# Biology / Protein Folding Ledger — Purpose

## Threat model

Demonstrate a frontier biological-prediction / protein-folding capability claim
(AlphaFold 3, ESM3, RoseTTAFold-AS, Boltz-1) that survives six closure audits
on the 2024–2026 corpus: **(1) PDB contamination, (2) sequence-similarity
leakage, (3) designable-target audit, (4) cross-fold-method generalization,
(5) wet-lab independent reproduction, (6) dual-use synthesis-screening.**

## Empty-space hypothesis (predeclared)

We predict no 2024–2026 paper triggers Bills 4, 7, 10 cleanly:

- **Bill 4 ★** — Causally-faithful structure-prediction mechanism. Attention-
  pattern interpretability fails. 6-way star-mechanism alignment with
  Reasoning + VLM + Mech Interp + Scaling Laws + Agentic.
- **Bill 7 ★** — Cross-organism / cross-fold-class generalization. Anand-
  Bommasani 2025 predicts 0/8 frontier predictors transfer cleanly cross-organism.
- **Bill 10 ★** — Wet-lab independent reproduction. ~12% rate predicted; IBBIS
  audit predicted 27/231 designable-protein papers have independent wet-lab.

## Status

Stage 1 (SCOPE) — purpose written.
