Moderation Challenges in Digital Gaming Spaces: Prevalence of Offensive Behaviors in Voice Chat

Take This partnered with Modulate, a voice moderation company, to produce what is described as the first known baseline on offensive behaviors in gaming voice chat. Data was collected via Modulate's ToxMod tool from three indie studios that each have a few hundred thousand active monthly users and consented to anonymized data use. The analysis covers 11,462 players over a preset 30-day period, split evenly between perceived adult and perceived underage players. ToxMod scores each flagged clip from 0 to 20, with scores above 10 treated as offensive and 13 or higher as severe.

In total, 25,291 offenses were flagged, and 26.43% of players had at least one offense of offensive language. Of all players, 21.39% had only non-severe offenses and 5.03% had at least one severe offense. Racial/cultural hate speech accounted for 53.35% of all offenses, followed by sexual vulgarity at 33.21% and gendered sexual hate speech at 12.42%.

Perceived adult players were more likely than perceived underage players to have at least one offense (36.28% versus 16.58%). Sexual vulgarity was the only category rated more often as severe than non-severe. The authors caution that ToxMod's design may partly shape these findings.