This work develops a verification-theoretic framework for quantifying audit and replay costs in black-box and device-independent settings. Rather than proposing new physical models or experimental protocols, the paper formalizes verification resistance as a structural property arising from asymmetry between a system’s generative process and an external verifier’s evidentiary access. Using the CHSH inequality as a worked example, the analysis derives an explicit replay cost κ associated with reproducing verification transcripts under constrained access. This cost is shown to scale independently of system correctness, revealing a divergence between logical verifiability and procedural auditability. The framework generalizes beyond quantum contexts, applying equally to cryptographic, computational, and institutional verification regimes. The contribution is methodological: it provides a precise language and cost model for discussing verification effort, audit friction, and evidence replay, without relying on hidden assumptions about trust, authority, or transparency. A reference implementation and analytic derivations are included to support independent inspection and extension. This work is intended as a theoretical and audit-oriented contribution, not a finalized applied system or experimental proposal.
Sanjit Mehat (Wed,) studied this question.