Stanford Study: X's Algorithm Prioritizes Anger-Inducing Content to Boost Engagement

A Stanford University study published in PNAS reveals that X's recommendation algorithm prioritizes content that conflicts with users' values, triggering anger to drive higher engagement. This dynamic creates a feedback loop where negative reactions prompt the system to recommend similar provocative content. This study highlights how engagement-optimization algorithms can inadvertently polarize public discourse by amplifying outrage. It underscores the need for greater transparency in social media algorithms, which shape public opinion without public oversight. The study tracked 715 US users using a browser extension to analyze their feeds alongside surveys based on the Schwartz theory of basic values. While X's former product head Nikita Bier claimed the platform has since reduced anger-inducing recommendations by an order of magnitude, the study confirms the algorithm's strong bias toward outrage during the research period.