Pearl Yu, "Managing Social Interactions on Two-Sided Live-Streaming Platforms"
Pearl Yu is currently a PhD candidate in Information Systems at NYU Stern School of Business, studying the creator economy and digital labor.
Research Summary:
My research seeks to understand human economic behaviors in environments with emerging technologies and following policy implications, especially relevant to the creator economy and digital labor. The creator and gig economy has developed the infrastructure, market, and monetization policies to enable individuals to become micro entrepreneurs. My research has explored the psychological dynamics of live streamer mental state transitions (understanding creator economy), how non-monetary engagement design interacts with monetary outcomes (platform reward and pricing designs), and how generative AI tools impact freelancers’ job choices and performance outcomes.
In this research project, we seek to understand how does audience engagement influence the content creators’ mental states and investigate content creators’ social interaction decisions.
Featuring synchronous interaction between content creators and audiences, the live-streaming industry has seen remarkable growth. While previous research has underscored the importance of social interaction in value creation for content creators and businesses, the costs associated with such interactions have received less attention. In this paper, we study how viewer engagement, particularly through live chats, influences the dynamics of streamers' interaction efforts. We employ a heterogeneous Hidden Markov Model (HMM) to quantify the dynamics of streamers' interaction effort decisions and estimate the model using data from Twitch.com, a leading live-streaming platform renowned for its commitment to streamer development.
Our findings suggest that streamers strategically navigate the balance between high interaction levels and potential burnout. Specifically, we identify two streamers' hidden mental states: a normal state and an exhaustion state characterized by reduced responsiveness to viewer chats. We find that higher interaction efforts facilitate transitions from the normal to the exhaustion state, and decrease the likelihood of returning to the normal state once exhausted. Additionally, the impact of social interaction on exhaustion is more pronounced among streamers with fewer loyal community members and those with lower status on the platform.
This study provides practical recommendations for digital platforms to acknowledge the efforts and challenges faced by content creators, enhance viewer engagement structures and support the growth and sustainability of live streaming. By bridging the gap between digital platform management and user engagement, this study contributes to the understanding of digital labor in the context of live streaming.