Flag and Flaggability in Automated Moderation: The Case of Reporting Toxic Behavior in an Online Game Community

Online platforms rely upon users or automated tools to flag toxic behaviors, the very first step in online moderation. While much recent research has examined online moderation, the role of flag remains poorly understood. This question becomes even more urgent in automated moderation, where flagging becomes a primary source of human judgment. We conducted a qualitative study of flagging practices in League of Legends (LoL), a popular eSports game. We found stark differences between how flag is designed to identify toxicity, and flaggability, or how players use and appropriate flag. Players distrust flag, but also appropriate flag for instrumental purposes. Thus, flaggability diverges decidedly from the conception of toxicity, and must be understood within the highly competitive gaming context of LoL. These findings help shed light on the situated nature of flaggability, the role of flag in online moderation, as well as implications for designing flag and moderation.