{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:W7O4RW2ZGHNIFOXSDFMTJA26OY","short_pith_number":"pith:W7O4RW2Z","schema_version":"1.0","canonical_sha256":"b7ddc8db5931da82baf2195934835e761a2b5991326aa6f0a8fd95919965450e","source":{"kind":"arxiv","id":"2605.23170","version":1},"attestation_state":"computed","paper":{"title":"Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chuyifei Zhang, Hongyu Cui, Jitao Sang, Xiaowen Huang","submitted_at":"2026-05-22T02:42:41Z","abstract_excerpt":"Position-controlled evaluation is standard for retrieval tasks such as Needle-in-a-Haystack and RULER, but mainstream reasoning benchmarks do not control positional placement of target tasks in long contexts. We audit 11 long-context benchmarks and find none jointly controls task position, filler content, and context length for reasoning. An audit of four flagship long-context releases finds no main result-table entry for NIAH, RULER, or LongBench-family benchmarks, while agentic and coding benchmarks appear in main result-tables across all four. We propose Context Rot Evaluation (CRE), a cont"},"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":"2605.23170","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-22T02:42:41Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"323fce018310459f2dd15bbbdf9aba3caa3e8b24747500be6b69d483c637018c","abstract_canon_sha256":"8eff69306e543ac7678b4c70b83726744cfefcaeefb99c4cee8e8368635da963"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:41.374633Z","signature_b64":"yxJ5yUNvTZQMQgNuXFW1G6twLv/udZnkfkCZHVxCHq7fr/k1VeI/p7a9ZnM2Y167VNHH5JqpNf0gXCTmy/kZCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b7ddc8db5931da82baf2195934835e761a2b5991326aa6f0a8fd95919965450e","last_reissued_at":"2026-05-25T02:01:41.373904Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:41.373904Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chuyifei Zhang, Hongyu Cui, Jitao Sang, Xiaowen Huang","submitted_at":"2026-05-22T02:42:41Z","abstract_excerpt":"Position-controlled evaluation is standard for retrieval tasks such as Needle-in-a-Haystack and RULER, but mainstream reasoning benchmarks do not control positional placement of target tasks in long contexts. We audit 11 long-context benchmarks and find none jointly controls task position, filler content, and context length for reasoning. An audit of four flagship long-context releases finds no main result-table entry for NIAH, RULER, or LongBench-family benchmarks, while agentic and coding benchmarks appear in main result-tables across all four. We propose Context Rot Evaluation (CRE), a cont"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.23170","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/2605.23170/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":"2605.23170","created_at":"2026-05-25T02:01:41.374029+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.23170v1","created_at":"2026-05-25T02:01:41.374029+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.23170","created_at":"2026-05-25T02:01:41.374029+00:00"},{"alias_kind":"pith_short_12","alias_value":"W7O4RW2ZGHNI","created_at":"2026-05-25T02:01:41.374029+00:00"},{"alias_kind":"pith_short_16","alias_value":"W7O4RW2ZGHNIFOXS","created_at":"2026-05-25T02:01:41.374029+00:00"},{"alias_kind":"pith_short_8","alias_value":"W7O4RW2Z","created_at":"2026-05-25T02:01:41.374029+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/W7O4RW2ZGHNIFOXSDFMTJA26OY","json":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY.json","graph_json":"https://pith.science/api/pith-number/W7O4RW2ZGHNIFOXSDFMTJA26OY/graph.json","events_json":"https://pith.science/api/pith-number/W7O4RW2ZGHNIFOXSDFMTJA26OY/events.json","paper":"https://pith.science/paper/W7O4RW2Z"},"agent_actions":{"view_html":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY","download_json":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY.json","view_paper":"https://pith.science/paper/W7O4RW2Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.23170&json=true","fetch_graph":"https://pith.science/api/pith-number/W7O4RW2ZGHNIFOXSDFMTJA26OY/graph.json","fetch_events":"https://pith.science/api/pith-number/W7O4RW2ZGHNIFOXSDFMTJA26OY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY/action/storage_attestation","attest_author":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY/action/author_attestation","sign_citation":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY/action/citation_signature","submit_replication":"https://pith.science/pith/W7O4RW2ZGHNIFOXSDFMTJA26OY/action/replication_record"}},"created_at":"2026-05-25T02:01:41.374029+00:00","updated_at":"2026-05-25T02:01:41.374029+00:00"}