Mass shooting narratives on social media: Content moderation and participatory discourse

Abstract

This study examines how mass shooting narratives emerge on social media through the interaction of platform governance and participatory discourse. Using a two-study design, we first compare moderation policies and enforcement approaches across major platforms, then analyze discourse surrounding five mass shooting events on X as visible outcomes of platform governance. For the two highest-volume events, 24,748 X posts from the Chiefs Parade shooting and 2043 X posts from the Hookah Bar shooting, we employ a retrieval-augmented generation (RAG)-based thematic analysis combining human-guided theme development with embedding-based semantic retrieval. Three lower volume events are examined through human thematic analysis, and 267 Reddit posts from the Chiefs Parade case serve as a qualitative comparative reference. Findings show that users shared breaking news, expressed grief, debated gun violence and media coverage, and interpreted events publicly. Platform governance appears associated with differences in discourse visibility, tone, and persistence.