Warrant by Structure

Software systems produce effects under mixtures of declared capability, ambient privilege, runtime convention, external dependency, human discretion, and post hoc audit. Warrant by structure asks whether an effect was produced through an explicit, authority-bound, rule-evaluated, receipt-producing path whose mediation boundary can itself be tested. The source manuscript reports a 30-system study and contrasts it with a proposed Veritas Task Engine (VTE). The supplied corpus does not contain the system list, coding guide, raw annotations, denominators, executable VTE artifact, evaluation output, or replication package needed to verify those results. One comparison cell is internally inconsistent: a reduction of 55.2 percentage points from 71.9% implies 16.7%, not approximately 0%. The empirical numbers are therefore retained here as source-reported claims requiring reconstruction, not as established findings. The durable conceptual contribution is a testable distinction between post hoc description and structurally mediated execution. That distinction requires tighter boundaries than the source supplies. A receipt can bind what a system claims occurred; it does not by itself prove legitimate authority, correct rule evaluation, physical effect, absence of bypass, or complete governance closure. Structural warrant is strongest when authorization, mediation, effect observation, custody, and independent verification are separately represented and jointly bounded. This revision formalizes those obligations, repairs the arithmetic presentation, defines the proposed metrics, integrates adjacent corpus machinery, supplies a warrant-strength ladder and adversarial tests, and locates the work in the universal GSM and realization funnel. No external research or implementation occurred.