Can an Algorithm be Agonistic? Ten Scenes from Life in Calculated Publics

Crawford K. Can an Algorithm be Agonistic? Ten Scenes from Life in Calculated Publics. Science, Technology, & Human Values. 2016;41(1):77-92. doi:10.1177/0162243915589635

This paper examines algorithms through political theory, suggesting "agonistic pluralism" as a design ideal for engineers and a way of viewing algorithms more broadly. Adopting the perspective Winner presents, that artifacts are political both socially and technically and must be examined in context, and the argument that Chantal Mouffe presents, that the "agonistic struggle" of conflict is the cornerstone of a functioning democracy, Crawford examines the context of algorithms against different politics. She argues that the common view of examining outputs of algorithms in isolation "forecloses more complex readings of the political spaces in which algorithms function, are produced, and modified."

Calculated public in this piece refers to the vision of public life displayed by an algorithm as it calculates unknowable variables behind the scenes. For example, how did Amazon decide to offer specific items to you after you purchased one - what public demographic is it representing? Humans and algorithms shape these publics together, through purchases, reviews, and other such relations. As humans game algorithms, algorithms further develop to resist such gaming, and the cycle continues. Search Engine Optimization is a great example of how humans also shape what is often perceived as solely the algorithm; individuals and entire institutions have arisen to shape how the "algorithm" displays information back to users.

DiSalvo's "adversarial design" embraces the endless loop of confrontation in the design process. He believes that labeling an artifact as adversarial "shifts the grounds for critique, prompting us to look at its designed qualities and what they bring to the fore."