Original title: Community corrections have divergent downstream effects across corrected accounts
Yuwei Chuai, Thomas Renault, Nicolas Pröllochs, Gabriele Lenzini +1 · August 2026
Fact-checking can stop bad posts from spreading, but does it make people post better in the future? Researchers looked at the Community Notes system on X to find out. This system lets everyday users attach warning notes to misleading posts. The researchers tracked what users posted before and after getting a note. Overall, getting a note led to a small increase in posting, but people reacted in opposite ways. Users who were corrected only once posted less often, used less toxic words, and shared less misleading content. But users who got corrected over and over did the opposite. They posted more often, shared more political content, used worse sources, and posted more highly misleading content. Warning notes fix single posts without improving the worst repeat posters.
Imagine you are teaching a friend to cook by pointing out when they add the wrong ingredient. If they rarely cook, they take the hint and improve. But if they cook often and you correct them every time, they might get frustrated and start adding even more wrong ingredients. Here, the community notes on X act like your corrections, and the repeatedly corrected accounts are like the friend who cooks often and reacts by getting worse.
This matters because platforms like X invest in community fact-checking to reduce misinformation, but the system may be counterproductive for the accounts that produce most of it. The 28% of accounts that are repeatedly corrected generate nearly three-quarters of fact-checked posts, and they respond by posting more toxic and misleading content. This means correcting content and changing user behavior are separate challenges, requiring different design approaches.
The abstract doesn't explicitly state limitations, but the quasi-experimental design means causal claims are tentative, and the four-week observation period may not capture long-term effects. The study does not explain why repeatedly corrected accounts react negatively, only that they do.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Community corrections have divergent downstream effects across corrected accounts — Yuwei Chuai, Thomas Renault, Nicolas Pröllochs et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.