As digital platforms increasingly rely on automated tools to govern user expression, an important practical question is whether algorithmic moderation can improve content quality without undermining user cooperation. Drawing on Wikipedia’s bot-based enforcement of neutrality rules, we find an unintended consequence: Contributors whose prior edits are moderated often respond with more politically slanted subsequent expression, rather than moving closer to neutrality. This pattern is stronger when moderation targets a contributor’s focal area of attention, among contributors with stronger prior political bias, and after repeated bot intervention. It is weaker when moderation occurs outside the contributor’s focal area and among contributors with greater experience in politically sensitive topics. Together, these findings suggest that effective platform governance requires more than scalable automated enforcement. For platform leaders, the results underscore the value of pairing bots with transparent explanations, context-sensitive messaging, and human oversight. For policymakers, the study indicates that algorithmic content governance should be evaluated not only by its ability to remove problematic content, but also by its downstream effects on user behavior, participation, and polarization. Well-designed governance systems must balance rule enforcement with users’ sense of autonomy.
Gu et al. (Tue,) studied this question.