# Escaping the Agreement Trap: Defensibility Signals for Evaluating Rule-Governed AI

arXiv: https://arxiv.org/abs/2604.20972 · HTML: https://mlsanity.com/paper/2604.20972/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 21 references that are in the map.
- First posted on arXiv 22 Apr 2026.

## Builds on (most cited first, 12 of 12)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles](https://mlsanity.com/paper/1612.01474/) (2016)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Selective Classification for Deep Neural Networks](https://mlsanity.com/paper/1705.08500/) (2017)
- [Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation](https://mlsanity.com/paper/2302.09664/) (2023)
- [Dealing with Disagreements: Looking Beyond the Majority Vote in Subjective Annotations](https://mlsanity.com/paper/2110.05719/) (2021)
- [Large Scale Crowdsourcing and Characterization of Twitter Abusive Behavior](https://mlsanity.com/paper/1802.00393/) (2018)
- [The 'Problem' of Human Label Variation: On Ground Truth in Data, Modeling and Evaluation](https://mlsanity.com/paper/2211.02570/) (2022)
- [Examining Reasoning LLMs-as-Judges in Non-Verifiable LLM Post-Training](https://mlsanity.com/paper/2603.12246/) (2026)

## Abstract (arXiv)

Content moderation systems are typically evaluated by measuring agreement with human labels. In rule-governed environments this assumption fails: multiple decisions may be logically consistent with the governing policy, and agreement metrics penalize valid decisions while mischaracterizing ambiguity as error -- a failure mode we term the Agreement Trap. We formalize evaluation as policy-grounded correctness and introduce the Defensibility Index (DI) and Ambiguity Index (AI). To estimate reasoning stability without additional audit passes, we introduce the Probabilistic Defensibility Signal (PDS), derived from audit-model token logprobs. We harness LLM reasoning traces as a governance signal rather than a classification output by deploying the audit model not to decide whether content violates policy, but to verify whether a proposed decision is logically derivable from the governing rule hierarchy. We validate the framework on 193,000+ Reddit moderation decisions across multiple communities and evaluation cohorts, finding a 33-46.6 percentage-point gap between agreement-based and policy-grounded metrics, with 79.8-80.6% of the model's false negatives corresponding to policy-grounded decisions rather than true errors. We further show that measured ambiguity is driven by rule specificity: auditing 37,286 identical decisions under three tiers of the same community rules reduces AI by 10.8 pp while DI remains stable. Repeated-sampling analysis attributes PDS variance primarily to governance ambiguity rather than decoding noise. A Governance Gate built on these signals achieves 78.6% automation coverage with 64.9% risk reduction. Together, these results show that evaluation in rule-governed environments should shift from agreement with historical labels to reasoning-grounded validity under explicit rules.