Balancing Acts: Unveiling the Dynamics of Post Removal on Social Media User-Generated Content
Abstract
User-generated content (UGC) is pivotal in driving the adoption and engagement of platforms, influencing both users and platform owners. Despite UGC’s positive impact, users often violate platform guidelines and generate low-quality content. In response to these challenges, online platforms commonly employ content moderation strategies, including the removal of posts that violate community guidelines (e.g., creating a civil and safe community). The impact of post removal on subsequent UGC creation remains uncertain. Some studies suggest that moderation tactics, such as user bans, induce a psychological backlash, potentially leading to increased UGC quantity but a decline in quality. Conversely, the learning theory suggests that the removal of posts can serve as a disciplinary measure, encouraging users to learn and internalize community guidelines and adjust their behavior accordingly. We explore the consequences of post removal on UGC quality on Reddit. Additionally, we explore the learning processes derived from users’ own experiences of being punished, from users’ observations of others facing similar consequences, and from the underlying reasons for moderation. Our findings indicate that post removal significantly decreases users’ future rule violations and encourages the generation of higher-quality posts (as evidenced by an increase in upvotes). These results suggest that post removal, as a disciplinary measure, fosters user learning, motivating users to adhere to community guidelines and contribute superior quality content. This study’s implications extend to theoretical and practical domains and provide valuable insights for platform operators on leveraging post removal as a content moderation strategy to guide users toward producing high-quality content.
History: Ram Gopal, Senior Editor; Xitong Li, Associate Editor.
Supplemental Material: The online appendix is available at https://doi.org/10.1287/isre.2023.0750.

