{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JTGXWGS3N42ROGYLFJ6RU4BJEN","short_pith_number":"pith:JTGXWGS3","schema_version":"1.0","canonical_sha256":"4ccd7b1a5b6f35171b0b2a7d1a7029237d12c08e8508eb2979e9bc0d5c3d0960","source":{"kind":"arxiv","id":"2607.08065","version":1},"attestation_state":"computed","paper":{"title":"When LLMs Agree, Are They Right? Auditing Self-Consistency and Cross-Model Agreement as Confidence Signals","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Kaihua Ding","submitted_at":"2026-07-09T02:46:51Z","abstract_excerpt":"LLM-as-judge (Zheng et al., 2023) is increasingly the default for evaluating AI systems in enterprise pipelines, often scaled to ensembles (Verga et al., 2024) or \"mixture-of-experts\" (Shazeer et al., 2017) panels of judges. These systems share a key assumption: that consistency -- agreement among judges, or among a model's own samples -- indicates correctness. We show this assumption is unreliable. Agreement is not accuracy: a model can agree with itself, and different models can agree with each other, out of shared bias, a memorized heuristic, or an option-position prior rather than truth. W"},"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":"2607.08065","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-09T02:46:51Z","cross_cats_sorted":[],"title_canon_sha256":"97472542c9c0b281f311a374ecff721d463fa997ff1fd6add09446b224acca11","abstract_canon_sha256":"cbbea24667a8099545dde38f56eb6510e46e89ea58c8f6479f58115b81fd0701"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-10T00:19:09.356571Z","signature_b64":"liLAKAX82VmkYmTUcb8lsOwrSpFzqiuJ0Mzhx67FgX0r1TuAzwwUzV1wa18T2ComcK7/SEhgsKoyLtyEfjy/Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4ccd7b1a5b6f35171b0b2a7d1a7029237d12c08e8508eb2979e9bc0d5c3d0960","last_reissued_at":"2026-07-10T00:19:09.356175Z","signature_status":"signed_v1","first_computed_at":"2026-07-10T00:19:09.356175Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When LLMs Agree, Are They Right? Auditing Self-Consistency and Cross-Model Agreement as Confidence Signals","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Kaihua Ding","submitted_at":"2026-07-09T02:46:51Z","abstract_excerpt":"LLM-as-judge (Zheng et al., 2023) is increasingly the default for evaluating AI systems in enterprise pipelines, often scaled to ensembles (Verga et al., 2024) or \"mixture-of-experts\" (Shazeer et al., 2017) panels of judges. These systems share a key assumption: that consistency -- agreement among judges, or among a model's own samples -- indicates correctness. We show this assumption is unreliable. Agreement is not accuracy: a model can agree with itself, and different models can agree with each other, out of shared bias, a memorized heuristic, or an option-position prior rather than truth. W"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.08065","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/2607.08065/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":"2607.08065","created_at":"2026-07-10T00:19:09.356239+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.08065v1","created_at":"2026-07-10T00:19:09.356239+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.08065","created_at":"2026-07-10T00:19:09.356239+00:00"},{"alias_kind":"pith_short_12","alias_value":"JTGXWGS3N42R","created_at":"2026-07-10T00:19:09.356239+00:00"},{"alias_kind":"pith_short_16","alias_value":"JTGXWGS3N42ROGYL","created_at":"2026-07-10T00:19:09.356239+00:00"},{"alias_kind":"pith_short_8","alias_value":"JTGXWGS3","created_at":"2026-07-10T00:19:09.356239+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/JTGXWGS3N42ROGYLFJ6RU4BJEN","json":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN.json","graph_json":"https://pith.science/api/pith-number/JTGXWGS3N42ROGYLFJ6RU4BJEN/graph.json","events_json":"https://pith.science/api/pith-number/JTGXWGS3N42ROGYLFJ6RU4BJEN/events.json","paper":"https://pith.science/paper/JTGXWGS3"},"agent_actions":{"view_html":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN","download_json":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN.json","view_paper":"https://pith.science/paper/JTGXWGS3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.08065&json=true","fetch_graph":"https://pith.science/api/pith-number/JTGXWGS3N42ROGYLFJ6RU4BJEN/graph.json","fetch_events":"https://pith.science/api/pith-number/JTGXWGS3N42ROGYLFJ6RU4BJEN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN/action/storage_attestation","attest_author":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN/action/author_attestation","sign_citation":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN/action/citation_signature","submit_replication":"https://pith.science/pith/JTGXWGS3N42ROGYLFJ6RU4BJEN/action/replication_record"}},"created_at":"2026-07-10T00:19:09.356239+00:00","updated_at":"2026-07-10T00:19:09.356239+00:00"}