# Debate Helps Weak Judges Reward Stronger Models

arXiv: https://arxiv.org/abs/2605.27483 · HTML: https://mlsanity.com/paper/2605.27483/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 33 of its 45 references that are in the map.
- First posted on arXiv 26 May 2026.

## Builds on (most cited first, 30 of 33)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://mlsanity.com/paper/2311.12022/) (2023)
- [MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark](https://mlsanity.com/paper/2406.01574/) (2024)
- [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023)
- [Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate](https://mlsanity.com/paper/2305.19118/) (2023)
- [Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2307.15217/) (2023)
- [BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions](https://mlsanity.com/paper/2406.15877/) (2024)
- [Humanity's Last Exam](https://mlsanity.com/paper/2501.14249/) (2025)
- [AI safety via debate](https://mlsanity.com/paper/1805.00899/) (2018)
- [Self-critiquing models for assisting human evaluators](https://mlsanity.com/paper/2206.05802/) (2022)
- [Debating with More Persuasive LLMs Leads to More Truthful Answers](https://mlsanity.com/paper/2402.06782/) (2024)
- [FrontierMath: A Benchmark for Evaluating Advanced Mathematical Reasoning in AI](https://mlsanity.com/paper/2411.04872/) (2024)
- [Measuring Progress on Scalable Oversight for Large Language Models](https://mlsanity.com/paper/2211.03540/) (2022)
- [QuALITY: Question Answering with Long Input Texts, Yes!](https://mlsanity.com/paper/2112.08608/) (2021)
- [ARC-AGI-2: A New Challenge for Frontier AI Reasoning Systems](https://mlsanity.com/paper/2505.11831/) (2025)
- [LLM Critics Help Catch LLM Bugs](https://mlsanity.com/paper/2407.00215/) (2024)
- [Language Models Learn to Mislead Humans via RLHF](https://mlsanity.com/paper/2409.12822/) (2024)
- [On scalable oversight with weak LLMs judging strong LLMs](https://mlsanity.com/paper/2407.04622/) (2024)
- [Prover-Verifier Games improve legibility of LLM outputs](https://mlsanity.com/paper/2407.13692/) (2024)
- [The Ideation-Execution Gap: Execution Outcomes of LLM-Generated versus Human Research Ideas](https://mlsanity.com/paper/2506.20803/) (2025)
- [Scalable AI Safety via Doubly-Efficient Debate](https://mlsanity.com/paper/2311.14125/) (2023)
- [Echoes in AI: Quantifying lack of plot diversity in LLM outputs](https://mlsanity.com/paper/2501.00273/) (2024)
- [CodeContests+: High-Quality Test Case Generation for Competitive Programming](https://mlsanity.com/paper/2506.05817/) (2025)
- [Debate Helps Supervise Unreliable Experts](https://mlsanity.com/paper/2311.08702/) (2023)
- [Single-Turn Debate Does Not Help Humans Answer Hard Reading-Comprehension Questions](https://mlsanity.com/paper/2204.05212/) (2022)
- [Scaling Laws For Scalable Oversight](https://mlsanity.com/paper/2504.18530/) (2025)
- [Two-Turn Debate Doesn't Help Humans Answer Hard Reading Comprehension Questions](https://mlsanity.com/paper/2210.10860/) (2022)

## Built on by (most cited first, 1 of 1)

- [Debate Training Reduces Reward Hacking in RLAIF](https://mlsanity.com/paper/2608.17776/) (2026)

## Abstract (arXiv)

Despite theoretical promise, debate as a scalable oversight protocol has produced mixed empirical results: gains in some settings, and null effects in others, especially when the judge does not have information hidden from it. We study proposer-critic debate in a stronger-debater/weaker-judge setting on programmatically verifiable code and logic tasks. Debate helps the judge over a consultancy baseline when the critic provides a usable advantage: the critic's classification ability must exceed the judge's, and the judge must treat critic speeches as claims to verify rather than testimony to summarize. On the three of five pairings where the condition holds, proposer-critic debate's gains are statistically significant over consultancy, and these pairings are the most capable model pairings. On the two non-responder pairings in our set, debate produces null effects, and judge verification rates drop by tens of percentage points once a critic enters the transcript. In these cases the critic's binary-classification ability and the judge's are within noise of each other, and the critic's disagreement is parsed as testimony rather than a claim to check. Ablating rebuttal rounds from debate produces no measurable change in judge performance: a single independent critique recovers the bulk of debate's benefit at lower inference cost. These findings suggest a cheaper primitive for training-free scalable oversight in verifiable domains (answer, critique, judge) and a pre-deployment audit (does the critic beat the judge, and will the judge verify it?) that predicts when debate will help.