Countering Propaganda, Disrupting Radicalization, Reducing Politico-Ideological Violence: Theorizing a Mechanistic Model, Using Prisma Scoping and Systematic Reviews of Evidence on Algorithmic Counter-Propaganda Tools

Scholars often focus on extremism as belief systems that excuse, justify, encourage, or engender the use of violence as a means to effect social and political change (i.e., politico-ideological violence (PIV), or “terrorism”). Whatever the “extremist beliefs” at issue, scholars agree that radicalization is a process, defined by an interlocked series of phenomena leading to a foreseeable result – precisely the kind of phenomenon amenable to mechanism-based explanation. We focus attention on propaganda, a “cog” cited across causal theories of radicalization. Researchers have developed a number of tools to detect and disrupt propaganda, thereby hindering radicalization and reducing PIV. Systematic reviews shed some light on the mechanisms underlying these tools: Celliers and Hattingh (2020) suggest what motivates propagandists in the first place (i.e., the initial cause), while Jahnke et al. (2022) suggest what might motivate otherwise “normal” individuals – as found by McGilloway et al. (2015) and Gill et al. (2021) – to engage in PIV (i.e., the outcome). Hassan et al. (2018) and Wolfowicz et al. (2022) show online connection as an effective means by which propagandists reach propagandees, and Williams et al. (2022) examine the content they use to persuade. In this study, we review the “most relevant” published studies over the last ten years (2013-2023) examining content detection and content moderation algorithms for their effectiveness against politico-ideological violence. We use findings from these studies to develop a mechanism scheme outlining what evidence suggests about how – and whether – counter-propaganda tools should work to reduce engagement in PIV. We discuss limitations (e.g., lack of data and algorithmic bias) in the scope and conclusiveness of the current science and directions for future research to address gaps and test our theoretical model.