Thread With Caution: Proactively Helping Users Assess and Deescalate Tension in Their Online Discussions

Incivility remains a major challenge for online discussion platforms, to such an extent that even conversations between well-intentioned users can often derail into uncivil behavior. Traditionally, platforms have relied on moderators to---with or without algorithmic assistance---take corrective actions as removing comments banning users. In this work we propose complementary paradigm directly empowers by proactively enhancing their awareness about existing tension in the conversation they are engaging and actively guides them drafting replies avoid further escalation. As proof of concept paradigm, design tool provides proactive information users, conduct user study popular platform. Through mixed methods approach combining surveys with randomized controlled experiment, uncover qualitative quantitative insights regarding how participants utilize react information. Most report finding valuable, noting it helps identify may otherwise missed prompts reflect own revise them. These effects corroborated comparison draft reply when our warns is at risk derailing behavior versus control condition where disabled.These preliminary findings highlight potential user-centered point concrete directions future implementations.