@inproceedings{chaudhury-etal-2026-structural,
title = "Structural Uncertainty from Self-Preference in Black-Box Language Models",
author = "Chaudhury, Baishali and
Wang, Mengdie Flora and
Park, Hyunji Hayley and
Ghosh, Rahul and
Hong, Sungmin and
Woo, Jae Oh",
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.699/",
pages = "16119--16147",
ISBN = "979-8-89176-459-0",
abstract = "Large language models can be confidently wrong, making uncertainty estimation especially important in black-box settings where token probabilities are unavailable. Existing post-hoc methods estimate uncertainty from output dispersion, treating sampled responses largely as exchangeable. We propose structural uncertainty, an observable signal derived from self-preference-induced ranking stability: given multiple sampled candidates, we elicit sparse pairwise self-preferences, aggregate them into ranking distributions via Bradley{--}Terry modeling with PageRank, and decompose uncertainty into two complementary entropy-based components{---}across-trial ranking instability and within-trial candidate ambiguity. Across five LLMs and eight benchmarks, self-preference-derived structural signals can provide information complementary to output dispersion, and their combination improves selective prediction on several reasoning and knowledge tasks. The two components relate differently to accuracy: on math reasoning, within-trial ambiguity can be positively correlated with correctness{---}consistent with settings where multiple plausible candidates remain competitive under self-preference ranking{---}whereas on factual retrieval the structural signal collapses toward uniformity, diagnosing a regime where preference-based uncertainty is uninformative. Structural uncertainty is not a universal replacement for dispersion but a regime-sensitive complementary signal."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chaudhury-etal-2026-structural">
<titleInfo>
<title>Structural Uncertainty from Self-Preference in Black-Box Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Baishali</namePart>
<namePart type="family">Chaudhury</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mengdie</namePart>
<namePart type="given">Flora</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyunji</namePart>
<namePart type="given">Hayley</namePart>
<namePart type="family">Park</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rahul</namePart>
<namePart type="family">Ghosh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sungmin</namePart>
<namePart type="family">Hong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jae</namePart>
<namePart type="given">Oh</namePart>
<namePart type="family">Woo</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>Large language models can be confidently wrong, making uncertainty estimation especially important in black-box settings where token probabilities are unavailable. Existing post-hoc methods estimate uncertainty from output dispersion, treating sampled responses largely as exchangeable. We propose structural uncertainty, an observable signal derived from self-preference-induced ranking stability: given multiple sampled candidates, we elicit sparse pairwise self-preferences, aggregate them into ranking distributions via Bradley–Terry modeling with PageRank, and decompose uncertainty into two complementary entropy-based components—across-trial ranking instability and within-trial candidate ambiguity. Across five LLMs and eight benchmarks, self-preference-derived structural signals can provide information complementary to output dispersion, and their combination improves selective prediction on several reasoning and knowledge tasks. The two components relate differently to accuracy: on math reasoning, within-trial ambiguity can be positively correlated with correctness—consistent with settings where multiple plausible candidates remain competitive under self-preference ranking—whereas on factual retrieval the structural signal collapses toward uniformity, diagnosing a regime where preference-based uncertainty is uninformative. Structural uncertainty is not a universal replacement for dispersion but a regime-sensitive complementary signal.</abstract>
<identifier type="citekey">chaudhury-etal-2026-structural</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.699/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>16119</start>
<end>16147</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Structural Uncertainty from Self-Preference in Black-Box Language Models
%A Chaudhury, Baishali
%A Wang, Mengdie Flora
%A Park, Hyunji Hayley
%A Ghosh, Rahul
%A Hong, Sungmin
%A Woo, Jae Oh
%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 chaudhury-etal-2026-structural
%X Large language models can be confidently wrong, making uncertainty estimation especially important in black-box settings where token probabilities are unavailable. Existing post-hoc methods estimate uncertainty from output dispersion, treating sampled responses largely as exchangeable. We propose structural uncertainty, an observable signal derived from self-preference-induced ranking stability: given multiple sampled candidates, we elicit sparse pairwise self-preferences, aggregate them into ranking distributions via Bradley–Terry modeling with PageRank, and decompose uncertainty into two complementary entropy-based components—across-trial ranking instability and within-trial candidate ambiguity. Across five LLMs and eight benchmarks, self-preference-derived structural signals can provide information complementary to output dispersion, and their combination improves selective prediction on several reasoning and knowledge tasks. The two components relate differently to accuracy: on math reasoning, within-trial ambiguity can be positively correlated with correctness—consistent with settings where multiple plausible candidates remain competitive under self-preference ranking—whereas on factual retrieval the structural signal collapses toward uniformity, diagnosing a regime where preference-based uncertainty is uninformative. Structural uncertainty is not a universal replacement for dispersion but a regime-sensitive complementary signal.
%U https://aclanthology.org/2026.emnlp-main.699/
%P 16119-16147
Markdown (Informal)
[Structural Uncertainty from Self-Preference in Black-Box Language Models](https://aclanthology.org/2026.emnlp-main.699/) (Chaudhury et al., EMNLP 2026)
ACL