# Blocking Automation Needs A Lifecycle Not A Launch Date ## Claim Automated enforcement that can block someone's work needs a graduated lifecycle — advisory first, blocking only after a defined promotion — or it reads to the people subject to it as arbitrary and sudden. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - Holds for any automated gate with real consequences for the person being checked (merge blocks, publish blocks, access denials) — not for purely informational tooling with no blocking power. - The lifecycle doesn't need to be exactly draft/approved/enforced/deprecated, but it needs at minimum an advisory stage people can observe and react to before a rule gains blocking power. - Doesn't claim gradual rollout eliminates all resistance — it claims skipping the advisory stage specifically produces the "surprise bureaucracy" failure mode. ## Evidence Cloudflare's standards move through lifecycle states: approved standards produce non-blocking findings; only enforced standards with a MUST-level violation can block a merge. This graduated model is presented as the mechanism that lets Cloudflare introduce nearly 250,000 flagged deviations and roughly 16,000 merge blocks without the enforcement reading as sudden — engineers see a rule as advisory before it can ever stop their work. ## Links - [[3. Resources/Permanent Notes/Automated Review Is Only As Accountable As The Standards Behind It]] — because extends: that note is about what the agent checks against; this note is about how that checking is introduced without breaking trust. - [[3. Resources/MOCs/AI Agent Safety MOC]] — because applies-in: any agent given blocking authority over human work needs this kind of staged rollout to remain legitimate to the people it affects. ## Tests / falsifiers - Compare engineer trust/appeal-rate metrics for rules introduced with an advisory stage versus rules introduced directly as blocking, holding rule content constant. - If engineers accept directly-blocking rules at the same trust rate as gradually-introduced ones, the lifecycle mechanism specifically wouldn't be doing the legitimacy work this claim attributes to it. ## Maintenance - Last tended: 2026-08-16 - Next review: 2026-11-16 ## Source - [[3. Resources/Readwise/Articles/How Cloudflare enforces engineering standards using AI]] — the approved/enforced lifecycle distinction and reported enforcement volume.