballots, default settings, confirmation dialogs, and search results. The e-commerce audits will examine search results, recommended products, private-label products, and buy boxes. The browser audits will study choice ballots and default settings. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.