Platforms as laboratories of the social: How digital capitalism matters for computational social research in North America.

Abstract

The contemporary prevalence of artificial intelligence and machine learning methods has resulted in a rich literature on the factors that shape computational research. This article draws on the laboratory studies literature to examine how platforms' socio-technical infrastructures shape contemporary computational social science research. Based on 18 months of online ethnography of a university laboratory and 15 in-depth interviews with its researchers, the article makes two main arguments. First, for computational social sciences, platforms function as laboratories where the social is selectively carved and transformed, to make it knowable with computational methods. Thus, it makes the case that platforms manufacture the objects of analysis in computational social research and provide the social as a domain. Second, because of the significance of social media platforms as data laboratories for computational research, in contrast to the claims of data sciences to be domainless, these sciences may derive some of their epistemological and occupational power, as well as their cultural authority, from digital capitalism.