{"id":"45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b","title":"Researchers test how agents build evidence to verify scientific consensus claims","argument":"The authors argue that scientific agents can better verify whether research literature supports or refutes a claim by maintaining a structured evidence state outside the model’s context, rather than accumulating evidence only within that context. They note that these agents are usually evaluated end-to-end without ground truth, leaving the effects of individual design choices unknown. They study two choices: what information the state contains and how a judge decides when enough evidence has been gathered. Their ProClaim instrument keeps this state separate so the choices can be examined while holding the other fixed.\n\nThe authors evaluate the approach on ProClaim-eval, a dataset of 718 claims with expert-curated SUPPORT and REFUTE labels from five databases in molecular biology, nuclear physics and astronomy. They report that the evidence state improves performance over in-context accumulation, and that what is stored matters. A trained judge that reads the state performs better than fixed-round stopping. A separately trained judge achieves the highest agreement with gold labels on three of five subsets, matching the off-the-shelf Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5. The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for this task.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported evaluation uses 718 claims with gold labels from five expert-curated databases in molecular biology, nuclear physics and astronomy; the abstract does not report results beyond this dataset. The abstract reports the separately trained judge’s highest agreement on three of five subsets, not all five.","speaker":{"name":"Ai, Lun, Wu, Yu-Heng, Schaul, Jonathan, Küchenhoff, Leonie, Turei, Denes, Carli, Francesco, Petroni, Fabio, Dugourd, Aurelien, Saez-Rodriguez, Julio","roleAtSource":"Article author"},"source":{"title":"Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.40906","publishedAt":"2026-10-29","dateBasis":"Date basis not provided."},"evidence":[{"text":"The paper examines what a scientific agent recursively builds as an evidence state and what controls when it stops gathering evidence.","excerpt":"we examined two design axes: (1) what the recursion builds (the state of evidence) and (2) what controls it (the judge that decides when to stop).","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"ProClaim-eval contains 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases across three scientific fields.","excerpt":"including 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases spanning molecular biology, nuclear physics, and astronomy.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"The authors report that maintaining an evidence state improves performance over accumulating evidence in the model’s context, and that the contents of the state matter.","excerpt":"We found that the evidence state improves performance over in-context accumulation, and it also matters what is stored in it.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"The authors report that a trained judge reading the evidence state performs better than stopping after a fixed number of rounds.","excerpt":"a trained judge that reads the evidence state performs better than fixed-round stopping.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"A separately trained judge achieved the highest agreement with gold labels on three of five subsets, matching the Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5.","excerpt":"A judge trained separately and never exposed to ProClaim-eval achieves the highest agreement with the gold labels on three of five subsets, where it matches the off-the-shelf Claude Sonnet~4.6 judge and outperforms Claude Haiku~4.5.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for verifying scientific claims in their task.","excerpt":"Our results support recursively maintaining a structured evidence state outside the model's context as an effective design for verifying claims in scientific agents.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.40906","checkedAt":"2026-11-11T05:02:08.669Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would the reported results change when ProClaim is evaluated on consensus claims beyond the five expert-curated databases?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Researchers test how agents build evidence to verify scientific consensus claims","summary":"The authors argue that scientific agents can better verify whether research literature supports or refutes a claim by maintaining a structured evidence state outside the model’s context, rather than accumulating evidence only within that context. They note that these agents are usually evaluated end-to-end without ground truth, leaving the effects of individual design choices unknown. They study two choices: what information the state contains and how a judge decides when enough evidence has been gathered. Their ProClaim instrument keeps this state separate so the choices can be examined while holding the other fixed.\n\nThe authors evaluate the approach on ProClaim-eval, a dataset of 718 claims with expert-curated SUPPORT and REFUTE labels from five databases in molecular biology, nuclear physics and astronomy. They report that the evidence state improves performance over in-context accumulation, and that what is stored matters. A trained judge that reads the state performs better than fixed-round stopping. A separately trained judge achieves the highest agreement with gold labels on three of five subsets, matching the off-the-shelf Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5. The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for this task.","question":"How would the reported results change when ProClaim is evaluated on consensus claims beyond the five expert-curated databases?","claims":[{"text":"The paper examines what a scientific agent recursively builds as an evidence state and what controls when it stops gathering evidence.","excerpt":"we examined two design axes: (1) what the recursion builds (the state of evidence) and (2) what controls it (the judge that decides when to stop).","support":"source-stated"},{"text":"ProClaim-eval contains 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases across three scientific fields.","excerpt":"including 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases spanning molecular biology, nuclear physics, and astronomy.","support":"source-stated"},{"text":"The authors report that maintaining an evidence state improves performance over accumulating evidence in the model’s context, and that the contents of the state matter.","excerpt":"We found that the evidence state improves performance over in-context accumulation, and it also matters what is stored in it.","support":"source-stated"},{"text":"The authors report that a trained judge reading the evidence state performs better than stopping after a fixed number of rounds.","excerpt":"a trained judge that reads the evidence state performs better than fixed-round stopping.","support":"source-stated"},{"text":"A separately trained judge achieved the highest agreement with gold labels on three of five subsets, matching the Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5.","excerpt":"A judge trained separately and never exposed to ProClaim-eval achieves the highest agreement with the gold labels on three of five subsets, where it matches the off-the-shelf Claude Sonnet~4.6 judge and outperforms Claude Haiku~4.5.","support":"source-stated"},{"text":"The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for verifying scientific claims in their task.","excerpt":"Our results support recursively maintaining a structured evidence state outside the model's context as an effective design for verifying claims in scientific agents.","support":"source-stated"}],"limitations":["The reported evaluation uses 718 claims with gold labels from five expert-curated databases in molecular biology, nuclear physics and astronomy; the abstract does not report results beyond this dataset.","The abstract reports the separately trained judge’s highest agreement on three of five subsets, not all five."],"mediaId":"cf02b25fdc7325f51edc15724cca2bc012f979bb710741ec7ecee0f3611e92c1","correctionReason":null,"id":"45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b","publicationId":"c163a282-824d-42e5-8bd8-8202690d2fba","version":1,"sourceUrl":"https://arxiv.org/abs/2610.40906","sourceTitle":"Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-29","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-02T05:01:45.269Z","createdAt":"2026-11-11T02:55:39.483Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"cf02b25fdc7325f51edc15724cca2bc012f979bb710741ec7ecee0f3611e92c1","kind":"article","url":"https://arxiv.org/abs/2610.40906","sourceUrl":"https://arxiv.org/abs/2610.40906","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-02T11:48:47.572Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.40906","checkedAt":"2026-11-11T05:02:08.669Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"cf02b25fdc7325f51edc15724cca2bc012f979bb710741ec7ecee0f3611e92c1","confidence":0.6799999999999999,"provider":"typesafe","model":"jev-1.13.0","inputHash":"662af6d29a1aa3d2b063d6d11989acbc7aa93afe664c0bb6bc0b639147c12561","decidedAt":"2026-11-02T11:48:48.436Z","redecided":{"from":"stream-admission-v5","under":"stream-admission-v5+confidence-0.5"}},"provenance":{"textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observationId":"38728a90d992fc6dd6ccc58ba37f679644385ee4dcdff622c2780275e0ba4d34","retrievedAt":"2026-11-02T11:48:47.472Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"1997292a-390c-4e70-8e34-7e56899e3db7"},"reviewer":{"model":"gpt-6-luna","runId":"baf9572c-ff32-435c-8b84-52cfe937b499"}},"audit":{"status":"passed","reviewedAt":"2026-11-11T02:55:39.483Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Ai, Lun","url":null,"kind":"unknown"},{"name":"Wu, Yu-Heng","url":null,"kind":"unknown"},{"name":"Schaul, Jonathan","url":null,"kind":"unknown"},{"name":"Küchenhoff, Leonie","url":null,"kind":"unknown"},{"name":"Turei, Denes","url":null,"kind":"unknown"},{"name":"Carli, Francesco","url":null,"kind":"unknown"},{"name":"Petroni, Fabio","url":null,"kind":"unknown"},{"name":"Dugourd, Aurelien","url":null,"kind":"unknown"},{"name":"Saez-Rodriguez, Julio","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Ai, Lun","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ai, Lun","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Wu, Yu-Heng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wu, Yu-Heng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Schaul, Jonathan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Schaul, Jonathan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Küchenhoff, Leonie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Küchenhoff, Leonie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Turei, Denes","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Turei, Denes","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Carli, Francesco","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Carli, Francesco","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Petroni, Fabio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Petroni, Fabio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Dugourd, Aurelien","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Dugourd, Aurelien","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Saez-Rodriguez, Julio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Saez-Rodriguez, Julio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}}],"participationCoverage":{"total":9,"returned":9,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.40906","snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b/source"}},"canonicalUrl":"https://escapemeridian.com/readings/researchers-test-how-agents-build-evidence-to-verify-scientific-45edd98c","formats":{"html":"/readings/researchers-test-how-agents-build-evidence-to-verify-scientific-45edd98c","markdown":"/readings/researchers-test-how-agents-build-evidence-to-verify-scientific-45edd98c.md","json":"/readings/researchers-test-how-agents-build-evidence-to-verify-scientific-45edd98c.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"},{"id":"claude","name":"Claude","kind":"model","context":"Claude is Anthropic’s family of AI models and assistants for tasks such as writing, analysis and coding.","referenceUrls":["https://claude.com/product/overview","https://www.anthropic.com/news","https://www.anthropic.com/research/team/economics"],"path":"/entities/claude","json":"/entities/claude.json","markdown":"/entities/claude.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b","revision":"c163a282-824d-42e5-8bd8-8202690d2fba","revisionKind":"publication-id","title":"Researchers test how agents build evidence to verify scientific consensus claims","url":"https://escapemeridian.com/readings/researchers-test-how-agents-build-evidence-to-verify-scientific-45edd98c"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.40906","revision":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","revisionKind":"sha256-source-text","title":"Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?","url":"https://arxiv.org/abs/2610.40906"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b","title":"Researchers test how agents build evidence to verify scientific consensus claims","argument":"The authors argue that scientific agents can better verify whether research literature supports or refutes a claim by maintaining a structured evidence state outside the model’s context, rather than accumulating evidence only within that context. They note that these agents are usually evaluated end-to-end without ground truth, leaving the effects of individual design choices unknown. They study two choices: what information the state contains and how a judge decides when enough evidence has been gathered. Their ProClaim instrument keeps this state separate so the choices can be examined while holding the other fixed.\n\nThe authors evaluate the approach on ProClaim-eval, a dataset of 718 claims with expert-curated SUPPORT and REFUTE labels from five databases in molecular biology, nuclear physics and astronomy. They report that the evidence state improves performance over in-context accumulation, and that what is stored matters. A trained judge that reads the state performs better than fixed-round stopping. A separately trained judge achieves the highest agreement with gold labels on three of five subsets, matching the off-the-shelf Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5. The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for this task.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported evaluation uses 718 claims with gold labels from five expert-curated databases in molecular biology, nuclear physics and astronomy; the abstract does not report results beyond this dataset. The abstract reports the separately trained judge’s highest agreement on three of five subsets, not all five.","speaker":{"name":"Ai, Lun, Wu, Yu-Heng, Schaul, Jonathan, Küchenhoff, Leonie, Turei, Denes, Carli, Francesco, Petroni, Fabio, Dugourd, Aurelien, Saez-Rodriguez, Julio","roleAtSource":"Article author"},"source":{"title":"Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.40906","publishedAt":"2026-10-29","dateBasis":"Date basis not provided."},"evidence":[{"text":"The paper examines what a scientific agent recursively builds as an evidence state and what controls when it stops gathering evidence.","excerpt":"we examined two design axes: (1) what the recursion builds (the state of evidence) and (2) what controls it (the judge that decides when to stop).","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"ProClaim-eval contains 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases across three scientific fields.","excerpt":"including 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases spanning molecular biology, nuclear physics, and astronomy.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"The authors report that maintaining an evidence state improves performance over accumulating evidence in the model’s context, and that the contents of the state matter.","excerpt":"We found that the evidence state improves performance over in-context accumulation, and it also matters what is stored in it.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"The authors report that a trained judge reading the evidence state performs better than stopping after a fixed number of rounds.","excerpt":"a trained judge that reads the evidence state performs better than fixed-round stopping.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"A separately trained judge achieved the highest agreement with gold labels on three of five subsets, matching the Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5.","excerpt":"A judge trained separately and never exposed to ProClaim-eval achieves the highest agreement with the gold labels on three of five subsets, where it matches the off-the-shelf Claude Sonnet~4.6 judge and outperforms Claude Haiku~4.5.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"},{"text":"The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for verifying scientific claims in their task.","excerpt":"Our results support recursively maintaining a structured evidence state outside the model's context as an effective design for verifying claims in scientific agents.","url":"https://arxiv.org/abs/2610.40906","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.40906","checkedAt":"2026-11-11T05:02:08.669Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"How would the reported results change when ProClaim is evaluated on consensus claims beyond the five expert-curated databases?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Researchers test how agents build evidence to verify scientific consensus claims","summary":"The authors argue that scientific agents can better verify whether research literature supports or refutes a claim by maintaining a structured evidence state outside the model’s context, rather than accumulating evidence only within that context. They note that these agents are usually evaluated end-to-end without ground truth, leaving the effects of individual design choices unknown. They study two choices: what information the state contains and how a judge decides when enough evidence has been gathered. Their ProClaim instrument keeps this state separate so the choices can be examined while holding the other fixed.\n\nThe authors evaluate the approach on ProClaim-eval, a dataset of 718 claims with expert-curated SUPPORT and REFUTE labels from five databases in molecular biology, nuclear physics and astronomy. They report that the evidence state improves performance over in-context accumulation, and that what is stored matters. A trained judge that reads the state performs better than fixed-round stopping. A separately trained judge achieves the highest agreement with gold labels on three of five subsets, matching the off-the-shelf Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5. The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for this task.","question":"How would the reported results change when ProClaim is evaluated on consensus claims beyond the five expert-curated databases?","claims":[{"text":"The paper examines what a scientific agent recursively builds as an evidence state and what controls when it stops gathering evidence.","excerpt":"we examined two design axes: (1) what the recursion builds (the state of evidence) and (2) what controls it (the judge that decides when to stop).","support":"source-stated"},{"text":"ProClaim-eval contains 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases across three scientific fields.","excerpt":"including 718 claims with SUPPORT and REFUTE gold labels from five expert-curated databases spanning molecular biology, nuclear physics, and astronomy.","support":"source-stated"},{"text":"The authors report that maintaining an evidence state improves performance over accumulating evidence in the model’s context, and that the contents of the state matter.","excerpt":"We found that the evidence state improves performance over in-context accumulation, and it also matters what is stored in it.","support":"source-stated"},{"text":"The authors report that a trained judge reading the evidence state performs better than stopping after a fixed number of rounds.","excerpt":"a trained judge that reads the evidence state performs better than fixed-round stopping.","support":"source-stated"},{"text":"A separately trained judge achieved the highest agreement with gold labels on three of five subsets, matching the Claude Sonnet 4.6 judge and outperforming Claude Haiku 4.5.","excerpt":"A judge trained separately and never exposed to ProClaim-eval achieves the highest agreement with the gold labels on three of five subsets, where it matches the off-the-shelf Claude Sonnet~4.6 judge and outperforms Claude Haiku~4.5.","support":"source-stated"},{"text":"The authors conclude that recursively maintaining a structured evidence state outside the model’s context is effective for verifying scientific claims in their task.","excerpt":"Our results support recursively maintaining a structured evidence state outside the model's context as an effective design for verifying claims in scientific agents.","support":"source-stated"}],"limitations":["The reported evaluation uses 718 claims with gold labels from five expert-curated databases in molecular biology, nuclear physics and astronomy; the abstract does not report results beyond this dataset.","The abstract reports the separately trained judge’s highest agreement on three of five subsets, not all five."],"mediaId":"cf02b25fdc7325f51edc15724cca2bc012f979bb710741ec7ecee0f3611e92c1","correctionReason":null,"id":"45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b","publicationId":"c163a282-824d-42e5-8bd8-8202690d2fba","version":1,"sourceUrl":"https://arxiv.org/abs/2610.40906","sourceTitle":"Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-29","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-02T05:01:45.269Z","createdAt":"2026-11-11T02:55:39.483Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"cf02b25fdc7325f51edc15724cca2bc012f979bb710741ec7ecee0f3611e92c1","kind":"article","url":"https://arxiv.org/abs/2610.40906","sourceUrl":"https://arxiv.org/abs/2610.40906","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-02T11:48:47.572Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.40906","checkedAt":"2026-11-11T05:02:08.669Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"cf02b25fdc7325f51edc15724cca2bc012f979bb710741ec7ecee0f3611e92c1","confidence":0.6799999999999999,"provider":"typesafe","model":"jev-1.13.0","inputHash":"662af6d29a1aa3d2b063d6d11989acbc7aa93afe664c0bb6bc0b639147c12561","decidedAt":"2026-11-02T11:48:48.436Z","redecided":{"from":"stream-admission-v5","under":"stream-admission-v5+confidence-0.5"}},"provenance":{"textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observationId":"38728a90d992fc6dd6ccc58ba37f679644385ee4dcdff622c2780275e0ba4d34","retrievedAt":"2026-11-02T11:48:47.472Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"1997292a-390c-4e70-8e34-7e56899e3db7"},"reviewer":{"model":"gpt-6-luna","runId":"baf9572c-ff32-435c-8b84-52cfe937b499"}},"audit":{"status":"passed","reviewedAt":"2026-11-11T02:55:39.483Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Ai, Lun","url":null,"kind":"unknown"},{"name":"Wu, Yu-Heng","url":null,"kind":"unknown"},{"name":"Schaul, Jonathan","url":null,"kind":"unknown"},{"name":"Küchenhoff, Leonie","url":null,"kind":"unknown"},{"name":"Turei, Denes","url":null,"kind":"unknown"},{"name":"Carli, Francesco","url":null,"kind":"unknown"},{"name":"Petroni, Fabio","url":null,"kind":"unknown"},{"name":"Dugourd, Aurelien","url":null,"kind":"unknown"},{"name":"Saez-Rodriguez, Julio","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Ai, Lun","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ai, Lun","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Wu, Yu-Heng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wu, Yu-Heng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Schaul, Jonathan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Schaul, Jonathan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Küchenhoff, Leonie","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Küchenhoff, Leonie","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Turei, Denes","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Turei, Denes","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Carli, Francesco","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Carli, Francesco","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Petroni, Fabio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Petroni, Fabio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Dugourd, Aurelien","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Dugourd, Aurelien","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}},{"actor":{"name":"Saez-Rodriguez, Julio","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Saez-Rodriguez, Julio","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.40906","locator":"head/meta[name=author|citation_author]"},"assertionId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T05:02:08.669Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","sourceUrl":"https://arxiv.org/abs/2610.40906","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z"}}],"participationCoverage":{"total":9,"returned":9,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.40906","snapshotId":"2cbd6615a00aac71efd71648de6b2c2829de4af4dc7a1bf3b5a8ad2c9793be31","contentSha256":"7eec83c65b17344e9a14413d4252c8bccd3cdd4b2f4c926c7ef8672130fe0476","textHash":"31deba25b85a7459be20853d7b1a37ec655d3a7a76730b893c50ac66afe4b365","observedAt":"2026-11-11T05:02:08.669Z","basis":"publisher-page-metadata","sourceLinks":47,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/45edd98cb1ae3b157f45b9621ef5d82d0fc41578c3aab008c4cfc356ea21916b/source"}}}],"audits":[{"publicationId":"c163a282-824d-42e5-8bd8-8202690d2fba","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The draft consistently attributes the findings and the conclusion to the paper's authors, using wording such as 'The authors argue', 'They report' and 'The authors conclude'. It does not present the results as established fact or as Escape Meridian's view. The paper is correctly identified as the arXiv submission by Lun Ai and colleagues.","excerpt":"Authors:Lun Ai, Yu-Heng Wu, Jonathan Schaul, Leonie Küchenhoff, Denes Turei, Francesco Carli, Fabio Petroni, Aurelien Dugourd, Julio Saez-Rodriguez"},{"dimension":"summary","status":"pass","explanation":"The title, summary and question faithfully reflect the abstract. The summary covers the two design axes, the ProClaim instrument, the 718-claim dataset and the main results, and it explains consensus verification in plain terms. Two details are compressed but not distorted. First, the draft omits that ProClaim ranks among the top three against verification systems and agent harnesses. Second, it omits that the separate judge was 'never exposed to ProClaim-eval'. Adding that phrase would better preserve the generalization point.","excerpt":"A judge trained separately and never exposed to ProClaim-eval achieves the highest agreement with the gold labels on three of five subsets"},{"dimension":"date","status":"pass","explanation":"The originalPublished value of 2026-10-29 matches the arXiv v1 submission date. It is not the later feed date of 2026-11-02.","excerpt":"[Submitted on 29 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"Each claim matches a verbatim excerpt from the abstract. The limitations correctly note the dataset scope and the three-of-five-subsets result. The question about generalizing beyond the five databases follows from that stated scope rather than implying an answer. The draft adds no unsupported benefits, costs or forecasts.","excerpt":"Our results support recursively maintaining a structured evidence state outside the model's context as an effective design for verifying claims in scientific agents."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct link to the arXiv abstract page, which is the original source.","excerpt":null}],"createdAt":"2026-11-11T02:55:39.483Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}