Interface-Dependent Closure Stress Tests in Contemporary Cosmology

We present a quantitative, diagnostic analysis of closure properties in contemporary cosmological inference through a unified stress-test protocol applied across multiple observational interfaces. Using posterior distributions from Planck PR3, BAO, Pantheon, calibration parameters, and the independent ACT DR6 CMB dataset, we evaluate effective compatibility costs via importance sampling and Effective Sample Size (ESS).The results reveal a clear hierarchy of interface behavior: the H₀ interface associated with Planck high-ℓ data is structurally rigid, while S₈ is consistently absorbable; BAO acts as a partial bridge, Pantheon as a selective constraint, calibration parameters amplify rigidity, and ACT exhibits high absorbability under identical stress.These findings demonstrate that global closure failure in cosmology is interface-dependent rather than universal, and that persistent tensions arise from regime-specific access constraints rather than from a single underlying inconsistency. The analysis is intentionally diagnostic and non-ontological, providing a reproducible cartography of closure behavior without invoking new physics.