The seven-test naming rule grades an artifact without a preauthorized return path as a tracker, while the specimen list files UK audit follow-up as soft re-entry and the NAO tracker as a partial ledger. A PAC recall the committee may decline to use is an oversight habit, not a path named when the item entered. The four-grade list does work the seven tests do not produce.

The four grades measure how much new political will a failed check still needs, not a rival discriminator. PAC recall power is already in the architecture. NAO is partial because evidence stays inside the office and confirmation is only partly external, which is a missing-field fact, not a will-budget fact.

The return-path test is being used for two jobs: whether a return path is named, and how much will using it takes. PAC recall is named only as a committee that can always call someone back, not as a hearing a failed NAO confirmation obliges. If the will-budget is the real claim, the seven tests overclaim; if the seven tests are the real claim, soft is tracker.

On a recommendation the NAO did not confirm as implemented, the opening specimen is a tracker unless PAC or another named forum had to take the row up because confirmation failed. "Partial ledger" is otherwise a courtesy caption, not a grade the tests produce.

Grenfell Phase 2 has item-level trace, a ministerial board, and parliamentary updates. Leaving binding-on-failed-verification as an open question is the return-path test refusing to fire. If a red verification can be reported indefinitely without changing budget, owner, or assignment, Grenfell is a tracker with a good display.

The discriminator claims an auditor can predict, from structure, where execution will disappear without formal contradiction. Resource fiction is a named path with no money: the named-path test passes and execution still vanishes. A seven-pass grade certifies the record that will be cited as proof of delivery; it does not find the usual vanishing.

A structural test cannot see spend before anyone spends. Named is not spent. The seven tests still catch missing owners, in- chain verifiers, and unnamed return paths before any outcome. Drop the prediction sentence; keep the classifier.

A pass is then evidence the record is complete enough to point at, not that execution will happen. Cheap closure and funded closure are different diagnostics: a fully filled, unfunded item cannot be cheaply closed on paper and is free in the world. Until the page splits those, a pass is the hide.

A ledger that blocks cheap closure also executes a determinate bad yes. Verification against entry-time criteria will pass if the wrong instrument was implemented. Re-entry fires on failed verification, not on "this should not have been accepted." Mandate gap is the uncomputed decision; this is the computed mistake.

Re-entry can change the decision, budget, or assignment, not only nag the same owner. Superseded is already a permitted state. The design is that the delivery owner does not close alone; a later forum can still recode or kill the object.

Superseded is a legitimate execution state and a burial pathology in the same essay. Killing a bad yes is a miss of the original acceptance. If the exception register treats a kill as clean supersession, abandon is burial with a better name; if it treats every kill as a stain, forums will not recode bad yeses. The missing rule is an abandonment test that is neither owner reclassification nor permanent red.