To What Extent Do Developers Tolerate Toxicity in Video Game Projects in OSS
收藏资源简介:
Open-source software (OSS) projects thrive on voluntary collaboration. Such collaboration typically occurs through public interactions such as issue reports, pull requests, discussion comments, and code reviews. While these interactions are essential for coordination, they can also incubate toxic communication. Although toxicity in OSS communities has been widely documented, we still lack a clear understanding of its temporal evolution and its specific impact on developer retention, particularly within the unique, cross-disciplinary ecosystems of open-source game development. In this paper, we present the first longitudinal analysis of toxicity across 99 GitHub-based game development projects. In particular, we use survival analysis to examine the incubation period of toxicity and subsequent developer dropout following targeted toxic exposure. Our project-level Kaplan–Meier analysis reveals that high-activity projects (those with 24+ pull requests per month) experience an accelerated timeline for toxicity, with 65.63% encountering their first toxic interaction within a year. At the developer level, we observe that a substantial portion of contributors disengage shortly after their first targeted exposure to toxicity, with 11.8% dropping out immediately and 23.3% within 30 days. A Cox proportional hazards model further indicates that developers with longer tenure and greater prior contribution history are slightly less likely to disengage after such exposure, whereas contributors with less experience in the project are more likely to leave following targeted toxicity. Our findings highlight the importance of early moderation practices, clear contribution guidelines, and proactive support for newcomers to sustain healthy collaboration in open-source game development communities.
开源软件(Open-source Software, OSS)项目依托自愿协作蓬勃发展。此类协作通常通过公开互动开展,包括议题报告、拉取请求(pull request)、讨论评论与代码审查等形式。尽管此类互动对于协作协调至关重要,但同时也可能滋生有毒沟通行为。尽管开源软件社区中的有毒沟通行为已被广泛记录,但我们仍未能清晰理解其随时间的演化轨迹,以及其对开发者留存率的具体影响——尤其是在开源游戏开发这一独特的跨学科生态系统中。 本研究首次针对99个基于GitHub的游戏开发项目开展有毒沟通行为的纵向分析。具体而言,我们采用生存分析方法,探究有毒沟通行为的孕育周期,以及开发者在遭遇针对性有毒沟通后随即发生的流失情况。我们的项目级Kaplan–Meier分析显示,高活跃度项目(月均拉取请求数≥24的项目)的有毒沟通行为出现时间线有所提前:其中65.63%的项目在成立一年内就出现了首次有毒互动。在开发者层面,我们观察到相当一部分贡献者在首次遭遇针对性有毒沟通后不久便脱离项目:11.8%的贡献者随即退出,23.3%则在30天内离开。Cox比例风险模型(Cox proportional hazards model)进一步表明,在项目中任职时长更长、过往贡献记录更多的开发者,在遭遇有毒沟通后脱离项目的概率略低;而项目经验较少的贡献者则更有可能在遭遇针对性有毒沟通后离开。本研究结果凸显了早期审核机制、清晰的贡献指南以及为新参与者提供主动支持的重要性,这对于维持开源游戏开发社区的健康协作至关重要。



