Online knowledge-sharing platforms such as Quora and Zhihu (a leading knowledge-sharing platform in China) often use both peer voting and followers to encourage user knowledge contributions. However, these platforms diverge in whether they highlight followers, upvotes, or both as reputation symbols. To better understand the consequences of these platform design choices, we conduct a field experiment with 1,696 focal users on Zhihu where we exogenously increase upvotes or/and followers for treated users over a 53-day intervention period. We monitor focal users’ activities for 303 days, covering both pre- and post-intervention periods. We further use a large language model to gauge the quality of 12,998 answers contributed by these users during this time. We find that increasing upvotes significantly boosts users’ answer contributions (in volume, total length, and quality), while increasing followers has mostly no overall effect on contributions. Additionally, increasing upvotes can encourage contribution for up to 100 days, particularly among lower-reputation (i.e., with fewer upvotes or followers), less active, female users, and those answering soft topic questions. In contrast, increasing followers only reduces contribution volume (not length or quality) for higher-reputation, more active, and male users. Moreover, while peer voting primarily only affects answer contributions, the follower treatment leads to negative spillovers, reducing users’ upvoting on fellow users’ answers, following fellow users, or purchasing Zhihu Lives hosted by fellow users. Our research suggests the possibility that users on knowledge-sharing platforms may associate peer voting with contributions, treating it as peer recognition that offers intrinsic motivation. In contrast, users may interpret additional followers as a symbol of status enhancement, which could, in some cases, dampen their motivation to contribute. Overall, our study suggests that platforms aiming to foster active, high-quality contributions should highlight upvotes rather than followers as a reputation symbol. To our knowledge, this study is among the first field experiments to identify and compare the causal effects of peer voting and followers on user contribution to knowledge-sharing platforms.
Zhang et al. (Fri,) studied this question.