Today's social knowledge is produced principally through privately controlled data extraction whose goal is to discriminate between social actors for economic advantage. Algorithms help “fram[e] problems” and provide the classifications on which broader social differentiations are constructed. But “many classifications are now embedded in markets,” a notable example being credit rating systems. Data-driven social classifications have profound consequences for the fairness or otherwise of the social world. Indeed, various entrenched features of the new social knowledge incline it toward enhancing not justice but inequality.

Nick Couldry & Ulises A. Mejias, The Costs of Connection, loc. 443