Game for Brainstorm: The Impact of a Badge System on Knowledge Sharing

Gamification has been widely adopted to engage users in various domains. To improve the effectiveness of gamification systems, we propose a generic framework to design and fine-tune a gamification system. Our study focuses on a badge system and considers three major design elements, namely, volume (the number of badges), variety (the number of badge categories), and valence (high-tier badges). We characterize the dynamics of user-system interactions on Stack Overflow and conduct policy experiments using a structural hidden Markov model (HMM). We sharpen our empirical analyses by incorporating Gaussian copulas into our HMM to address potential endogeneity from badge elements. Our HMM-copula model quantifies both the short-term and long-term impacts of the three design elements on the corresponding user knowledge contributions and engagement-state transitions. Our results demonstrate that the badge system encourages the quantity and quality of user contributions. These positive impacts vary across the four engagement states: Inactive , Gentle , Active , and Vigorous . Specifically, high badge volume and valence keep users staying or moving up to higher engagement states, whereas high badge variety discourages them from sharing more. More importantly, we use the individual structural parameter estimates to run counterfactual experiments to understand the impacts of badging strategies. Overall, this study provides a novel design aspect to contribute to the gamification literature, proposes a generic copula approach to address endogeneity in HMM, and delivers actionable implications for platform managers and gamification system designers.