{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YCW5M656WQN6RHL6BSR6V27B2U","short_pith_number":"pith:YCW5M656","schema_version":"1.0","canonical_sha256":"c0add67bbeb41be89d7e0ca3eaebe1d536b2b053f3514f0bf10b4dccb26210cd","source":{"kind":"arxiv","id":"2606.05932","version":1},"attestation_state":"computed","paper":{"title":"A Pre-Registered Causal Partition of Self-Consistency Elicitation and Reward Design in RLVR","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Yuze Gao","submitted_at":"2026-06-04T09:35:54Z","abstract_excerpt":"Reinforcement learning from verifiable rewards (RLVR) improves reasoning even when the reward signal is spurious -- assigning credit to the group-plurality answer rather than a ground-truth verifier. Practitioners commonly interpret naive = acc(TRUE) - acc(RANDOM) as the reward-design effect. We prove this estimand is systematically biased: it conflates self-consistency elicitation (sharpening the policy toward its modal answer via majority pseudo-reward) with genuine reward-design signal. Using a controlled tabular-GRPO simulator we derive an exact telescoping decomposition total = null + eli"},"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":"2606.05932","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-04T09:35:54Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a483e70e3dbd69c82b8ecdaa644654ad4bef9daac0391248ec1cbd9587c4140a","abstract_canon_sha256":"10c66ef00ed80f178a2f38b9ad72182b180c0e85c0518d1ba3d4f7fb85a6fa96"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:15:28.096156Z","signature_b64":"z9BZ2RNeCMmGzCnnRRKSwSkQXNc23vIEynzpTpnSjDU4R4IA9RRcN5lNDxXNQDLuw5gtQWtk0yTaSSVxbdWNAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c0add67bbeb41be89d7e0ca3eaebe1d536b2b053f3514f0bf10b4dccb26210cd","last_reissued_at":"2026-06-05T01:15:28.095415Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:15:28.095415Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Pre-Registered Causal Partition of Self-Consistency Elicitation and Reward Design in RLVR","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Yuze Gao","submitted_at":"2026-06-04T09:35:54Z","abstract_excerpt":"Reinforcement learning from verifiable rewards (RLVR) improves reasoning even when the reward signal is spurious -- assigning credit to the group-plurality answer rather than a ground-truth verifier. Practitioners commonly interpret naive = acc(TRUE) - acc(RANDOM) as the reward-design effect. We prove this estimand is systematically biased: it conflates self-consistency elicitation (sharpening the policy toward its modal answer via majority pseudo-reward) with genuine reward-design signal. Using a controlled tabular-GRPO simulator we derive an exact telescoping decomposition total = null + eli"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.05932","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.05932/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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":"2606.05932","created_at":"2026-06-05T01:15:28.095523+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.05932v1","created_at":"2026-06-05T01:15:28.095523+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.05932","created_at":"2026-06-05T01:15:28.095523+00:00"},{"alias_kind":"pith_short_12","alias_value":"YCW5M656WQN6","created_at":"2026-06-05T01:15:28.095523+00:00"},{"alias_kind":"pith_short_16","alias_value":"YCW5M656WQN6RHL6","created_at":"2026-06-05T01:15:28.095523+00:00"},{"alias_kind":"pith_short_8","alias_value":"YCW5M656","created_at":"2026-06-05T01:15:28.095523+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/YCW5M656WQN6RHL6BSR6V27B2U","json":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U.json","graph_json":"https://pith.science/api/pith-number/YCW5M656WQN6RHL6BSR6V27B2U/graph.json","events_json":"https://pith.science/api/pith-number/YCW5M656WQN6RHL6BSR6V27B2U/events.json","paper":"https://pith.science/paper/YCW5M656"},"agent_actions":{"view_html":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U","download_json":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U.json","view_paper":"https://pith.science/paper/YCW5M656","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.05932&json=true","fetch_graph":"https://pith.science/api/pith-number/YCW5M656WQN6RHL6BSR6V27B2U/graph.json","fetch_events":"https://pith.science/api/pith-number/YCW5M656WQN6RHL6BSR6V27B2U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U/action/storage_attestation","attest_author":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U/action/author_attestation","sign_citation":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U/action/citation_signature","submit_replication":"https://pith.science/pith/YCW5M656WQN6RHL6BSR6V27B2U/action/replication_record"}},"created_at":"2026-06-05T01:15:28.095523+00:00","updated_at":"2026-06-05T01:15:28.095523+00:00"}