Protection and Usable Correction: A conditional UCM comparison

Can a restriction preserve an allowed state while preventing a correction that would preserve it too? This fourth Ev5 working paper examines that question through a conditional comparison informed by the Unified Coherence Model (UCM). Three decision policies share an observation rule, corrective actions and a clearance screen. Under the stated protection assumptions, a warning-bound hold remains protected but stalled. A release policy permits progress, while a simpler controller reproduces its complete trajectory. Their equivalence identifies an unnecessary restriction without demonstrating the sacrifice of useful protection. Proofs and eight completed model scenarios separate screening, corrective action, opportunity timing and release. Delayed or blocked access, inaccurate clearance and persistent forcing expose different limits. Later target attainment can coexist with an earlier protection breach, while a bounded response can fail to complete the task. A separate synthetic execution of recovered software demonstrates output suppression with continued internal updating and automatic resumption. It establishes neither repaired readiness nor a performance benefit. The contribution is an inspectable account of how protection and usable correction can work together. UCM supplies the organising question; ordinary control explains the model mechanism. Conditional correction and implemented resumption are demonstrated separately. Maintenance of the conditions enabling later correction remains unestablished. The findings do not constitute physical validation or validation of the complete UCM.