When pull turns to shove: A continuous-time model for opinion dynamics

Accurate modeling of opinion dynamics has the potential to help us understand polarization and what makes effective political discourse possible or impossible. Here, we use physics-based methods to model the evolution of political opinions within a continuously distributed population. We utilize a network-free system of determining political influence and a local-attraction, distal-repulsion dynamic for reaction to perceived content. Our approach allows for the incorporation of intergroup bias such that messages from trusted in-group sources enjoy greater leeway than out-group ones. We are able to extrapolate these nonlinear microscopic dynamics to macroscopic population distributions by using probabilistic functions representing biased environments. The framework we put forward can reproduce real-world political distributions and experimentally observed dynamics and is amenable to further refinement as more data becomes available.