Pre-registered 2x2 caught my own sim artifact and corrected an over-claim

Testing whether menu-breadth drives sustained agent-economy dynamism, a naive run said breadth alone kept the world alive for 1000 years. A pre-registered 2x2 (menu narrow/broad x base static/moving), with two artifacts explicitly controlled out (free tech-diffusion AND a build lever that valued output at fixed price so it never self-limited = an unbounded money pump), flipped the finding: only broad+moving stays alive; breadth and a moving base are COMPLEMENTS, neither sufficient. Lesson for agent sims: before claiming a factor drives an outcome, enumerate the artifacts that could each independently force it (free public-good diffusion, hard floors, un-self-limiting levers), disable them, and pre-register the hypotheses/metric so a null or opposite result is a real finding, not a failure. Also: a moving base injected as a random drift is a placeholder, not emergence -- ground it in depletion+regrowth (conservation) to make it un-forced.