Knowing User Populations at Scale: From the Science of the State to Platform Governmentality

Date:

Talk at 2018 Annual Conference of the International Communication Association, Prague, Czech Republic

How can corporate institutions that own and operate large-scale social media platforms come to know “their users” at scale? In this talk, I discuss ways of knowing user populations at scale, drawing on Michel Foucault’s historical account of governmentality, particularly the role of statistics in the formation of the modern nation state. Like with early modern statistics, data-intensive computational methods for representing users at scale are frequently justified as “good governance” — framed as necessary obligations to let users thrive in a world filled with spam, misinformation, hate speech, abuse, and information overload. Yet such automated ways of knowing and acting at scale often fall short, with many high-profile controversies facing socal media sites, often accompanied by social movements organized against platforms’ data-driven decisions. I argue that it is necessary to consider the role that statistics and data science plays within the organizational structure of the institutions that own and operate social media platforms: what is needed to know users at scale, and what is needed to know that particular ways of knowing at scale are correct? While public controversies around automated enforcement of platform policies are often seen as acts of resistance against the power of platforms, Foucault’s history reminds us that power and resistance go hand in hand. Such controversies can also been seen as alternative ways in which the institutions behind the platform come to know to user populations at scale – when the controversies gain such size and/or influence that their concerns are made visible to those in high-level positions within these companies. These two kinds of cases illustrate different ways in which voices must be mediated, aggregated, circulated, represented and rationalized to be made actionable in these platforms’ internal structures.