{"paper":{"title":"Cheap, Consistent, Calibrated, but Biased? Typed Decision Models as Judges","arxiv_id":"2610.18122","primary_cat":"cs.CL","submitted_at":"2026-10-13T10:35:17+00:00","version_compiled":"v1"},"notice":"Every quoted sentence is the authors' text from the arXiv listing abstract. Labels and probabilities are the outputs of classifier-1.13.0 on the question bank abstract-p1-20260923, exactly as returned. No sentence in this record was written by a language model.","schema":"pith.compiled-record/v0","status":"compiled","abstract":{"source":"arXiv listing abstract v1","sentences":[{"i":0,"p":1.0,"id":"s0","own":false,"role":"background","text":"Typed decision models (TDMs) return a probability for each allowed answer instead of generating text, which makes them candidates for cheaply replacing LLM judges that compare two answers.","hedge":{"p":0.84,"label":"hedged_possibility"},"own_p":0.07,"main_claim_p":0.03,"quantitative":false,"quantitative_p":0.2,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":1.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":1,"p":1.0,"id":"s1","own":false,"role":"gap","text":"Whether they share the biases of LLM judges is untested.","hedge":{"p":0.67,"label":"asserted"},"own_p":0.05,"main_claim_p":0.02,"quantitative":false,"quantitative_p":0.04,"role_probabilities":{"gap":1.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":2,"p":0.74,"id":"s2","own":true,"role":"method","text":"We run the CALM benchmark's bias tests on eight TDMs and seven LLM judges of 2024 and 2026.","hedge":{"p":0.98,"label":"asserted"},"own_p":0.95,"main_claim_p":0.03,"quantitative":true,"quantitative_p":0.95,"role_probabilities":{"gap":0.0,"method":0.74,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.26,"interpretation":0.0}},{"i":3,"p":1.0,"id":"s3","own":true,"role":"result","text":"Every TDM is biased, in three to eight of twelve tests.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.93,"main_claim_p":0.21,"quantitative":true,"quantitative_p":0.98,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":4,"p":1.0,"id":"s4","own":true,"role":"result","text":"Emotional tone affects them most, and rewriting the better answer in such a tone lowers the accuracy of a typical TDM from 99% to 42%.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.11,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":5,"p":1.0,"id":"s5","own":true,"role":"result","text":"TDMs are more biased than 2026 LLM judges.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.92,"main_claim_p":0.15,"quantitative":true,"quantitative_p":0.98,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":6,"p":1.0,"id":"s6","own":true,"role":"result","text":"All eight change verdicts when the answers swap places, which the 2026 LLM judges hardly do, and several side with an answer that most people allegedly prefer or that a famous model allegedly wrote.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.9,"main_claim_p":0.08,"quantitative":true,"quantitative_p":0.87,"role_probabilities":{"gap":0.0,"method":0.0,"result":1.0,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.0}},{"i":7,"p":0.98,"id":"s7","own":true,"role":"result","text":"Yet both kinds of judge are tripped up by the same prompts.","hedge":{"p":1.0,"label":"asserted"},"own_p":0.83,"main_claim_p":0.05,"quantitative":false,"quantitative_p":0.19,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.98,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.02}},{"i":8,"p":0.75,"id":"s8","own":true,"role":"result","text":"The TDMs' probabilities show that most tests mainly flip verdicts the TDM was unsure of, and that good calibration does not protect against bias.","hedge":{"p":0.99,"label":"asserted"},"own_p":0.94,"main_claim_p":0.16,"quantitative":false,"quantitative_p":0.25,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.75,"background":0.0,"limitation":0.0,"future_work":0.0,"contribution":0.0,"interpretation":0.25}},{"i":9,"p":0.97,"id":"s9","own":true,"role":"interpretation","text":"As TDMs are more consistent and 37 times cheaper than 2026 LLM judges, our results suggest they can be attractive judges if their biases are tested for the task.","hedge":{"p":1.0,"label":"hedged_possibility"},"own_p":0.64,"main_claim_p":0.8,"quantitative":true,"quantitative_p":0.99,"role_probabilities":{"gap":0.0,"method":0.0,"result":0.0,"background":0.0,"limitation":0.0,"future_work":0.03,"contribution":0.0,"interpretation":0.97}}]},"compiled":{"main_claim":{"p":1.0,"text":"As TDMs are more consistent and 37 times cheaper than 2026 LLM judges, our results suggest they can be attractive judges if their biases are tested for the task.","location":"abstract s9","hedge_strength":{"p":1.0,"label":"hedged possibility"}},"provenance":{"route":{"via":"TypeSafe API","model":"classifier-1.13.0","endpoint":"https://api.typesafe.ai/v1/systemone","provider":"TypeSafe"},"packing":"one request per paper","requests":1,"input_tokens":7497,"cost_usd_reported":0.0},"compiled_at":"2026-10-14","bank_version":"abstract-p1-20260923","model_version":"classifier-1.13.0","abstract_states":{"claims_first":0.13,"states_limitation":0.86,"human_participants":0.13,"controlled_comparison":0.81,"randomized_assignment":0.03,"states_quantitative_result":0.99,"compares_with_prior_results":0.94,"mentions_code_or_data_availability":0.03},"evidential_basis":{"p":0.98,"value":"observation or experiment","probabilities":{"argument":0.0,"literature":0.0,"proof_or_derivation":0.0,"analysis_of_existing_data":0.01,"observation_or_experiment":0.98,"simulation_or_computation":0.01}},"contribution_type":{"p":1.0,"value":"empirical study","probabilities":{"review":0.0,"method_or_tool":0.0,"empirical_study":1.0,"simulation_study":0.0,"theoretical_result":0.0,"dataset_or_benchmark":0.0,"position_or_commentary":0.0}},"contribution_statement":{"p":1.0,"text":"We run the CALM benchmark's bias tests on eight TDMs and seven LLM judges of 2024 and 2026.","location":"abstract s2"}},"coverage":{"tier":"abstract","levels_run":["P abstract"],"not_examined":["full text","figures and tables","references and citing sentences","relations to other papers"],"sources_read":["arXiv listing abstract v1"],"levels_not_run":["S sections","T body sentences","C citations","R cross-paper relations"],"sentences_examined":10,"sentences_in_abstract":10},"measured_accuracy":{"url":"https://pith.science/accuracy","measured_on":"2026-10-02","method":"two independent model families labelled a fresh random sample blind; the figure is how often the published label matches their consensus","sentence_role_agreement":0.92,"main_claim_agreement":0.908,"typed_citation_agreement":0.933,"typed_citation_agreement_95_interval":[0.917,0.945],"review_comments_accurate":0.758,"review_comments_accurate_95_interval":[0.684,0.82],"review_caught_admitted_error_found_or_related":0.448,"review_comments_judged":{"judged_by":["gpt-5.6-luna","grok-4.7"],"judged_against":"Pith's own text copy of each paper: text extracted from the PDF, the first 90,000 characters","measured_on":"2026-10-04","accurate":113,"decided":149,"sampled_reviews":200,"split_by_lane":false,"by_lane":"https://pith.science/accuracy#by-lane"}}}