An automated tool that requires users to provide more information in their ban appeals before they are sent to the moderators.

To decrease the workload on moderators and the amount of toxic appeals that moderators are exposed to when reviewing appeals.

Atreja et al. (2023) conducted a field experiment with a Reddit community that looked at how ban appeals differed when users were asked to provide an additional set of information in their appeals by a bot compared to those who did not have to do so. The bot would only forward completed ban appeals to moderators.

By shifting the effort to the users to provide the reasons and evidence for overturning their ban, the authors observed a 70% decrease in the number of appeals that moderators received which led to a similar 60% drop in the number of appeals they reviewed. (Note: all results we report in this review are statistically significant, unless otherwise stated.) Furthermore, when the bot was in place there was a 90% decrease in the number of toxic appeals that moderators were exposed to.

Mod teams might be concerned that, while AppealMod decreases their workload, it could have a negative impact on users, specifically by decreasing the number of successful appeals or by creating resentment among users who prefer their appeals to go directly to mods. The authors, however, did not find any evidence of those negative effects. The number of successful appeals was not significantly different when AppealMod was used and they similarly did not see any difference in the number of toxic messages sent to mods.

Volunteer moderators offer their time and effort to keep their social media communities thriving. Yet their status also leaves them open to ill-intentioned actions by members of the community. Bans, which have strong consequences for members, may spur greater toxicity and retaliation that manifest in the appeals. By adding friction to the ban appeal process, the tool enhances the moderation experience by alleviating the moderator’s workload and protecting their well-being.

While the tool offers clear benefits to the moderators, in this study the researchers were not able to examine its effect on  all users or the community as a whole.  Further research would be needed to be conducted to see if there is likewise a positive, neutral or negative impact.