Why Conspiracy Theory Communities Endure: The Interplay of Feedback, Habits, and Community Context

Abstract

Online conspiracy theory (CT) communities pose challenges for digital governance, yet the behavioral mechanisms sustaining their participation remain unclear. We analyze 2.1 million posts from 2,711 users across 4,435 subverses on Voat, an alternative platform with minimal moderation. Using survival analysis and hierarchical mixed-effects models, we show that habitual posting reduces sensitivity to social feedback, decoupling participation from the upvotes and downvotes that users receive. The type of content that users contribute varies in its effect: while group-oriented CT signals attenuate community engagement, trait-oriented signals, such as allegations of secrecy, have little effect. While posting habit is associated with disengagement from subverses in general, ideological alignment mitigates this effect in CT subverses. Together, these results reveal a multi-scale framework in which social reinforcement, routine participation, and community context interact to sustain CT engagement. By isolating mechanisms beyond CT alignment, we explain how fringe communities persist despite volatile rewards, providing new insights into the structural resilience of CT spaces.