Evidence item
Cross-judge evaluation found meaningfully distinct decision processes and pronounced heterogeneity across judge pairs. DIRECT_TEXT
Source: Do Judges Behave Like Algorithms? · locator {} · excerpt hash sha256:d8685da3609e445659fb5c6ac89d812d0fc803cdc0b528968d144d0ed4e499e9
Exact excerpt
We find that it cannot: judges exhibit meaningfully distinct decision-making processes. Specifically, when we evaluated each judge-specific model on cases decided by other judges, performance deteriorated substantially, revealing pronounced heterogeneity across judge pairs.
Linked claims
| Claim | Direction | Relevance | Steps | Counted |
|---|---|---|---|---|
| e1122017 | SUPPORT | DIRECT | 0 | yes |
Audits
None.