Haters Gonna Hate? How Removing Downvote Option Impacts Discussions on Online Forum
40 Pages Posted: 5 May 2021 Last revised: 18 Apr 2022
Date Written: April 15, 2022
Social technologies have brought enormous changes to how consumers interact with businesses as well as how individuals interact with one another. A key component of social technologies is peer evaluation where feedback for user-generated content is provided through dynamic social interactions. Several social media platforms have experimented with different versions of peer evaluation systems in recent years. A natural question that arises is how the design of a peer evaluation system affects user behavior and engagement which ultimately has substantial implications for the long-term sustainability of platform operations. In this study, we examine the effect of removing the downvote option on users' content-generating behaviors using data collected from a popular North American online discussion forum. Platform-level analyses reveal that after the removal of this option, users became more active on the platform in both creating new content and posting replies. Moreover, while the average sentiment of the replies did not show a significant change, the average toxicity of the replies decreased significantly. More granular analysis of replies' topics shows that the topic distribution of the replies became more diverse and that users posted replies that were richer in information content after the policy change. Overall, the platform-level analysis results suggest that removing the downvote option had a positive impact on the platform's discussion culture. Through user-level analyses, we further establish that the effect of the policy change was stronger on newer users, and that heavy users became more toxic than light users after the policy change. Actively managing toxic behaviors as opposed to relying solely on the current reporting system may further improve the platform's discussion environment.
Keywords: Online platform, peer evaluation, negative evaluation, user-generated content
Suggested Citation: Suggested Citation