Verify, Then Vote: Overturning Wrong Majorities in Multi-Agent LLMs

A November 2026 paper, "Verify, Then Vote," from the grainulation/papers collection reports that a rule letting checked passage evidence overturn a multi-agent LLM majority vote improves accuracy on the evaluated benchmark, but the original scored rule misses its pre-specified harm criterion. The paper states its formal result identifies an inspectable failure condition rather than guaranteeing a low error rate, and reports the gains, failures, and the tradeoff against a stronger model judge together, with code, data, and proof checks released for reproduction under Apache 2.0 and CC BY 4.0 terms. The collection is being prepared for its first public release, with study freezes recording when analyses were specified.

Papers you can read. Results you can check. Research on evidence and verification in multi-agent systems. Manuscripts, formal proofs, code, and the recorded data behind the results. Paper map https://github.com/grainulation/papers/blob/main/CONTENTS.md · Supplement https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/supplement.pdf · Reproduce the results https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/artifact/README.md · Cite https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/README.md citation · Contribute https://github.com/grainulation/papers/blob/main/CONTRIBUTING.md Overturning Wrong Majorities in Multi-Agent LLMs · November 2026 When agents share a mistake, their majority can be wrong. This paper studies a rule that lets checked passage evidence overturn the vote, while recording which evidence licensed each change. Verification improves accuracy on the evaluated benchmark, but the original scored rule misses its pre-specified harm criterion. The formal result identifies an inspectable failure condition; it does not guarantee a low error rate. The paper reports the gains, failures, and tradeoff against a stronger model judge together. | Start here | What you get | |---|---| | Read the paper https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/paper.pdf | The rule, main results, controls, and limitations. | | Supporting material https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/supplement.pdf | Methods, proofs, pooled-bound construction, and full replay tests. | | Reproduce the results https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/artifact/README.md | Commands and pinned toolchains for the code, analyses, and formal proofs. | | Source and citation https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/README.md | LaTeX build instructions, artifact map, and BibTeX. | Each paper lives alongside the material needed to inspect its claims: - Read the argument in the main paper and the supporting methods in its supplement. - Inspect the study protocols, recorded inputs, checker labels, and reported outputs. - Reproduce the analyses and run the implementation and proof checks using the artifact guide. The paper map https://github.com/grainulation/papers/blob/main/CONTENTS.md collects abstracts and links. catalogue.yaml https://github.com/grainulation/papers/blob/main/catalogue.yaml provides the machine-readable index. With Tectonic https://tectonic-typesetting.github.io/ installed, or latexmk with pdfLaTeX, run from the repository root: make -C papers/verify-then-vote/tex This builds the main paper and supplement. Add the arxiv target to create a source bundle containing both standalone documents. Code, data, and proof checks have separate requirements; follow the artifact guide https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/artifact/README.md . The collection is being prepared for its first public release . Study freezes record when analyses were specified; they are not earlier publications. After publication, tagged releases and the history https://github.com/grainulation/papers/blob/main/history.md will record corrections and revisions. For a correction or reproduction report, open an issue https://github.com/grainulation/papers/issues with the paper, section, claim, and supporting evidence or commands. See CONTRIBUTING.md https://github.com/grainulation/papers/blob/main/CONTRIBUTING.md for the contribution and release process. | Material | Terms | |---|---| | Code | Apache License 2.0 https://github.com/grainulation/papers/blob/main/LICENSE | | Manuscripts and figures | CC BY 4.0 https://github.com/grainulation/papers/blob/main/LICENSE-CC-BY-4.0 | | Data we produced | CC BY 4.0 unless the artifact's DATA LICENSES.md states otherwise | | Third-party data and model outputs | Their own dataset or provider terms; EXAONE outputs are not under CC BY 4.0 | The artifact's data and component licenses https://github.com/grainulation/papers/blob/main/papers/verify-then-vote/artifact/DATA LICENSES.md take precedence over this summary.