Experimental Data: First Principles Learning for Sociotechnical World Models (72-cycle LLM Agent Pod)

Complete experimental record from a 72-cycle longitudinal experiment in which an LLM agent (Claude, Anthropic) constructed a causal world model of AI public trust dynamics using the First Principles Learning protocol. Contains 546 events with per-node certainty scores, 34 causal edges, 71 raw observations, and 8 agent decision logs. The agent discovered three structural errors in its initial world model through prediction-error learning and exhibited a systematic behavioral-introspective gap over 50+ cycles. See the accompanying README for schema documentation.