OpenReview mOcTXKawFY · local evidence package
Causal Modeling of Selection in Evolution
Exact enumeration supports the five graphical claims. The paper-native finite-sample rerun falsifies the compound empirical-validation claim without weakening those oracle-premise theorems.
| # | Verdict | Decisive result | Page |
|---|---|---|---|
| 1 | Verified | 26,064/26,064 exact constructions pass | Claim 1 |
| 2 | Verified | 0/1,433,520 violations; 17,712 witnesses | Claim 2 |
| 3 | Verified | 1,433,520/1,433,520 d-separation matches | Claim 3 |
| 4 | Verified | All 93,798 exact CPDAG clause checks pass | Claim 4 |
| 5 | Verified | 0 lost; strict gain in 88,176/269,328 | Claim 5 |
| 6 | Falsified literally | PC advantage in 3/15 cells and 0/5 at d=20 | Claim 6 |