Harmful online information can lead to many forms of online anti-X hate, where X can be races, ethnicity, or even vaccines. Such anti-X hate can drive certain groups to extremism and induce dangerous misbehavior, such as the incident of the US capital riot in January 2021. Having computational models that can rigorously describe the online information dynamics and predict the formation process of anti-X communities could potentially help policy makers and online moderators to develop better strategies to prevent damages, but the inherent complexities in information dynamics — such as the rapid fusion–fission dynamics, the increasing number of Internet users, and the cross-platform information sharing — pose substantial challenges to develop a rigorous model for accurate prediction and proper mitigation. In a recent work, Pedro D. Manrique and colleagues generalized a nonlinear fluid dynamics model for describing online information dynamics and further demonstrated that their model can explain the rapid growth of online hate and propose prevention strategies.
The proposed model considers two important processes for the formation of online anti-X communities: fusion and fission. While fusion corresponds to heterogeneous individuals aggregating based on character similarity, fission corresponds to the fragmentation process based on dissimilarity. These processes resemble a phenomenon in liquid where bubbles can aggregate, grow, and pop off the liquid surface, and the authors make use of this analogy to develop a generalized nonlinear fluid equation that can account for both fusion and fission. More importantly, the model can describe higher-scale aggregation that forms clusters of communities, corresponding to larger communities across different social-media platforms. Notably, the rapid growth of anti-X groups can be explained by a shockwave solution, which corresponds to a strong and fast propagating nonlinear wave. The authors further demonstrated that the formation of anti-X groups can be intervened or even prevented by adjusting the flux of new users and the fusion probability that describes the diversity of the online population. For instance, it was shown that the onset of an anti-X shockwave can be delayed by decreasing the fusion probability. When applied to real cases — such as foreign anti-US communities on VKontakte and anti-US government communities on Facebook — the model predicted community growth curves that agree well with empirical data.
This is a preview of subscription content, access via your institution