Transportable Proofs: A Receipt Protocol for Mathematical Claims That Survive Tool Drift
Mathine: Proof-Transport Receipt Machine
Link: https://doi.org/10.5281/zenodo.18397556
Modern mathematical claims increasingly depend on evolving toolchains: SAT/SMT solvers, proof assistants, large libraries, numerical pipelines, and brittle environment assumptions. In that setting, “accepted today” can fail tomorrow—not because the claim is false, but because the verification path no longer travels cleanly across machines, versions, or institutions.
This paper proposes a receipt-first protocol for transportable proofs: instead of treating a proof as a static artifact, treat it as a governed sequence of admissible moves, each accompanied by a minimal checkable receipt. The receipts are designed so verification is cheaper than reproduction, and so the decisive steps remain independently checkable under bounded resources.
A central theme is tool drift resistance. The protocol formalizes identity, canonicalization, and replay constraints so that changes in compilers, libraries, solvers, or hardware don’t silently change what was “proved.” Where drift cannot be controlled, the framework requires explicit HOLD states that localize which obligation stopped being checkable, rather than letting closure degrade into narrative trust.
The practical payoff is portability. A claim can be transported across teams and time without importing someone else’s entire environment or relying on informal “it worked for us.” If mathematical trust is going to scale under AI-driven claim volume, it needs this kind of zero-trust, receipt-based closure—where truth is something you can re-check, not just inherit.
Mathine — Evidence In, Trust Out 