# Verified retrospective ablation

Source revision `a173e8ab187a11ea58b4e67f2c7293450b71cd52`. Sixteen fixed methods; no candidate ranking promoted, no new observations or physical truth. Numerical and allocation failures are preserved and disclosed. [Full interpretation](results.md), [exact table](comparison.csv), [all metrics, label curves and controls](summary.json).

- [results.md](results.md) — `docs/ABLATION_RESULTS_20261014.md`
- [ablation-01-ABLATION_RESEARCH_GOAL_20261014.md](ablation-01-ABLATION_RESEARCH_GOAL_20261014.md) — `docs/ABLATION_RESEARCH_GOAL_20261014.md`
- [ablation-02-ABLATION_PROTOCOL_20261014.md](ablation-02-ABLATION_PROTOCOL_20261014.md) — `docs/ABLATION_PROTOCOL_20261014.md`
- [ablation-03-ABLATION_EXECUTION_AMENDMENT_20261014.md](ablation-03-ABLATION_EXECUTION_AMENDMENT_20261014.md) — `docs/ABLATION_EXECUTION_AMENDMENT_20261014.md`
- [ablation-04-ABLATION_VERIFICATION_AMENDMENT_20261014.md](ablation-04-ABLATION_VERIFICATION_AMENDMENT_20261014.md) — `docs/ABLATION_VERIFICATION_AMENDMENT_20261014.md`
- [ablation-05-ABLATION_NUMERICAL_TIE_AMENDMENT_20261014.md](ablation-05-ABLATION_NUMERICAL_TIE_AMENDMENT_20261014.md) — `docs/ABLATION_NUMERICAL_TIE_AMENDMENT_20261014.md`
- [ablation-06-run_ablation_diagnostic.py](ablation-06-run_ablation_diagnostic.py) — `research/run_ablation_diagnostic.py`
- [ablation-07-verify_ablation_diagnostic.py](ablation-07-verify_ablation_diagnostic.py) — `research/verify_ablation_diagnostic.py`
- [ablation-08-build_ablation_report_20261014.py](ablation-08-build_ablation_report_20261014.py) — `research/build_ablation_report_20261014.py`
- [ablation-09-ablation_registration_20261014.json](ablation-09-ablation_registration_20261014.json) — `artifacts/ablation_registration_20261014.json`
- [ablation-10-ablation_execution3_registration_20261014.json](ablation-10-ablation_execution3_registration_20261014.json) — `artifacts/ablation_execution3_registration_20261014.json`
- [ablation-11-ablation_verification_execution6_registration_20261014.json](ablation-11-ablation_verification_execution6_registration_20261014.json) — `artifacts/ablation_verification_execution6_registration_20261014.json`
- [summary.json](summary.json) — `artifacts/ablation_diagnostic_execution3_20261014/summary.json`
- [comparison.csv](comparison.csv) — `artifacts/ablation_diagnostic_execution3_20261014/comparison.csv`
- [ablation-14-checkpoint.json](ablation-14-checkpoint.json) — `artifacts/ablation_diagnostic_execution3_20261014/checkpoint.json`
- [ablation-15-verification.json](ablation-15-verification.json) — `artifacts/ablation_verification_execution6_20261014/verification.json`
- [checkpoint.json](checkpoint.json) — `artifacts/ablation_checkpoint_20261014.json`
- [ablation-17-ablation_reviewed_supplement_20261014.json](ablation-17-ablation_reviewed_supplement_20261014.json) — `artifacts/graph/ablation_reviewed_supplement_20261014.json`

[Exact source hashes](source-inventory.csv). The checkpoint retains hash inventories for the full local 19,328-row ledger and numerical artifacts. Compact exports are evidence for a diagnostic, not a public physical classification catalogue.