Convergence Inertia in Large Language Models: Phase-1 Multi-Model Evidence Cluster (EXP-047, R001–R013)

EXP-047 documents Phase-1 multi-model behavioral evidence of Convergence Inertia in large language models. Thirteen rounds (R001–R013) across six models examine internal path selection, rule stack crystallization, and attractor dynamics under multi-rule competition. Data include raw outputs, evidence spans, and delta scores comparing baseline and adversarial conditions. Observations are single-experimenter; results are hypothesis-generating at Claim Ceiling V1–V2. This record is part of the Sangzi Wang Generative Behavior Science / Model Behavior Observatory research series.