{"id":"W4415315888","doi":"10.48550/arxiv.2510.06557","title":"The Markovian Thinker: Architecture-Agnostic Linear Scaling of Reasoning","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Samsung; Alliance de recherche numérique du Canada; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research; Microsoft Research","keywords":"Initialization; Reinforcement learning; Context (archaeology); Decoupling (probability); Matching (statistics); Continuation; Constant (computer programming); Markov process","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008943237,0.0003158557,0.0004006488,0.000140153,0.0002925216,0.0001295376,0.002946404,0.0002520249,0.000004973342],"category_scores_gemma":[0.0009355952,0.0002426883,0.0002326958,0.0002660772,0.00009737186,0.00007712224,0.003529314,0.001189686,0.00001568887],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006626256,"about_ca_system_score_gemma":0.0003666205,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002570495,"about_ca_topic_score_gemma":0.00004834115,"domain_scores_codex":[0.9975258,0.0002045434,0.0006087171,0.0008076377,0.0004127616,0.0004405942],"domain_scores_gemma":[0.9963031,0.0009092905,0.000349108,0.002190917,0.0001562632,0.00009135412],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00006162752,0.0001701851,0.2608843,0.001372719,0.0006288597,0.0001172831,0.01020933,0.3957655,0.0005851878,0.04829252,0.0006119477,0.2813005],"study_design_scores_gemma":[0.0005186489,0.00005794643,0.04578247,0.003443002,0.000128344,0.00002445279,0.00009667408,0.913384,0.003501079,0.0259685,0.00607097,0.001023943],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2940662,0.001423776,0.6969308,0.002290627,0.001919771,0.0003699991,0.000007613812,0.0002471386,0.002744036],"genre_scores_gemma":[0.8971021,0.0001790977,0.1010595,0.0003709343,0.0003523609,0.00003438942,0.000006281593,0.00001965489,0.0008757069],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6030358,"threshold_uncertainty_score":0.9896544,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02605805089703114,"score_gpt":0.2681770662058424,"score_spread":0.2421190153088113,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}