Fewer Reviews but Better Content: When an Online Review Platform Disables Downvotes
研究了一个餐厅评论平台禁用差评功能后的影响，发现评论数量减少但质量提升，为平台设计平衡公平、动机和内容价值的评价系统提供了启示。
Most user-generated content platforms utilize peer evaluation systems that incorporate upvotes and downvotes to distinguish high-quality from low-quality content. However, downvotes are often misused to target others and discourage content creators, and some platforms, such as Amazon and TripAdvisor, have removed the downvote option altogether. Meanwhile, the real-world impact of this platform-level intervention on user contributions remains underexplored, and we fill this gap by analyzing a restaurant review platform that disabled downvotes. We apply the regression discontinuity in time (RDiT) method and find that removing downvotes reduces the quantity of reviews but increases their quality. Further mechanism analysis suggests that the former may stem from perceived unfairness, and the latter from a motivation of continuing reviewers to stand out. Additionally, our heterogeneity analysis shows that the decline in review quantity is stronger among long-tenure reviewers, whereas the improvements in multiple quality dimensions are concentrated among reviewers with fewer followers. Our study reveals the nuanced behavioral effects of a platform-level design change and offers practical insights for designing peer evaluation systems that balance fairness, motivation, and content value.