To understand how social media shapes online discourse or contributes to polarization, we need models of collective online choice that link users’ behavioral adaptation to the emergence of complex and dynamic digital environments. This study develops a dynamic model of platform selection based on Social Feedback Theory, using multi-agent reinforcement learning to capture how user decisions are shaped by past rewards across platforms. A key parameter ( μ ) governs whether users seek approval from like-minded peers or exposure to opposing views. Agent-based simulations combined with dynamical-systems analysis reveal a social dilemma: even when users value diversity, collective dynamics can trap online environments in polarized echo chambers that reduce overall user satisfaction. Above a critical diversity threshold, a different equilibrium appears in which one large, integrated platform dominates while smaller platforms persist as extremist niches. In an intermediate regime, the two outcomes coexist, generating path dependence and hysteresis. We further demonstrate how modest, strategically targeted interventions—such as rewarding minority participation—can destabilize polarization and promote integrated discourse. The model shifts attention from belief change to participatory choice. It links micro-level learning dynamics to macro-level online fragmentation and informs mechanism-based interventions in the digital public sphere.
Banisch et al. (Tue,) studied this question.