Most online content platforms thrive on active user participation. However, new research from the University of Miami’s Patti and Allan Herbert Business School suggests that, in some cases, too much engagement can be detrimental.

Kevin Hong, associate dean for research, Centennial Endowed Chair, and professor of business technology, has found that on synchronous content platforms such as Twitch and YouTube Live, larger audiences may discourage rather than encourage interaction.

“When there are a lot of people participating, conversations often fragment, with multiple discussions happening at once,” Hong explained. “This creates a chaotic environment that can drive viewers away. Imagine an online class where every student is shouting questions in the Zoom chat at the same time—it quickly becomes unmanageable.”

Hong’s study, Lost in the Crowd: How Group Size and Content Moderation Shape User Engagement in Live Streaming, analysed data from 7,074 Twitch playbacks and their associated chat histories, a project that took years to complete. The research, recently published in Information Systems Research, was co-authored with scholars from Pennsylvania State University, the University of South Florida, the University of Illinois at Chicago, and Ohio State University.

The findings suggest that large live-stream audiences not only create cognitive overload but also foster emotional polarisation, which further reduces the willingness of viewers to participate. This contrasts sharply with earlier studies on asynchronous platforms such as Yelp, Reddit, and Wikipedia, where a larger number of participants typically correlates with increased engagement.

“Researchers have studied asynchronous platforms for decades, where one person’s activity rarely affects another’s experience,” Hong noted. “Synchronous platforms are relatively new, and the evidence indicates that big crowds can have a negative impact. The question is: how do we mitigate that?”

One promising solution is the use of AI-powered moderation tools. While human moderators can handle smaller surges in participation, they struggle when hundreds of people are speaking at once. “Bots, particularly those using artificial intelligence, can scale rapidly and maintain lower costs,” Hong said. “Although their capabilities are still limited, AI moderators are improving every day. They have significant potential to organise conversations and make them more coherent.”

More information: Keran Zhao et al, Lost in the Crowd: How Group Size and Content Moderation Shape User Engagement in Live Streaming, Information Systems Research. DOI: 10.1287/isre.2022.0086

Journal information: Information Systems Research Provided by University of Miami