Python, stdlib-only synthetic companion that generates a discovery rate Rd from a label-blind interface searcher on a Highly Optimized Tolerance (HOT) rule graph. The patch-gap G = Rd/Rp and the stocks Sknown, Sactive are outputs, not fitted legal measurements. v0.2.0 adds Lean 4 proofs of the paper’s Section 4.3 stock algebra, a recomputation of the Section 4.4 regime arithmetic against the paper text, a 162-cell searcher sweep frozen before its first run (failed v1 spec kept), and a figure/prediction ledger. Paper this illustrates (separate Zenodo concept): concept 10.5281/zenodo.21910874; this PDF 10.5281/zenodo.23262565 (v1.2). Reproduce: python verify_patchgap_claims.py; python run_all_gates.py. No paper PDF. No real legal corpus.
AI governance · open texture · reward hacking · regulatory arbitrage · highly optimized tolerance · governance patch-gap · legal exploit discovery · computational claim gate · synthetic rule graph