Bounded Propagation Without Ignition: Structural Limits of Local Coupling in Governance Network Simulations

We investigate disturbance propagation in locally coupled governance network simulations across four topologies: chain, star, random-sparse, and fully connected. The full simulator evolves trust, disturbance, stability, and cognitive distortion with local coupling, interventions, and stochastic updates; the main analysis uses a reduced disturbance-centered interpretation for clarity. Propagation is classified with a strict conjunction rule: affected share ≥ 20% of non-seed nodes, cascade depth ≥ 5, and sustained failure ≥ 3 consecutive steps. Baseline and weak-coupling families often show high propagation, while targeted damping, gating, nonlinear propagation, feedback, and hysteresis produce suppression regimes. Fine-grained scans over intrinsic decay reveal topology-dependent threshold windows rather than a single universal flip point. Within the tested model family and parameter ranges, we do not observe a robust sustained ignition phase as a stable attractor.