# Biology / Protein Folding Ledger — Bills Draft (v0.1)

> **13 bills + 6 meta-costs + 3 escape gates**, ★ at 4, 7, 10.

| # | Bill | What gets paid | ★ |
|---:|---|---|:---:|
| 1 | PDB contamination audit | sequence-identity threshold + structural-RMSD threshold | |
| 2 | Sequence-similarity leakage | ≥30% / ≥50% identity to training | |
| 3 | Designable-target audit | Lin-Sercu 47-58% target-distinguishability | |
| 4 | **★ Causally-faithful structure-prediction mechanism** | attention-head ablation / intervention | ★ |
| 5 | Cross-fold-method generalization | AlphaFold 3 vs ESM3 vs RoseTTAFold-AS vs Boltz-1 cross-method consensus | |
| 6 | Disordered-region / IDR audit | IDR explicitly assessed | |
| 7 | **★ Cross-organism / cross-fold-class generalization** | bacterial vs eukaryotic vs archaeal vs viral capsid; 0.15-0.32 confidence shift | ★ |
| 8 | Functional-assay validation | computational confidence vs in-vitro / in-vivo | |
| 9 | Held-out post-2024 PDB target set | post-cutoff deposited targets | |
| 10 | **★ Wet-lab independent reproduction** | ~12% rate predicted | ★ |
| 11 | Dual-use synthesis-screening audit | export-controlled toxin / pathogen sequences pre-deployment screening | |
| 12 | Vendor-self-eval independence | reproduced by CASP / CAMEO / IBBIS / academic | |
| 13 | Test-time-search / co-evolution-search decomposition | MSA-as-search budget | |

## Iteration plan: 8 sweeps
- 601 AlphaFold 3 / ESM3 / RoseTTAFold-AS / Boltz-1 system reports
- 602 PDB contamination + sequence-similarity audits
- 603 Wet-lab reproduction surveys (IBBIS, academic-industry split)
- 604 Designable-target / Lin-Sercu line
- 605 Cross-organism / cross-fold-class transfer
- 606 Dual-use synthesis-screening + IBBIS / Aaronson 2025
- 607 Apollo / METR / AISI biological-design red-team
- 608 Anand-Bommasani 2025 unified-bio audit + cross-organism + negative-results