The Power of Social Norms: How Initial Responses to Toxicity Shape Conversations on Twitter

A new study published on arXiv by researchers including Ana Aleksandric and Mohit Singhal reveals how the first response to a toxic tweet can determine the trajectory of an entire conversation. Analyzing a random sample of 187,000 tweets across approximately 9,000 conversations, the researchers found compelling evidence for social norm contagion in digital spaces. The study specifically examined two key factors: the number of users present before toxicity appears (bystander effect) and the toxicity level of the first direct reply.

The research discovered that larger conversation groups before a toxic incident actually reduced the likelihood of non-toxic responses, suggesting a diffusion of responsibility. More significantly, when the first reply to a toxic tweet was itself toxic, it established a harmful norm that increased toxicity in subsequent replies by approximately 40%. This pattern held across different types of conversations and demonstrates how quickly antisocial behavior can become normalized in online interactions.

These findings have direct implications for platform design and moderation strategies. The researchers argue that understanding these dynamics is crucial for developing interventions that encourage positive initial responses to toxicity, potentially preventing escalation. The study will be presented at the 7th International Workshop on Cyber Social Threats at ICWSM 2026, highlighting its relevance to current social media challenges.