All data

Vision-Language Capability Ledger Data Receipts

Public Draft v0.2 REAL DATA

8 parallel Opus research-agent sweeps yielded ~403 raw papers, deduplicated and hand-arbitrated to 397 unique. Bills 4, 7, 10 ★ NO CLEAN TRIGGER YET (0 clean triggers each). Rebuttal density 22.4%.

Receipts

ArtifactLinkPurpose
Bill definitionsbills_draft.md12 bills + 6 meta-costs + 3 escape gates + ★ 4, 7, 10 empty-space verification with real fire counts
Threat modelpurpose.mdThreat model, scope, empty-space hypothesis, cousin-ledger coupling
Corpus union JSON_batch_1_union.json397 unique papers (deduplicated from ~403 raw across 8 sweeps), with full metadata
Classifierbill_classifier.pyRegex rule engine + hand-arbitration. Run with --arbitrate-union
Aggregatoraggregate_batch_1.pyDeduplicates raw sweep JSONs into the corpus union
READMEREADME.mdReproducibility README with run order

Real fire counts

BillCands.CleanRebuttalsGated
1 — Image-search / web-snapshot contamination11920
2 — OCR-extracted-text leakage11542
3 — Vision-tokenizer-format brittleness311669
4 ★ Causally-faithful vision-grounding mechanism3201319
5 — Cross-VLM-architecture portability211542
6 — Tool-augmented-vision decomposition8323
7 ★ Cross-benchmark generalization6033
8 — Multi-image / video / interleaved generalization151320
9 — Vendor-self-eval independence181413
10 ★ Universal vision-task coverage5032
11 — Anti-saturation construction352753
12 — Distilled-cousin / open-weight VLM audit7610

Public draft v0.2 (2026-05-09). Sweep JSONs live in the source repo at ProjectForty2 public evidence bundle: vision_language/deep_loops/. Target v0.3 lock 2026-Q3.