Setting the Record Straighter on Shadow Banning

Abstract

Shadow banning consists for an online social network in limiting the\nvisibility of some of its users, without them being aware of it. Twitter\ndeclares that it does not use such a practice, sometimes arguing about the\noccurrence of "bugs" to justify restrictions on some users. This paper is the\nfirst to address the plausibility or not of shadow banning on a major online\nplatform, by adopting both a statistical and a graph topological approach. We\nfirst conduct an extensive data collection and analysis campaign, gathering\noccurrences of visibility limitations on user profiles (we crawl more than 2.5\nmillion of them). In such a black-box observation setup, we highlight the\nsalient user profile features that may explain a banning practice (using\nmachine learning predictors). We then pose two hypotheses for the phenomenon:\ni) limitations are bugs, as claimed by Twitter, and ii) shadow banning\npropagates as an epidemic on user-interactions ego-graphs. We show that\nhypothesis i) is statistically unlikely with regards to the data we collected.\nWe then show some interesting correlation with hypothesis ii), suggesting that\nthe interaction topology is a good indicator of the presence of groups of\nshadow banned users on the service.\n