{"slug": "why-averaging-llm-benchmark-scores-is-fundamentally-broken", "title": "Why averaging LLM benchmark scores is fundamentally broken", "summary": "A new study finds that averaging benchmark scores produces misleading rankings when evaluation data is sparse and item difficulty varies widely, with Spearman rank correlation dropping from 1.000 to 0.809 under such conditions. The researchers show that Item Response Theory (IRT) maintains near-perfect accuracy (ρ ≥ 0.996) across all tested scenarios, including in autonomous vehicle safety and cybersecurity domains.", "body_md": "# Computer Science > Machine Learning\n\n[Submitted on 11 May 2026]\n\n# Title:The Scaling Law of Evaluation Failure: Why Simple Averaging Collapses Under Data Sparsity and Item Difficulty Gaps, and How Item Response Theory Recovers Ground Truth Across Domains\n\n[View PDF](/pdf/2605.11205)\n\n[HTML (experimental)](https://arxiv.org/html/2605.11205v1)\n\nAbstract:Benchmark evaluation across AI and safety-critical domains overwhelmingly relies on simple averaging. We demonstrate that this practice produces substantially misleading rankings when two conditions co-occur: (1) the evaluation matrix is sparse and (2) items vary substantially in difficulty. Through controlled simulation experiments across four domains -- NLP (GLUE), clinical drug trials, autonomous vehicle safety, and cybersecurity -- we show that Spearman rank correlation $\\rho$ between simple-average rankings and ground-truth rankings degrades from $\\rho = 1.000$ at 100% coverage to $\\rho = 0.809$ at 67% coverage with high difficulty heterogeneity (mean over 20 seeds). A standard two-parameter logistic (2PL) Item Response Theory (IRT) model maintains $\\rho \\geq 0.996$ across all conditions. A 150-condition grid sweep over sparsity $S \\in [0, 0.70]$ and difficulty gap $D \\in [0.5, 5.0]$ confirms that ranking error forms a failure surface with a strong $S \\times D$ interaction ($\\gamma_3 = +0.20$, $t = 13.05$), while IRT maintains $\\rho \\geq 0.993$ throughout. We discuss implications for Physical AI benchmarking, where evaluation matrices are often incomplete and difficulty gaps are extreme.\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer\n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers\n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps\n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations\n\n*(*[What are Smart Citations?](https://www.scite.ai/))# Code, Data and Media Associated with this Article\n\nalphaXiv\n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers\n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub\n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub\n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face\n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast\n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower\n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender\n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))\nIArxiv Recommender\n\n*(*[What is IArxiv?](https://iarxiv.org/about))# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [ Learn more about arXivLabs](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/why-averaging-llm-benchmark-scores-is-fundamentally-broken", "canonical_source": "https://arxiv.org/abs/2605.11205", "published_at": "2026-07-01 09:40:20+00:00", "updated_at": "2026-07-01 09:51:31.985936+00:00", "lang": "en", "topics": ["machine-learning", "ai-safety", "ai-research"], "entities": ["GLUE", "Item Response Theory", "IRT"], "alternates": {"html": "https://wpnews.pro/news/why-averaging-llm-benchmark-scores-is-fundamentally-broken", "markdown": "https://wpnews.pro/news/why-averaging-llm-benchmark-scores-is-fundamentally-broken.md", "text": "https://wpnews.pro/news/why-averaging-llm-benchmark-scores-is-fundamentally-broken.txt", "jsonld": "https://wpnews.pro/news/why-averaging-llm-benchmark-scores-is-fundamentally-broken.jsonld"}}