{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:N7ZGSYXBVROCMHSLONW7LMC5SH","short_pith_number":"pith:N7ZGSYXB","schema_version":"1.0","canonical_sha256":"6ff26962e1ac5c261e4b736df5b05d91f3c5018b9a436ae67d31b385570c0745","source":{"kind":"arxiv","id":"2605.06840","version":5},"attestation_state":"computed","paper":{"title":"Extracting Search Trees from LLM Reasoning Traces Reveals Myopic Planning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"LLMs expand deep nodes in game reasoning traces but choose moves using only shallow lookahead.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Ji-An Li, Kayla Lee, Marcelo G. Mattar, Saner Cakir, Sinan Akcali, Sixing Chen","submitted_at":"2026-05-07T18:45:46Z","abstract_excerpt":"Large language models (LLMs), especially reasoning models, generate extended chain-of-thought (CoT) reasoning that often contains explicit deliberation over future outcomes. Yet whether this deliberation constitutes genuine planning, how it is structured, and what aspects of it drive performance remain poorly understood. In this work, we introduce a new method to characterize LLM planning by extracting and quantifying search trees from reasoning traces in the four-in-a-row board game. By fitting computational models on the extracted search trees, we characterize how plans are structured and ho"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.06840","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-07T18:45:46Z","cross_cats_sorted":[],"title_canon_sha256":"60850fa24959f61381d5a871661d9f7da8643698369c4898ea4d588309c94124","abstract_canon_sha256":"69450002addcf1c692679267e9fc35077c4296348cbcd105343d7524328ba109"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:22.331427Z","signature_b64":"vyj2/W5ocwgSxMRLfXt0tsa201j004WzIqCucHV9oFQFM2kCIue7kGLH3SZPO06WO1DLAjv+B1ZB8gW7Z4gFAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6ff26962e1ac5c261e4b736df5b05d91f3c5018b9a436ae67d31b385570c0745","last_reissued_at":"2026-05-25T02:01:22.330700Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:22.330700Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Extracting Search Trees from LLM Reasoning Traces Reveals Myopic Planning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"LLMs expand deep nodes in game reasoning traces but choose moves using only shallow lookahead.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Ji-An Li, Kayla Lee, Marcelo G. Mattar, Saner Cakir, Sinan Akcali, Sixing Chen","submitted_at":"2026-05-07T18:45:46Z","abstract_excerpt":"Large language models (LLMs), especially reasoning models, generate extended chain-of-thought (CoT) reasoning that often contains explicit deliberation over future outcomes. Yet whether this deliberation constitutes genuine planning, how it is structured, and what aspects of it drive performance remain poorly understood. In this work, we introduce a new method to characterize LLM planning by extracting and quantifying search trees from reasoning traces in the four-in-a-row board game. By fitting computational models on the extracted search trees, we characterize how plans are structured and ho"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"although LLMs expand deep nodes in their traces, their move choices are best explained by a myopic model that ignores those nodes entirely","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the extracted search trees faithfully represent the LLM's internal planning process and that the fitted computational models correctly identify the causal drivers of move selection from those trees.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LLMs exhibit myopic planning in games, with move choices driven by shallow nodes despite deep reasoning traces, in contrast to human deep-search reliance.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"LLMs expand deep nodes in game reasoning traces but choose moves using only shallow lookahead.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ea519b3eac84138229e9f211cf5a75339b3b61b5c082b13decfe7716f0e9fe52"},"source":{"id":"2605.06840","kind":"arxiv","version":5},"verdict":{"id":"c4df59b9-e655-4117-9a10-f2c946beb069","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:52:19.675931Z","strongest_claim":"although LLMs expand deep nodes in their traces, their move choices are best explained by a myopic model that ignores those nodes entirely","one_line_summary":"LLMs exhibit myopic planning in games, with move choices driven by shallow nodes despite deep reasoning traces, in contrast to human deep-search reliance.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the extracted search trees faithfully represent the LLM's internal planning process and that the fitted computational models correctly identify the causal drivers of move selection from those trees.","pith_extraction_headline":"LLMs expand deep nodes in game reasoning traces but choose moves using only shallow lookahead."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.06840/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-20T12:02:03.749707Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-20T07:34:03.322141Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T18:01:18.681517Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T12:22:29.919344Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"170126c51586d8d30960af7f0b2cb74c663a9218653d37082ef92e0b2ac592bf"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.06840","created_at":"2026-05-25T02:01:22.330804+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.06840v5","created_at":"2026-05-25T02:01:22.330804+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.06840","created_at":"2026-05-25T02:01:22.330804+00:00"},{"alias_kind":"pith_short_12","alias_value":"N7ZGSYXBVROC","created_at":"2026-05-25T02:01:22.330804+00:00"},{"alias_kind":"pith_short_16","alias_value":"N7ZGSYXBVROCMHSL","created_at":"2026-05-25T02:01:22.330804+00:00"},{"alias_kind":"pith_short_8","alias_value":"N7ZGSYXB","created_at":"2026-05-25T02:01:22.330804+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH","json":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH.json","graph_json":"https://pith.science/api/pith-number/N7ZGSYXBVROCMHSLONW7LMC5SH/graph.json","events_json":"https://pith.science/api/pith-number/N7ZGSYXBVROCMHSLONW7LMC5SH/events.json","paper":"https://pith.science/paper/N7ZGSYXB"},"agent_actions":{"view_html":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH","download_json":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH.json","view_paper":"https://pith.science/paper/N7ZGSYXB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.06840&json=true","fetch_graph":"https://pith.science/api/pith-number/N7ZGSYXBVROCMHSLONW7LMC5SH/graph.json","fetch_events":"https://pith.science/api/pith-number/N7ZGSYXBVROCMHSLONW7LMC5SH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH/action/storage_attestation","attest_author":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH/action/author_attestation","sign_citation":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH/action/citation_signature","submit_replication":"https://pith.science/pith/N7ZGSYXBVROCMHSLONW7LMC5SH/action/replication_record"}},"created_at":"2026-05-25T02:01:22.330804+00:00","updated_at":"2026-05-25T02:01:22.330804+00:00"}