Two copies of the same model co-fail on 90% of missions.

In a preregistered 18,000-mission evaluation scored by deterministic code with no LLM judge, two instances of one model in a two-agent handoff co-failed on 90.0% of missions where either failed (log OR 6.66, phi 0.916). Swapping in a different model reduced the association in six of six contrasts. Swapping vendor while already using a different model did not, a registered null. arXiv 2608.12895 Multiplying component reliabilities over-credits redundancy exactly when your agents share a base model. Your verifier-checks-generator pattern is not two independent samples.