On live-streaming platforms, bigger audiences don’t always mean better engagement. In fact, sudden surges in viewers can make participants less likely to speak up, unless moderation keeps the crowd in check.
A new study, published in Information Systems Research, co-authored by Tengteng Ma, assistant professor at the University of South Florida, examines what happens when group size spikes on Twitch.
Leveraging Twitch’s “Raid” feature, which unexpectedly redirects viewers from one live stream to another, the researchers analyzed more than 7,000 streaming sessions to isolate the causal impact of audience growth.
They found a clear congestion effect. When a large group suddenly joins, existing viewers are less likely to comment and post fewer messages overall. Conversations become fragmented across multiple topics, and emotional intensity in the chat rises, both of which discourage participation.
Moderation makes a difference. Automated bot moderators are particularly effective at maintaining topic coherence when large groups arrive. Human moderators, meanwhile, are better at calming emotional surges when smaller groups join.
The findings challenge assumptions drawn from asynchronous platforms, where larger audiences often increase engagement. In live, synchronous settings, scale without structure can undermine the user experience.
Authors: Keran Zhao, Pennsylvania State University; Yili Hong, University of Miami; Tengteng Ma, University of South Florida; Yingda Lu, University of Illinois Chicago; Yuheng Hu, Ohio State University.