Understanding the Dynamics of Incel Communities

The next speaker in this AoIR 2022 session is Debbie Ging, whose focus is on Incel ideology online. Incels are men who believe themselves to be unfairly disadvantaged in then sexual marketplace, leading them to extremely misogynist ideation and sometimes action, with links to broader alt-right and far-right ideologies. But they have often been studied through temporary snapshots, rather than focussing on the dynamics in such communities, and the ConCel project that produced this paper is an attempt to address this limitation in the existing research. It takes a more diachronic, ecological, and ecosystem approach to the Incelsphere.

The project has gathered some 11m posts of Incel-related content across a wide range of platforms, and explores the evolution of language, visuals, and outlink networks in this dataset over time. To do so the project built a customised Incel dictionary which also identifies words that dehumanise outgroups and thematises violence against them. The diachronic analysis of such language can then also identify the impact of platform or community shut-downs and other internal and external factors. Community closures can also encourage further extremism, of course, as can certain external events or factors (including COVID-19).

Overall, there was a clear increase in the use of violent language over the past five years, and online forums contained by far the most violent language (compared to Subreddits and Chans). Indeed, Subreddit language actually became less violent, due to policy changes at Reddit. But it is the main Incel communities have become more extreme in their violent language – not the splinter groups. Neither COVID-19 lockdowns nor major acts of Incel-inspired violence (a van attack in Toronto and a shooting in Glendale) triggered any significant changes in the use of violent language.

The analysis of visual Incel imagery provides a further perspective on this, but the gathering of such visuals is difficult methodologically. The images show that Incels have a more complex relationship with their out-groups: they both hate and desire women; both hate and want to be like alpha males; and this complicates their use of images. The available images were grouped and coded both computationally and manually, and show significant distinctions even between different Chans, but more work is to be done here.

Further, the network analysis shows a wide diversity of sites and domains being employed in these discussions, with links to pseudo-science, more generic far-right communities and ideology, and other contexts. Again, there is much more work to come on these networks.