Latent Risk Geometry Classification (LRGC) A Causal Taxonomy of Hidden Failure Geometries in Infrastructure and Complex Systems

Latent Risk Geometry Classification (LRGC) introduces a constrained causal taxonomy for hidden failure behavior in infrastructure and complex systems. Rather than treating resilience loss as a single universal phenomenon, the framework separates latent-risk dynamics into four non-interchangeable geometries: Corridor Fragility, Hidden Scarring, Visible Lock-In, and Symmetry-Protected Binary Failure. These geometries arise from four dominant causal terms: topology (T), delayed degradation memory (M), synchronization (S), and reinforcement/pruning (R).The branch was developed through frozen observables, frozen thresholds, null-controlled mechanism tests, synthetic falsification, and real-world infrastructure interrogation. Key findings include: synchronization actively suppresses heterogeneity formation; bottleneck topology injects inherited fragility at time zero; reinforcement alone cannot generate lock-in without active pruning; and rerouting alone cannot generate Hidden Scarring without delayed degradation memory.Real-world validation includes ERCOT WESTEX transmission corridors, IEEE 118-bus benchmark comparison, EAGLE-I outage data, and Con Edison utility-scale reliability analysis. The work demonstrates that latent failure geometry is scale-dependent and that some mechanisms do not merely generate geometries — they suppress the emergence of others.This paper does not propose a universal resilience law. It presents a falsifiable, mechanism-separated taxonomy for identifying and distinguishing latent-risk geometries in real systems.