# Adversarial review log — cancer-atavism (2026-10-09 revised edition)

- **Authoring model:** Sol 6.1 (HSEF v5.3 pipeline)
- **Adversarial reviewer:** Claude Opus 5.5 (Anthropic)
- **Date:** 2026-10-14 (round 2)
- **Canonical file reviewed:** `consensus_consistency_audit_2026-10-09/revised/cancer-atavism/cancer-atavism.review.md`
- **Level before → after:** Γ (Moderate, epiphenomenon) → Γ (Moderate, epiphenomenon) — unchanged
- **Momentum before → after:** Decisive work underway → **Mechanistic only**
- **Corrections applied:** 5 (0 Critical, 0 Major, 4 Minor, 1 Editorial)
- **PMIDs verified:** 59/59 title + first author + journal + year; 59/59 DOI strings; 0 failures; 0 retractions; 1 Expression of Concern, correctly handled
- **Load-bearing claims spot-checked:** 9 (all supported)
- **Status:** Approved as revised

## Relation to the earlier edition

A full diff against `hsef_v53_popularity_ranks_31_35_2026-10-08/cancer-atavism/cancer-atavism.review.md` shows the 2026-10-09 revision added **exactly one block** — the "Consensus provenance reconciliation" section (18 lines) — and changed nothing else. No letter, confidence, role, subclaim, count, reference or evidence card differs. **No revision-introduced substantive regression.** The one revision-introduced defect is the false approval clause inside the spliced block (finding D2). No earlier adversarial approval of this slug exists anywhere in the deliverables tree, so this is its first adversarial adjudication.

## Findings

| ID | Severity | Location | Original → corrected | Evidence |
|---|---|---|---|---|
| D1 | Minor | AT A GLANCE; M/S/D section; scorecard `extended_scoring_MSD.momentum` | Momentum "Decisive work underway" → "Mechanistic only" | The artifact names 2024 mapping work, 2022 GOT2 context tests and 2026 correction notices — all published, the last being source-integrity notices — and then disclaims any ongoing recruiting trial. The framework reserves "Decisive work underway" for an experiment someone is actually running that could cross the next threshold. Momentum modifies no letter. |
| D2 | Minor | Consensus provenance reconciliation block | "The earlier adversarial approval refers to its original edition" → rescoped to state that this case carried no earlier adversarial approval | False template boilerplate spliced in by the 9 October integration step. Verified absent: no adversarial record for this slug exists in the tree, and the earlier edition has neither banner nor approval section. Rescoped, not deleted. |
| D3 | Minor | REVIEWER NOTES | "Expression of Concern PMID10440377" → "The 1999 report PMID10440377, which carries the 2026 Editorial Expression of Concern PMID41986727" | The notes labelled the parent article's PMID as the notice. PMID 41986727 is the real Editorial Expression of Concern (Nature 652(8112):E18, April 2026); the evidence card already paired them correctly. |
| D4 | Minor | scorecard `final_score.confidence_rationale` | Access depth removed as a stated basis for the band | The framework ties confidence to evidence density and reproducibility, not to the reviewer's access. Re-derived on density and reproducibility alone — 59 reports but the distinctive claim rests on a few analyses sharing TCGA data and gene-age assignments with no independent biological replication, against an openly heterogeneous metabolic literature — which is Moderate. **Band unchanged**; only its basis corrected. Historical `score_before` / `score_after` snapshots inside `consensus_validation` were left as-is. Lead policy of 2026-10-14 (two-step test) applied: access depth struck, band re-derived. Step 2 keeps **Moderate** — the distinctive coherent-program claim rests on a small set of analyses sharing TCGA data and gene-age assignments with no independent biological replication, against an openly heterogeneous metabolic literature — which are legitimate determinants; the High band's "extensive independent replication across labs, models, and assay modalities" is not met, and access depth was not the only thing holding the band down. |
| D5 | Editorial | 17 lines | Lost inter-word spaces restored: "ErratumIn: PMIDnot supplied.", "Fig1b", "Fig4a", "Fig4C" | Known upstream builder defect. Whitespace-only repair verified by whitespace-stripped comparison; p53, Yap1, Hand2, beta1-integrin, INK4A, CDK4, PP2A, hTERT, GOT2, MCT1, SOX11 and SOX4 were protected. |

## Rejected findings (checked, no change)

1. Aggregation — SC3 primacy with a mandatory paired SC2/SC3 minimum, all three alternatives reported (weakest-link Β, primary-subclaim Γ, scope-matched Γ); the Β sensitivity comes from the narrower SC1/SC4 components, reported separately. Deviation from the index wording is disclosed.
2. Β/Γ boundary for the nominal — Γ is right: network and phylostratigraphic support without a selective whole-program disruption-and-rescue experiment.
3. Causal role `epiphenomenon` (Step 2) — correct for the distinctive coherent-program account.
4. Competing models at Ε and Δ above the nominal Γ — permitted; the tournament does not set the level.
5. **Renderer defect (nominal printed as its own "Best-supported rival") — checked, absent.** `models_evaluated[0]` is rank 1 with `is_nominal` false; the nominal ("Coherent ancestral program") is rank 4.
6. **Reviewer-notes/consensus-block contradiction — checked, absent.** The preserved `METHODS AND CONSENSUS CONTRIBUTION` section records signed-in Consensus browser use, so it agrees with the spliced block.
7. Three evidence cards record "ErratumIn: PMID not supplied" — the identifiers are genuinely absent from retrieved metadata; retained as a residual limitation.
8. Print/issue vs e-pub year — all within ±1.
9. Validator warning `v5.claim_form` — the claim states re-expression **plus** the consequence that the program drives tumorigenesis; documented exception.

## Validator

`hsef-hypothesis-review-v5/tools/validate_review.py` on the revised scorecard + prose: **33 pass, 1 warn, 1 fail**. Warn = `v5.claim_form` (rejected by convention); fail = `v5.lay_summary`, marked "mandatory from v5.4" and inapplicable to this v5.3 package. Output in `validator.txt`; pre-review `validation-report.json` preserved in `sol61-original/`.

## Files changed

- `cancer-atavism.review.md` (banner, momentum, rescoped provenance clause, EoC pairing, approval section, whitespace)
- `cancer-atavism.scorecard.json` (`extended_scoring_MSD.momentum.value` / `.artifact` / `.justification`; `final_score.confidence_rationale`; surgical text edits, formatting preserved)

## Propagation

The inventory records no `propagate_to` copies. The divergent copy at `deliverables/hsef_v53_popularity_ranks_31_35_2026-10-08/cancer-atavism/cancer-atavism.review.md` was **not** modified, as instructed — note that it still carries the uncorrected Momentum value and the uncorrected Expression-of-Concern sentence.