@inproceedings{yang-etal-2026-beyond,
title = "Beyond Leaderboards: A Frozen-Rubric Contract for Position-Invariant Winning-Argument Detection",
author = "Yang, Zida and
Zeng, Weiqing and
You, Guquan and
Zheng, Jiexin",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1270/",
pages = "28937--28958",
ISBN = "979-8-89176-459-0",
abstract = "When large language models serve as evaluation infrastructure, progress cannot be reduced to a single accuracy number. We study winning-argument (WA) detection as a stress test: can one system jointly improve accuracy, enforce speaker-order invariance, emit rubric-conditioned diagnostics, improve a proper scoring rule, and cut inference cost? TAPIR freezes a survey-grounded rubric into a dual-artifact protocol, distils privileged signals by mode-split self-distillation, and predicts through an anti-symmetric scalar head paired with a structured ledger. An algebraic swap identity (Winning-Position Invariance, WPI=1) holds on every test pair, and a tied-order wrapper preserves it on the deployed router. On the standard 807-pair WA split, TAPIR beats our same-backbone Qwen3-14B re-implementation of Labruna et al.{'}s MS-PS-MLP, 65.80{\%} vs. 58.12{\%} (paired McNemar p=3.41{\texttimes}10{\ensuremath{-}}4), at a lower Brier score and 39.41{\texttimes} faster inference per pair. Reusing the frozen protocol without retuning carries the mechanism to a second task and a second backbone, both at WPI=1. Under domain shift it only matches length-only controls, so we report contract preservation rather than superiority. The contribution is the joint contract under a paired protocol, with an explicit amortisation boundary, not a leaderboard delta."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yang-etal-2026-beyond">
<titleInfo>
<title>Beyond Leaderboards: A Frozen-Rubric Contract for Position-Invariant Winning-Argument Detection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zida</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weiqing</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guquan</namePart>
<namePart type="family">You</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiexin</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>When large language models serve as evaluation infrastructure, progress cannot be reduced to a single accuracy number. We study winning-argument (WA) detection as a stress test: can one system jointly improve accuracy, enforce speaker-order invariance, emit rubric-conditioned diagnostics, improve a proper scoring rule, and cut inference cost? TAPIR freezes a survey-grounded rubric into a dual-artifact protocol, distils privileged signals by mode-split self-distillation, and predicts through an anti-symmetric scalar head paired with a structured ledger. An algebraic swap identity (Winning-Position Invariance, WPI=1) holds on every test pair, and a tied-order wrapper preserves it on the deployed router. On the standard 807-pair WA split, TAPIR beats our same-backbone Qwen3-14B re-implementation of Labruna et al.’s MS-PS-MLP, 65.80% vs. 58.12% (paired McNemar p=3.41×10\ensuremath-4), at a lower Brier score and 39.41× faster inference per pair. Reusing the frozen protocol without retuning carries the mechanism to a second task and a second backbone, both at WPI=1. Under domain shift it only matches length-only controls, so we report contract preservation rather than superiority. The contribution is the joint contract under a paired protocol, with an explicit amortisation boundary, not a leaderboard delta.</abstract>
<identifier type="citekey">yang-etal-2026-beyond</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1270/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>28937</start>
<end>28958</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Leaderboards: A Frozen-Rubric Contract for Position-Invariant Winning-Argument Detection
%A Yang, Zida
%A Zeng, Weiqing
%A You, Guquan
%A Zheng, Jiexin
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F yang-etal-2026-beyond
%X When large language models serve as evaluation infrastructure, progress cannot be reduced to a single accuracy number. We study winning-argument (WA) detection as a stress test: can one system jointly improve accuracy, enforce speaker-order invariance, emit rubric-conditioned diagnostics, improve a proper scoring rule, and cut inference cost? TAPIR freezes a survey-grounded rubric into a dual-artifact protocol, distils privileged signals by mode-split self-distillation, and predicts through an anti-symmetric scalar head paired with a structured ledger. An algebraic swap identity (Winning-Position Invariance, WPI=1) holds on every test pair, and a tied-order wrapper preserves it on the deployed router. On the standard 807-pair WA split, TAPIR beats our same-backbone Qwen3-14B re-implementation of Labruna et al.’s MS-PS-MLP, 65.80% vs. 58.12% (paired McNemar p=3.41×10\ensuremath-4), at a lower Brier score and 39.41× faster inference per pair. Reusing the frozen protocol without retuning carries the mechanism to a second task and a second backbone, both at WPI=1. Under domain shift it only matches length-only controls, so we report contract preservation rather than superiority. The contribution is the joint contract under a paired protocol, with an explicit amortisation boundary, not a leaderboard delta.
%U https://aclanthology.org/2026.emnlp-main.1270/
%P 28937-28958
Markdown (Informal)
[Beyond Leaderboards: A Frozen-Rubric Contract for Position-Invariant Winning-Argument Detection](https://aclanthology.org/2026.emnlp-main.1270/) (Yang et al., EMNLP 2026)
ACL