Content advocating partisan violence represents one of the most dangerous forms of divisive posting, according to research identifying such content as a significant polarization driver. When posts expressing support for political violence were amplified in users’ feeds, animosity toward opponents increased substantially, raising urgent concerns about platform responsibility for preventing violence.
The study analyzed X posts during the 2024 presidential election for various markers of divisive content. Among the most concerning were posts advocating partisan violence—content suggesting that political opponents deserve physical harm or that violence represents a legitimate political tool. Though such posts might seem obviously problematic, engagement-optimized algorithms can amplify them if they generate strong reactions.
Researchers manipulated feeds for over 1,000 users to include slightly more or less content advocating partisan violence. Those exposed to more such content showed measurably increased polarization and political animosity. This demonstrates a direct causal link between algorithmic amplification of violence-advocating content and deterioration of democratic norms.
The implications for platform governance are serious. If algorithms systematically amplify content that promotes political violence, platforms may bear some responsibility for any real-world violence that results. While direct causation would be difficult to prove in specific cases, the research establishes that such content measurably increases political hostility and presumably the likelihood of violence.
Current content moderation policies typically prohibit direct incitement to violence but may allow subtler advocacy that falls below enforcement thresholds. Meanwhile, engagement-optimized algorithms may amplify whatever content generates reactions, potentially boosting violence-adjacent content that doesn’t quite violate stated rules. This mismatch between moderation and amplification policies deserves urgent attention.
20