Changes in YouTube's Content Moderation Policy Had Little Detectable Impact on Election Denial Content

Abstract

While a growing body of research examines the effect of deplatforming, the blanket banning of certain types of content and users, on social media discourse, comparatively little research examines the effect of replatforming, when social media companies remove such blanket bans. We exploit such a policy change, YouTube’s June 2nd decision to stop removing content denying the validity of the 2020 US Presidential Election, to examine the effect of replatforming content. Using data from YouTube, Facebook, Telegram conspiracy groups, and Google, we find no evidence that YouTube's policy change increased demand for or supply of election denial content over the short term. These results are consistent across all platforms and three different methodologies: regression discontinuity models, structural topic models, and a Bayesian structural time series model. This suggests that replatforming alone, when conducted after the effective marginalization of the targeted content, has minimal short-term effects on the spread of previously-prohibited content