After Deplatforming: The Return of Trace Research for the Study of Content Moderation

When platforms were considered ‘intermediaries', or conduits through which content flows unfettered (Kleis Nielsen and Ganter, 2018), they could serve as sites for the study of user traces: online actions and behaviours interactively registered by the platforms and made available to researchers in the form of hit logs, links, likes, retweets, shares and so forth. One could observe user behaviour and attitude unobtrusively, as if ‘in the wild’ (Webb et al., 1966). These measures would provide insights into collective mood and sentiments, give indicators of opinion and perhaps even capture animal spirits behind stock price or currency movements (Lazer et al., 2008, 2020; Watts and Dodds, 2007).

But there is a certain artificiality to platform data that results from platform as well as user actions. On the platform side, there are attempts to retain users with ‘sticky content’ to increase revenue tied to ‘watch time'; there are even ‘dark patterns’ underlying user interfaces that steer users towards forms of conversion. While certain content is promoted, other kinds are demoted. Deplatforming, labelling and downranking are content moderation techniques that adversely affect the visibility of platform data.