Get the latest tech news
Persistent interaction patterns across social media platforms and over time
Long conversations online consistently exhibit higher toxicity, yet toxic language does not invariably discourage people from participating in a conversation, and toxicity does not necessarily escalate as discussions evolve.
When exposed to such uncivil language, users are more likely to interpret these messages as hostile, influencing their judgement and leading them to form opinions based on their beliefs rather than the information presented and may foster polarized perspectives, especially among groups with differing values 22. Our findings show that, across all datasets, the participation of users in the evolution of conversations, averaged over almost all considered threads, is decreasing, as indicated by the results of Mann–Kendall test—a nonparametric test assessing the presence of a monotonic upward or downward tendency—shown in Extended Data Table 1. Our results, showing user resilience to toxic content, indicate the potential for creating advanced, context-aware moderation tools that can accurately navigate the complex influence of antagonistic interactions on community engagement and discussion quality.
Or read this on Hacker News
