Economic incentives drive algorithmic choices that increase polarization, according to research revealing tensions between platform profitability and democratic health. While down-ranking divisive content reduced political animosity, it also slightly decreased engagement metrics that determine advertising revenue, creating financial pressures that discourage healthier algorithms.
The study examined X during the 2024 US presidential election, a period when divisive content proliferated and engagement soared. Researchers manipulated what over 1,000 users saw in their feeds, increasing divisive content for some participants while decreasing it for others. The goal was to understand both psychological and business implications of different algorithmic approaches.
Results confirmed that divisive content drives engagement. Posts expressing anti-democratic attitudes and partisan hostility provoke strong emotional reactions that keep users scrolling, viewing more posts, and spending more time on platforms. These metrics directly translate to advertising impressions and revenue, creating powerful financial incentives to optimize algorithms for maximum controversy rather than constructive dialogue.
However, the research also revealed nuance in engagement patterns. While down-ranking divisive content slightly reduced time spent on the platform and number of posts viewed, users actually liked and reposted content more frequently. This suggests that healthier algorithms might foster deeper, more meaningful engagement even if reducing overall scrolling time.
The findings illuminate a critical choice facing platform owners. Current business models optimize for engagement volume, which divisive content effectively generates. But algorithms could instead be designed to promote quality engagement that builds community rather than fueling conflict. Whether platforms will voluntarily make this shift or whether regulatory intervention might be necessary remains an open and urgent question.