MRU v19 Route-History Toy Branch: Previous-Bin Persistence Descriptor Effect Under No-Smuggling Guardrails

This archive contains the frozen MRU v19 route-history toy bookkeeping branch. The branch tests whether immediate previous-bin persistence is sufficient for a robust descriptor-level route-history effect in a toy harness under strict no-smuggling guardrails. The archive includes the polished report, source scripts, output archives, closure registers, manifests, and SHA256 receipts. The staged workflow includes output-only baseline testing, temporal-specific descriptors, classifier repair, dose-response robustness, ablation collapse, independent replication, seed-block replication, label permutation tests, within-run segment stability, decoy specificity repair, minimal component isolation, persistence-only robustness, surrogate challenge/repair, size/horizon robustness, dose-level signature stability, blind mode-label recovery, and final closure audit. The strongest allowed claim is narrow: in this toy bookkeeping harness, previous-bin persistence is sufficient for a robust, reproducible descriptor-level route-history effect under the tested controls, challenges, repairs, and blind recovery. Previous-bin persistence-only descriptor effect under strict no-smuggling guardrails. This archive does not claim physical inertia, force, field behavior, gravity, physical motion, orbit, cosmology, inverse-square derivation, Landauer foundation, thermodynamic foundation, or real physical causality. DISCLAIMER AI Assistance Disclosure: This work was developed, designed and ochestrated by Malin Hess with assistance from GPT-5.5 Thinking for code generation, drafting, workflow organization, formatting, and claim-boundary auditing. Final interpretation, archive decisions, and publication responsibility remain with the human author. AI assistance is not independent validation or peer review. The archive should be read only within its stated toy bookkeeping scope, with no physical mechanism claim.