{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:C2F3SPJCRVGJBASRU6OPTMGI5T","short_pith_number":"pith:C2F3SPJC","schema_version":"1.0","canonical_sha256":"168bb93d228d4c908251a79cf9b0c8ecfcb68b31359095c317b7c5688364f672","source":{"kind":"arxiv","id":"2606.28661","version":1},"attestation_state":"computed","paper":{"title":"When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Kathleen A. Yearick, Yong Yi Bay","submitted_at":"2026-06-27T00:37:33Z","abstract_excerpt":"People overthink; language models over-sample, and the extra effort can talk both into a worse answer. Reasoning systems answer a hard question by sampling it many times (test-time scaling), and the more they draw, the more often a correct answer turns up somewhere, so coverage, the fraction of problems with at least one correct try, climbs and appears to be progress. But a deployed system must return one answer, and choosing it, not knowing which try is right, is selection; selection is capped, and past a point extra samples only make the model surer of a confident mistake, even as every draw"},"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":"2606.28661","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-27T00:37:33Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"title_canon_sha256":"14b525ee2cd1a9ad568a85a6ad27980d27796d5884223b972f040f8c4adfaea1","abstract_canon_sha256":"4525f666404e5d7aa54ba1185ae7c8385b75b4e90d7e8c9edce093c871b4dd57"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:16:46.360836Z","signature_b64":"yfWaQ9hm7x4FwDn84T5wigVqQSt27+YkEhbCFM7knUk2vYDzzE876qUsC6t2MDWY1X1RTclzXGfhrtrvZNErAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"168bb93d228d4c908251a79cf9b0c8ecfcb68b31359095c317b7c5688364f672","last_reissued_at":"2026-06-30T01:16:46.360224Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:16:46.360224Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Kathleen A. Yearick, Yong Yi Bay","submitted_at":"2026-06-27T00:37:33Z","abstract_excerpt":"People overthink; language models over-sample, and the extra effort can talk both into a worse answer. Reasoning systems answer a hard question by sampling it many times (test-time scaling), and the more they draw, the more often a correct answer turns up somewhere, so coverage, the fraction of problems with at least one correct try, climbs and appears to be progress. But a deployed system must return one answer, and choosing it, not knowing which try is right, is selection; selection is capped, and past a point extra samples only make the model surer of a confident mistake, even as every draw"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.28661","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/2606.28661/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":"2606.28661","created_at":"2026-06-30T01:16:46.360310+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.28661v1","created_at":"2026-06-30T01:16:46.360310+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.28661","created_at":"2026-06-30T01:16:46.360310+00:00"},{"alias_kind":"pith_short_12","alias_value":"C2F3SPJCRVGJ","created_at":"2026-06-30T01:16:46.360310+00:00"},{"alias_kind":"pith_short_16","alias_value":"C2F3SPJCRVGJBASR","created_at":"2026-06-30T01:16:46.360310+00:00"},{"alias_kind":"pith_short_8","alias_value":"C2F3SPJC","created_at":"2026-06-30T01:16:46.360310+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.00152","citing_title":"GRPO, Dr. GRPO, and DAPO Are Three Operations on One Number: The Group-Standard-Deviation Identity","ref_index":12,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T","json":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T.json","graph_json":"https://pith.science/api/pith-number/C2F3SPJCRVGJBASRU6OPTMGI5T/graph.json","events_json":"https://pith.science/api/pith-number/C2F3SPJCRVGJBASRU6OPTMGI5T/events.json","paper":"https://pith.science/paper/C2F3SPJC"},"agent_actions":{"view_html":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T","download_json":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T.json","view_paper":"https://pith.science/paper/C2F3SPJC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.28661&json=true","fetch_graph":"https://pith.science/api/pith-number/C2F3SPJCRVGJBASRU6OPTMGI5T/graph.json","fetch_events":"https://pith.science/api/pith-number/C2F3SPJCRVGJBASRU6OPTMGI5T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T/action/storage_attestation","attest_author":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T/action/author_attestation","sign_citation":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T/action/citation_signature","submit_replication":"https://pith.science/pith/C2F3SPJCRVGJBASRU6OPTMGI5T/action/replication_record"}},"created_at":"2026-06-30T01:16:46.360310+00:00","updated_at":"2026-06-30T01:16:46.360310+00:00"}