@inproceedings{bhandarkar-nanesha-2026-vouch,
title = "Vouch: Counterfactual Contracts for Multi-Agent Coordination",
author = "Bhandarkar, Abhay and
Nanesha, Vedanth",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {IJCNLP}-{AACL} 2026",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-aacl.222/",
pages = "4345--4368",
ISBN = "979-8-89176-444-6",
abstract = "Multi-agent LLM coordination typically reduces to voting or arbitration over self-reported labels and reasoning that is uncoupled from agent behavior and fails predictably under structured bias. A shortcut-reliant agent is confident about its shortcut, and its predictions survive any rule that consults only self-reports. We propose VOUCH, a coordination protocol in which every agent claim is coupled to a deterministic test the protocol runs itself. An agent commits to a counterfactual prediction about an intervention{'}s outcome; the protocol applies the intervention and verifies the contract only if the actual outcome matches. The natural single-side instantiation is unsound as a self-coherent agent satisfies its own commitment by construction. We resolve this with a dual-side construction where the proposer additionally commits to a second agent{'}s post-intervention prediction, a cross-side claim that self-coherence cannot fulfill. Across multimodal classification (Hateful Memes, CrisisMMD, MM-IMDb), code (MBPP, HumanEval), and math (GSM8K, MATH, SVAMP), the dual-side fix reduces wrong-verified events by 46{\%}{--}85{\%} per domain while never falling more than 1 point below the Ensemble baseline on accuracy. VOUCH is therefore designed not to outperform ensembling, but to preserve task accuracy while making coordination decisions testable and auditable."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="bhandarkar-nanesha-2026-vouch">
<titleInfo>
<title>Vouch: Counterfactual Contracts for Multi-Agent Coordination</title>
</titleInfo>
<name type="personal">
<namePart type="given">Abhay</namePart>
<namePart type="family">Bhandarkar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vedanth</namePart>
<namePart type="family">Nanesha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-444-6</identifier>
</relatedItem>
<abstract>Multi-agent LLM coordination typically reduces to voting or arbitration over self-reported labels and reasoning that is uncoupled from agent behavior and fails predictably under structured bias. A shortcut-reliant agent is confident about its shortcut, and its predictions survive any rule that consults only self-reports. We propose VOUCH, a coordination protocol in which every agent claim is coupled to a deterministic test the protocol runs itself. An agent commits to a counterfactual prediction about an intervention’s outcome; the protocol applies the intervention and verifies the contract only if the actual outcome matches. The natural single-side instantiation is unsound as a self-coherent agent satisfies its own commitment by construction. We resolve this with a dual-side construction where the proposer additionally commits to a second agent’s post-intervention prediction, a cross-side claim that self-coherence cannot fulfill. Across multimodal classification (Hateful Memes, CrisisMMD, MM-IMDb), code (MBPP, HumanEval), and math (GSM8K, MATH, SVAMP), the dual-side fix reduces wrong-verified events by 46%–85% per domain while never falling more than 1 point below the Ensemble baseline on accuracy. VOUCH is therefore designed not to outperform ensembling, but to preserve task accuracy while making coordination decisions testable and auditable.</abstract>
<identifier type="citekey">bhandarkar-nanesha-2026-vouch</identifier>
<location>
<url>https://aclanthology.org/2026.findings-aacl.222/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>4345</start>
<end>4368</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Vouch: Counterfactual Contracts for Multi-Agent Coordination
%A Bhandarkar, Abhay
%A Nanesha, Vedanth
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-444-6
%F bhandarkar-nanesha-2026-vouch
%X Multi-agent LLM coordination typically reduces to voting or arbitration over self-reported labels and reasoning that is uncoupled from agent behavior and fails predictably under structured bias. A shortcut-reliant agent is confident about its shortcut, and its predictions survive any rule that consults only self-reports. We propose VOUCH, a coordination protocol in which every agent claim is coupled to a deterministic test the protocol runs itself. An agent commits to a counterfactual prediction about an intervention’s outcome; the protocol applies the intervention and verifies the contract only if the actual outcome matches. The natural single-side instantiation is unsound as a self-coherent agent satisfies its own commitment by construction. We resolve this with a dual-side construction where the proposer additionally commits to a second agent’s post-intervention prediction, a cross-side claim that self-coherence cannot fulfill. Across multimodal classification (Hateful Memes, CrisisMMD, MM-IMDb), code (MBPP, HumanEval), and math (GSM8K, MATH, SVAMP), the dual-side fix reduces wrong-verified events by 46%–85% per domain while never falling more than 1 point below the Ensemble baseline on accuracy. VOUCH is therefore designed not to outperform ensembling, but to preserve task accuracy while making coordination decisions testable and auditable.
%U https://aclanthology.org/2026.findings-aacl.222/
%P 4345-4368
Markdown (Informal)
[Vouch: Counterfactual Contracts for Multi-Agent Coordination](https://aclanthology.org/2026.findings-aacl.222/) (Bhandarkar & Nanesha, Findings 2026)
ACL