{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:4SODO3Y4OIVZA4CQEGTKSBEWK4","short_pith_number":"pith:4SODO3Y4","canonical_record":{"source":{"id":"2605.17273","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T05:48:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dd51faa155daded0058f63e1b3d91932f98a9e1270ab0705ce4436148f0d1863","abstract_canon_sha256":"794e2ff12d9423714216c0a789c6f6be093293f38c60f73aa4944a60912b8203"},"schema_version":"1.0"},"canonical_sha256":"e49c376f1c722b90705021a6a9049657234099d8854c1e2378919093d622eed5","source":{"kind":"arxiv","id":"2605.17273","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.17273","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"arxiv_version","alias_value":"2605.17273v1","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.17273","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"pith_short_12","alias_value":"4SODO3Y4OIVZ","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"pith_short_16","alias_value":"4SODO3Y4OIVZA4CQ","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"pith_short_8","alias_value":"4SODO3Y4","created_at":"2026-05-20T00:03:49Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:4SODO3Y4OIVZA4CQEGTKSBEWK4","target":"record","payload":{"canonical_record":{"source":{"id":"2605.17273","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T05:48:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dd51faa155daded0058f63e1b3d91932f98a9e1270ab0705ce4436148f0d1863","abstract_canon_sha256":"794e2ff12d9423714216c0a789c6f6be093293f38c60f73aa4944a60912b8203"},"schema_version":"1.0"},"canonical_sha256":"e49c376f1c722b90705021a6a9049657234099d8854c1e2378919093d622eed5","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:49.159400Z","signature_b64":"ezyIkwIO70yKPcbOzM5WF3hk1VDTzUMOUO09/SQogg90F0ZRb4Tz2vfMxyt/R2s5P0060p6MUNSPFAgcWyfqDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e49c376f1c722b90705021a6a9049657234099d8854c1e2378919093d622eed5","last_reissued_at":"2026-05-20T00:03:49.158501Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:49.158501Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.17273","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:03:49Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Aql5h3lg1O1flthE3KdpflnRZKNtkkKMpJbr8F6ESGDfp7cw6Vf0Ibm7cYHnyZaTjrH6lNpTp3Tuuo8dBzFIDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T00:14:21.840486Z"},"content_sha256":"1ab9b3a3745b6262ea00ad7f10c5711eb0004f2e98413517a17f8a62f863f084","schema_version":"1.0","event_id":"sha256:1ab9b3a3745b6262ea00ad7f10c5711eb0004f2e98413517a17f8a62f863f084"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:4SODO3Y4OIVZA4CQEGTKSBEWK4","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"State-of-the-Art Claims Require State-of-the-Art Evidence","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"YongKyung Oh","submitted_at":"2026-05-17T05:48:10Z","abstract_excerpt":"State-of-the-Art (SOTA) claims pervade Artificial Intelligence (AI) and Machine Learning (ML) research. These claims rest on benchmark evaluations, where models are ranked by aggregate scores across tasks. Public benchmarks or leaderboards are the most visible instance, but the same structure appears in paper tables throughout the literature. However, such minimal evidence often cannot support these strong claims. We identify a widespread claim-evidence gap in AI benchmarking. Claiming SOTA carries implicit assumptions beyond mean score superiority, suggesting that a model meaningfully outperf"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.17273","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/2605.17273/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-19T22:01:57.834587Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.776894Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"9025c8d6524da973b3b47273b9f029a696b75b834097e10afaa1583b1aeb0aa7"},"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:03:49Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"i0r1AsnblaVBdfWocrluViAGug4nFbTGFsQVN8I1De3/nRuWg2oYK7MlGmQNQ4vXhHbto9dx/MohN1zYUQebCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T00:14:21.841292Z"},"content_sha256":"36b1847e92604d3a8ac87033082c28eb014780539a508a3a815a966aadf11ab9","schema_version":"1.0","event_id":"sha256:36b1847e92604d3a8ac87033082c28eb014780539a508a3a815a966aadf11ab9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4/bundle.json","state_url":"https://pith.science/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-07T00:14:21Z","links":{"resolver":"https://pith.science/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4","bundle":"https://pith.science/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4/bundle.json","state":"https://pith.science/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/4SODO3Y4OIVZA4CQEGTKSBEWK4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:4SODO3Y4OIVZA4CQEGTKSBEWK4","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"794e2ff12d9423714216c0a789c6f6be093293f38c60f73aa4944a60912b8203","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T05:48:10Z","title_canon_sha256":"dd51faa155daded0058f63e1b3d91932f98a9e1270ab0705ce4436148f0d1863"},"schema_version":"1.0","source":{"id":"2605.17273","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.17273","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"arxiv_version","alias_value":"2605.17273v1","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.17273","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"pith_short_12","alias_value":"4SODO3Y4OIVZ","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"pith_short_16","alias_value":"4SODO3Y4OIVZA4CQ","created_at":"2026-05-20T00:03:49Z"},{"alias_kind":"pith_short_8","alias_value":"4SODO3Y4","created_at":"2026-05-20T00:03:49Z"}],"graph_snapshots":[{"event_id":"sha256:36b1847e92604d3a8ac87033082c28eb014780539a508a3a815a966aadf11ab9","target":"graph","created_at":"2026-05-20T00:03:49Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"claim_evidence","ran_at":"2026-05-19T22:01:57.834587Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.776894Z","status":"skipped","version":"1.0.0"}],"endpoint":"/pith/2605.17273/integrity.json","findings":[],"snapshot_sha256":"9025c8d6524da973b3b47273b9f029a696b75b834097e10afaa1583b1aeb0aa7","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"State-of-the-Art (SOTA) claims pervade Artificial Intelligence (AI) and Machine Learning (ML) research. These claims rest on benchmark evaluations, where models are ranked by aggregate scores across tasks. Public benchmarks or leaderboards are the most visible instance, but the same structure appears in paper tables throughout the literature. However, such minimal evidence often cannot support these strong claims. We identify a widespread claim-evidence gap in AI benchmarking. Claiming SOTA carries implicit assumptions beyond mean score superiority, suggesting that a model meaningfully outperf","authors_text":"YongKyung Oh","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T05:48:10Z","title":"State-of-the-Art Claims Require State-of-the-Art Evidence"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.17273","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:1ab9b3a3745b6262ea00ad7f10c5711eb0004f2e98413517a17f8a62f863f084","target":"record","created_at":"2026-05-20T00:03:49Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"794e2ff12d9423714216c0a789c6f6be093293f38c60f73aa4944a60912b8203","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-17T05:48:10Z","title_canon_sha256":"dd51faa155daded0058f63e1b3d91932f98a9e1270ab0705ce4436148f0d1863"},"schema_version":"1.0","source":{"id":"2605.17273","kind":"arxiv","version":1}},"canonical_sha256":"e49c376f1c722b90705021a6a9049657234099d8854c1e2378919093d622eed5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e49c376f1c722b90705021a6a9049657234099d8854c1e2378919093d622eed5","first_computed_at":"2026-05-20T00:03:49.158501Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:03:49.158501Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ezyIkwIO70yKPcbOzM5WF3hk1VDTzUMOUO09/SQogg90F0ZRb4Tz2vfMxyt/R2s5P0060p6MUNSPFAgcWyfqDA==","signature_status":"signed_v1","signed_at":"2026-05-20T00:03:49.159400Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.17273","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1ab9b3a3745b6262ea00ad7f10c5711eb0004f2e98413517a17f8a62f863f084","sha256:36b1847e92604d3a8ac87033082c28eb014780539a508a3a815a966aadf11ab9"],"state_sha256":"623fd3ce461f33edb6cdadba98388eefec37611a056cbf6851d6adeb2372007f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vFg7gIUIdNTa9haXf7aLvS30OZraFzb5LkxoNE1ltrnjYoT1iA7e6IgCZ+zvU980BF+3pwgGZQjjQIoQ5uNVCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-07T00:14:21.845042Z","bundle_sha256":"689670800ad78a8e4c7f694c0f39fb1f1c743ba7ccb16626250c29b7a8f9539e"}}