Algorithmic Trust: Aversion vs. Acceptance in Programmatic Advertising

• In Event: Algorithms at Work: The Practice of Prediction - II: Case Studies of Prediction in Practice

Abstract

While advocates for algorithmic decision-making routinely argue that algorithms produce impartial and objective decisions, critics point to instances of biased and discriminatory outcomes. This dissent has focused on the inherence of bias during the creation of algorithms and within algorithmic systems themselves. More recently have scholars begun investigating the meanings, understandings, and interpretations held by professionals that use “algorithms in practice” (Christin 2017). Drawing on ethnographic research in the field of programmatic advertising, this paper contributes to the growing literature on algorithms in professional and everyday life. In this rapidly growing field, advertisements are sold through an auction marketplace where “traders” input bids into an algorithmically determined platform. Data on each campaign are fluid; traders receive performance feedback and adjust their bids to attain more clicks, increase ad views, and achieve sales goals. I show that when campaigns are optimized and bids are providing maximum returns, traders increase trust in the technology. Utilizing a conceptual framework based on Goffman’s interaction order, I argue that algorithmic trust is situational. Situational knowledge and understandings of the advertising marketplace reduce “algorithmic aversion” and increase “algorithmic acceptance.” This presents a counter to scholarship that predominantly frames algorithms as points for doubt and resistance.

Author

• Adam Peri, University of Chicago