OpenReview LhGxRnGmGJ
| Claim | Verdict | Headline |
|---|---|---|
| 1 bias decays exponentially in N | falsified | power law, slope −0.92 (R²0.90) vs exponential R²0.51 |
| 2 KM-ADD bias bound | reproduced | 0.031% relative bias at N=6,400 |
| 3 truncation bias < conventional | reproduced | ~100× smaller on 3 detection laws |
| 4 asymptotically unbiased if support ≤ boundary | reproduced | −0.003 within vs −22.4 beyond, at any N |
| 5 reduced variance vs LB | falsified | KM variance is 2.4× higher; RMSE 46× better |
| 6 accurate from 10% to 90% censoring | falsified | 0.85% at 48% censoring, 24% at 87% |