Shocking the Crowd: The Effect of Censorship Shocks on Chinese Wikipedia

Abstract

Collaborative crowdsourcing has become a popular approach to organizing work across the globe. Being global also means being vulnerable to shocks — unforeseen events that disrupt crowds — that originate from any country. In this study, we examine changes in collaborative behavior of editors of Chinese Wikipedia that arise due to the 2005 government censorship in mainland China. Using the exogenous variation in the fraction of editors blocked across different articles due to the censorship, we examine the impact of reduction in group size, which we denote as the shock level, on three collaborative behavior measures: volume of activity, centralization, and conflict. We find that activity and conflict drop on articles that face a shock, whereas centralization increases. The impact of a shock on activity increases with shock level, whereas the impact on centralization and conflict is higher for moderate shock levels than for very small or very high shock levels. These findings provide support for threat rigidity theory — originally introduced in the organizational theory literature — in the context of large-scale collaborative crowds.

Participation of New Editors after Times of Shock on Wikipedia

“…While we interpret reverts as negative feedback, other studies have associated reverts with conflict and controversy (Kittur et al 2007;Kittur and 2010;Zhang et al 2017b;2017a). One key difference is that we focus on reverts to article newcomers, who due to their lack of expertise, are more likely to learn about the norms and rules of the article by being reverted than other more seasoned editors.…”

“…Unfortunately, we know little about the impact of shocks on crowd future participation. This is despite the fact that crowds operate in a rapidly changing environments often impacted by external events with little forewarning (Zhang et al 2017a). The lack of formal boundaries and the ability of crowd members to leave at any time means that crowd participation are likely to be particularly susceptible to shocks (Robert and Romero 2017).…”

“…Theory suggests competing hypotheses for the effect of attention shocks on centralization -the extent to which an article is controlled and produced by a small fraction of editors. Threat rigidity would suggest that crowds would react to attention shocks by becoming more centralized (Staw, Sandelands, and Dutton 1981;Zhang et al 2017a). Yet, an influx of new article editors unfamiliar with work practices and norms would make it difficult to maintain the current social order (Buechler 2013;Useem 1998), which involves maintaining a highly centralized structure (Romero, Huttenlocher, and Kleinberg 2015;Kittur and Kraut 2008).…”

“…Centralization. Similar to related work (Zhang et al 2017a;2017b), we measure centralization of workload in a group using the normalized Gini coefficient of the distribution of number of revisions contributed by each editor. The Gini coefficient measures the level of inequality in a distribution (Dorfman 1979).…”

“…We analyze three types of collaborative features that are associated with levels of participation of members of online communities (Zhu et al 2013a;Halfaker et al 2013). Specifically, we focus on 1) negative feedback measured by reverts-revisions that undo other ed-itors' contributions (Zhu et al 2013a;Halfaker et al 2013), 2) centralization of contributions as measured by distribution of editor efforts (Zhang et al 2017a;2017b), and 3) levels of coordination measured by discussions on talk pages (Romero, Huttenlocher, and Kleinberg 2015). Through mediation analysis, we quantify the direct and indirect effects of shocks and identify specific collaborative features that lead to higher newcomer participation.…”