# MUDDLE: Measuring Understanding of Documents under Distractor and Length Effects

arXiv: https://arxiv.org/abs/2608.29477 · HTML: https://mlsanity.com/paper/2608.29477/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 20 references that are in the map.
- First posted on arXiv 30 Aug 2026.

## Builds on (most cited first, 17 of 17)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](https://mlsanity.com/paper/1606.05250/) (2016)
- [Dense Passage Retrieval for Open-Domain Question Answering](https://mlsanity.com/paper/2004.04906/) (2020)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding](https://mlsanity.com/paper/2308.14508/) (2023)
- [RULER: What's the Real Context Size of Your Long-Context Language Models?](https://mlsanity.com/paper/2404.06654/) (2024)
- [DocVQA: A Dataset for VQA on Document Images](https://mlsanity.com/paper/2007.00398/) (2020)
- [Approximate Nearest Neighbor Negative Contrastive Learning for Dense Text Retrieval](https://mlsanity.com/paper/2007.00808/) (2020)
- [Large Language Models Can Be Easily Distracted by Irrelevant Context](https://mlsanity.com/paper/2302.00093/) (2023)
- [Making Retrieval-Augmented Language Models Robust to Irrelevant Context](https://mlsanity.com/paper/2310.01558/) (2023)
- [The Power of Noise: Redefining Retrieval for RAG Systems](https://mlsanity.com/paper/2401.14887/) (2024)
- [MMLongBench-Doc: Benchmarking Long-context Document Understanding with Visualizations](https://mlsanity.com/paper/2407.01523/) (2024)
- [Document Understanding Dataset and Evaluation (DUDE)](https://mlsanity.com/paper/2305.08455/) (2023)
- [LongDocURL: a Comprehensive Multimodal Long Document Benchmark Integrating Understanding, Reasoning, and Locating](https://mlsanity.com/paper/2412.18424/) (2024)
- [UNIDOC-BENCH: A Unified Benchmark for Document-Centric Multimodal RAG](https://mlsanity.com/paper/2510.03663/) (2025)
- [IRPAPERS: A Visual Document Benchmark for Scientific Retrieval and Question Answering](https://mlsanity.com/paper/2602.17687/) (2026)

## Built on by (most cited first, 2 of 2)

- [LongHarness Bench: Stress-Testing Language Model Harnesses for Long-Context Reasoning](https://mlsanity.com/paper/2609.38137/) (2026)
- [State Representation Shapes Long-Horizon Agent Learning and Transfer: An Empirical Study](https://mlsanity.com/paper/2610.15407/) (2026)

## Abstract (arXiv)

Document question-answering systems increasingly answer questions over collections of retrieved documents rather than one clean source, so robustness to distracting context matters as much as reading ability. When such systems fail, it is often unclear whether the context was too long or the distractors were too close to the topic, because prior work tends to conflate these two effects. We present MUDDLE, a controlled benchmark that separates them. MUDDLE uses 270 human-annotated questions, each tied to a single source document, and instantiates every question in five conditions: the source alone, the source with two or four topically similar hard negatives, and the source with two or four random distractors. The random distractors are matched to the hard negatives in length and provenance, so an accuracy gap between the two arms reflects topical similarity rather than length. All five conditions are rendered in markdown, page images, and raw PDF, but the distractor sweep reported here is run in markdown, since a source plus its distractors exceeds current image and PDF input limits. We score answers with an LLM judge across three model families. In the complete markdown sweep, hard negatives lower accuracy more than length-matched random documents at both context sizes for gpt-5-mini, while random documents stay near the no-distractor baseline. The effect is small but directionally consistent, and for gpt-5-mini hard negatives significantly underperform length-matched random distractors when pooled across context sizes. We release the data and evaluation code for a reproducible study of context degradation.