A feed-ranking algorithm that promotes posts in which members of one political group express views usually associated with the other group (e.g., a conservative account voicing a liberal position.)

The goal is to interrupt caricatured views and misperceptions about political opponents by showing the other side is less uniform than stereotypical depictions suggest, and, in turn, reduce political animosity.

Stray et al. (2026) tested this approach as one arm of a large independent field experiment across Reddit, Facebook, and X during the 2024 U.S. electoral cycle. They used a browser extension and a Large Language Model (GPT-4o) to identify and promote stereotype-challenging posts. The results demonstrated a small, but non-significant, reduction in partisan animosity (-0.02 standard deviations) (Note: all effects we include are statistically significant, unless otherwise stated. We report effect sizes using the metrics in the authors’ paper.) Two secondary effects did reach significance, however: users reported a worse experience on the platform (-0.06 standard deviations), but the condition also resulted in the largest increase in engagement of any algorithm tested in the study.

Stereotyping and misperceptions of political opponents have been associated with partisan animosity, and there is a long history of research proposing exposure to diverse viewpoints and counter-stereotypical examples can produce more humanized perceptions of others. Though this study offers an initial large scale test of this idea in the field, the lack of significant evidence leaves this an emergent option requiring further testing.