Description
Social media platforms provide large-scale examples of interacting many-body systems, in which collective behaviour emerges from local interactions between heterogeneous agents. In this work, we study ideological asymmetry on Twitter/X using tools motivated by statistical physics and condensed matter theory. Users are assigned positions on a left-right ideological axis from their retweet activity, allowing the population to be coarse-grained into ideological classes analogous to components or phases in an interacting system. A particular challenge in the ‘social particle’ case is that the broader interacting populations are largely invisible and transient. We examine reply interactions between the classes to infer the effective population structure that gives rise to the observed interaction rates. We develop a develop a simple interaction model that provides a statistical-mechanical description of macroscopic ideological activity (‘temperature’) in terms of support and threat activity (‘pressure’) from ideological groups. We find that the structure of activity appears consistent across contexts, but that the specific behaviour depends on the environment. With this work we connect empirical studies of online behaviour with opinion dynamics, non-equilibrium statistical physics, and models of interacting social particles.
| I am the presenting author | Yes |
|---|