Feedback Design and Equilibrium Knowledge-Sharing

Abstract

Feedback systems are widely used by digital platforms. We study how feedback design shapes equilibrium content provision in the context of Stack Overflow. Using a regression discontinuity design based on answer timing, we show that competition shapes feedback and influences contribution. We develop and estimate an empirical model that links feedback design to equilibrium outcomes among heterogeneous contributors. The model allows the platform to continuously adjust leniency (the tendency to give positive feedback) and harshness (the tendency to give negative feedback). We decompose feedback’s impact into a direct effect on individual contributors and an indirect competitive effect. We find that although a contributor favors more leniency and less harshness when considered alone, she may prefer less leniency and higher harshness in equilibrium. This reversal arises from competition and depends on user quality, with higher-quality contributors being more likely to alter their preferences under equilibrium incentives. We further show how the platform’s targeted quality levels, its ability to distinguish contributor quality, and the distribution of contributor skill shape competition and inform optimal feedback design. Finally, we consider the case of endogenous feedback preferences, where contributors’ valuations of feedback depend on the equilibrium quality induced by the mechanism.

Keywords: Feedback system, Positive feedback, Negative feedback, Platform design, Knowledgesharing platforms, Competition, Oblivious equilibrium, UGC