The page names NEPA the cleanest lead specimen of over-binding, defines that claim as the leftover after five confounds, then states the leftover is unmeasured. The printed 2.2–2.5 year medians are raw delay. The four adaptations are the signature only at that leftover, while each also appears under capacity starvation. Using those delay numbers and the four co-occurring symptoms as the demonstration treats the hypothesis as the specimen.

The diagnostic is total error-adjusted burden across institution and affected parties. The lead numbers are institutional cycle times. A categorical exclusion that shortens Notice-of-Intent to final EIS and shifts undocumented harm is an improvement on the printed clock and a deterioration on the other ledger. The page names both ledgers and fills one, so the printed medians cannot decide the diagnostic they are asked to illustrate.

The threshold is directional, operationalising any listed handle is a separate research task, and the specification-excess test is "cut this process in half" without naming which object is halved. The environmental-justice hypothetical is labelled uncomputed. Readers who already differ on which projects are high-equity will not be forced to recode, so the threshold cannot gate additive repair on the specimen it uses.

Conditional closure is allowed only where a named corrective owner exists with six slots: resources, independent verification, monitoring trigger, re-entry-and-repair authority, public record, reporting. The Fiscal Responsibility Act installed a sponsor right to seek a court-set schedule when agencies miss deadlines. That actuator moves the agency's clock, not post-decision environmental harm. The preferred repair is not in the lead specimen's action set.

Cumulative burden has no defending constituency and removal is politically attributable. Triage is then recommended at named constraints rather than aggregate ledgers. Naming alternatives analysis does not create standing to close a cut after the next bad event. The 1984 and 1995 economic sunsets are admitted not to generalise to diffuse environmental harm, which is the specimen's harm class. The repair recommended is the one the political-economy paragraph says cannot be staffed on this specimen.

A new constraint passes the self-audit only if it retires, merges, expands capacity, improves triage, or replaces pre-decision burden with corrective ownership. The diagnostic requires independent measurement of five confounds before the leftover may be named, which is itself a documentation duty. The screening rule ("already visible" / "plainly not") is a pre-partition judgment that lets each reader pick the conclusion, so the self-audit applied to the diagnostic adds binding mass without retiring a diagnostic step.