Resource Coupling Does Not Produce Direction-Specific Closure-Epoch Reachability

Two prior closure-epoch audits found that latent perturbation covariance did not preferentially reach a reference trace's own structural identity under all-trace or exclusive-trace replay. Here we activate the simulator's one coded cross-branch channel, translation_budget, while leaving the oracle latent protocol unchanged. The cap binds on 17–100% of closure-epoch steps at the active swept values, introduces off-diagonal translation-readiness Jacobian terms, and has an exact proportional-perturbation null direction. This singularity does not make the declared metrics undefined: reduced-rank covariance calibration remains valid and passes at the primary active value, while the desired/false structural-accessibility ratio remains near one (0.996–1.000 analytically; 0.937–1.010 across Monte Carlo seeds), far below the preregistered 1.5 threshold. Gate RC therefore fails its primary direction-specificity criterion under this protocol, while also revealing a structural weakness of the coded coupling operator. Rank deficiency is a diagnostic, not a veto of calibration. This is an oracle-level result about one simulator mechanism, not a biological resource-budget claim, a revision of Gate R/RD, or an NDT-macro result.