A feed-ranking algorithm that upranks posts and comments demonstrating markers of civility (e.g., reasoning, compassion, respect, curiosity).

By increasing the chances of users seeing more respectful, civil, content, the goal of the intervention is to lower hostility among platform users and, in this specific case, affective polarization between opposing political groups.

Stray et al. (2026) tested this approach as one arm of a large independent field experiment across Facebook, Reddit, and X, using a browser extension and Google Jigsaw’s Perspective API to score how civil each post was and then upranking posts/comments with higher civility scores. The study was carried over six months during the 2024 U.S. election cycle, and the results indicated that upranking civil content achieved a small (~0.02 standard deviations), but non-significant, reduction in partisan animosity (Note: we report effect sizes using the metrics in the authors’ paper.) It also had no significant effect on user experience, social trust, intergroup empathy, or political knowledge. Because this is the only field study testing this intervention, and it found no significant effect, the evidence is limited to a single null result.

Civil, reasoned discussion is widely valued as a building block of healthy online conversation, and upranking content with higher scores on civility markers is a relatively low-cost intervention for platforms. The current evidence, however, suggests that elevating civil content on its own may not be enough to meaningfully reduce levels of partisan animosity, at least not without also downranking hostile content (see entry on Downranking Toxic Content).