Metraipolis › archive › dc_372945f5ee4f
Pair Difficulty Matters: Rethinking Pairwise LLM-as-a-Judge Evaluation and Consistency — arXiv:2609.37577
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.37577
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T16:57:04.153Z
- SHA-256 of the archived text
9bc0718467db4c0e46c98b1978e1a75866fe995596a56daeb50817e8c9f5e558- Stable URL of this page
https://metraipolis.com/documento/dc_372945f5ee4f
How to cite
Pair Difficulty Matters: Rethinking Pairwise LLM-as-a-Judge Evaluation and Consistency — arXiv:2609.37577 — arXiv (en), https://arxiv.org/abs/2609.37577 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_372945f5ee4f (Metraipolis archive dc_372945f5ee4f, sha256 9bc0718467db4c0e46c98b1978e1a75866fe995596a56daeb50817e8c9f5e558)
BibTeX
@misc{dc372945f5ee4f,
title = {Pair Difficulty Matters: Rethinking Pairwise LLM-as-a-Judge Evaluation and Consistency — arXiv:2609.37577},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.37577},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_372945f5ee4f; sha256 9bc0718467db4c0e46c98b1978e1a75866fe995596a56daeb50817e8c9f5e558},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.37577. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large Language Model judges are widely used to rank texts and text-generating systems through pairwise comparison, and their reliability is typically assessed via three proxies: position bias, transitivity, and pairwise agreement (self- or human-labeled). Because these proxies drive judge selection and benchmarking, a substantial literature reporting that judges perform poorly on them risks steering practitioners away from otherwise capable evaluators. We argue this assessment is misleading. Under the Bradley--Terry geometry underlying pairwise aggregation, each proxy is dominated by close-rank-gap pairs, where inconsistency is information-theoretically expected and individual verdicts contribute little to the aggregate ranking; far-gap pairs carry the ranking signal but barely move the proxies. We formalize this argument and validate it in a controlled simulation and on two human-rated corpora: the proxies correlate only weakly with ranking accuracy against gold, and their predictive component concentrates in the far-gap regime. Judges should therefore be assessed on rank-gap-conditional metrics, ideally against human rankings. Code at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.37577 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T16:27:03.318Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).