{"id":"f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399","title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","argument":"Authors Carolina Fortuna and Blaz Bertalanic report that, in the tested Fermi estimation tasks, combining answers from multiple language models delivered little improvement because the models shared item-level error: a common error on a particular question. The paper tests the usual averaging argument that independent, unbiased errors shrink as the number of answers grows. Across 2.51M answers from 15 open-weight models, teammates’ errors were uncorrelated given the item, but the item-level component accounted for a median 0.906 of error variance over items. Teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%. The authors report that three rounds of debate reduced within-team dispersion but left the item-level component unchanged; revision without peers achieved the same small gain. Mixing five models of similar accuracy reduced error relative to their homogeneous teams, but no mixture outperformed the most accurate model. The authors say the findings are conditional on the models, benchmarks, prompts and settings tested.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported findings are conditional on the models, benchmarks, prompts and settings tested; the evidence does not establish whether they generalize beyond those conditions.","speaker":{"name":"Fortuna, Carolina, Bertalanic, Blaz","roleAtSource":"Article author"},"source":{"title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.16315","publishedAt":"2026-11-12","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors report that teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%.","excerpt":"Teams of thirty therefore reduce a single agent's error by a median 1.7% (at most 5.4%), whereas draws that were also unbiased on each item would reduce it by 74-79%.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"In the tested answers, the item-level component accounted for a median 0.906 of error variance over items.","excerpt":"the item-level component accounts for a median 0.906 of the error variance over items.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"The authors report that debate reduced within-team dispersion but did not change the item-level component, and that revision without peers achieved the same small gain.","excerpt":"Three rounds of debate reduce within-team dispersion by 93% and leave the item-level component unchanged; revision without peers attains the same small gain.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"The authors report that mixing five models of similar accuracy reduced error relative to their homogeneous teams, but did not outperform the most accurate model.","excerpt":"Mixing five models of similar accuracy reduces error by 6.6% relative to their own homogeneous teams, but no mixture improves on the most accurate model.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"The paper’s findings are conditional on the evaluated models, benchmarks, prompts and settings.","excerpt":"The results are conditional on the models, benchmarks, prompts and settings tested.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.16315","checkedAt":"2026-11-13T11:48:07.263Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Do the reported limits on aggregation persist with models, benchmarks, prompts or settings beyond those tested?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","summary":"Authors Carolina Fortuna and Blaz Bertalanic report that, in the tested Fermi estimation tasks, combining answers from multiple language models delivered little improvement because the models shared item-level error: a common error on a particular question. The paper tests the usual averaging argument that independent, unbiased errors shrink as the number of answers grows. Across 2.51M answers from 15 open-weight models, teammates’ errors were uncorrelated given the item, but the item-level component accounted for a median 0.906 of error variance over items. Teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%. The authors report that three rounds of debate reduced within-team dispersion but left the item-level component unchanged; revision without peers achieved the same small gain. Mixing five models of similar accuracy reduced error relative to their homogeneous teams, but no mixture outperformed the most accurate model. The authors say the findings are conditional on the models, benchmarks, prompts and settings tested.","question":"Do the reported limits on aggregation persist with models, benchmarks, prompts or settings beyond those tested?","claims":[{"text":"The authors report that teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%.","excerpt":"Teams of thirty therefore reduce a single agent's error by a median 1.7% (at most 5.4%), whereas draws that were also unbiased on each item would reduce it by 74-79%.","support":"source-stated"},{"text":"In the tested answers, the item-level component accounted for a median 0.906 of error variance over items.","excerpt":"the item-level component accounts for a median 0.906 of the error variance over items.","support":"source-stated"},{"text":"The authors report that debate reduced within-team dispersion but did not change the item-level component, and that revision without peers achieved the same small gain.","excerpt":"Three rounds of debate reduce within-team dispersion by 93% and leave the item-level component unchanged; revision without peers attains the same small gain.","support":"source-stated"},{"text":"The authors report that mixing five models of similar accuracy reduced error relative to their homogeneous teams, but did not outperform the most accurate model.","excerpt":"Mixing five models of similar accuracy reduces error by 6.6% relative to their own homogeneous teams, but no mixture improves on the most accurate model.","support":"source-stated"},{"text":"The paper’s findings are conditional on the evaluated models, benchmarks, prompts and settings.","excerpt":"The results are conditional on the models, benchmarks, prompts and settings tested.","support":"source-stated"}],"limitations":["The reported findings are conditional on the models, benchmarks, prompts and settings tested; the evidence does not establish whether they generalize beyond those conditions."],"mediaId":"230e075913833340416682f4867155beaeb932dc78f75ec907c02e5c7fdb715b","correctionReason":null,"id":"f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399","publicationId":"0a1560c8-e133-4d95-80f0-68198afd0691","version":1,"sourceUrl":"https://arxiv.org/abs/2611.16315","sourceTitle":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-12","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-13T07:39:33.241Z","createdAt":"2026-11-13T11:32:02.541Z","historical":false,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"230e075913833340416682f4867155beaeb932dc78f75ec907c02e5c7fdb715b","kind":"article","url":"https://arxiv.org/abs/2611.16315","sourceUrl":"https://arxiv.org/abs/2611.16315","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-13T11:17:41.531Z","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/2611.16315","checkedAt":"2026-11-13T11:48:07.263Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v7","mediaId":"230e075913833340416682f4867155beaeb932dc78f75ec907c02e5c7fdb715b","confidence":0.68,"provider":"typesafe","model":"jev-1.13.0","inputHash":"68061d886c946962a94734fe2d76b97ca7035f729a6166c86f66e1bb52bbdca4","decidedAt":"2026-11-13T11:17:42.842Z"},"provenance":{"textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observationId":"106e022c543d79cebed3b6349890288e1456987702dac9ab025207b5a4e2e066","retrievedAt":"2026-11-13T11:17:41.478Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"b6183082-4973-42ef-8b16-b7d29f62f71c"},"reviewer":{"model":"gpt-6-luna","runId":"abfb01e5-d74a-4888-8d9e-3c1d4d7dd168"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-13T11:32:02.541Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Fortuna, Carolina","url":null,"kind":"unknown"},{"name":"Bertalanic, Blaz","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Fortuna, Carolina","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Fortuna, Carolina","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.16315","locator":"head/meta[name=author|citation_author]"},"assertionId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T11:48:07.263Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d","sourceUrl":"https://arxiv.org/abs/2611.16315","contentSha256":"31a7433689fd432835335a0643f2c33924f51932e9cbb2e41557d6d74843327d","textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observedAt":"2026-11-13T11:48:07.263Z"}},{"actor":{"name":"Bertalanic, Blaz","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Bertalanic, Blaz","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.16315","locator":"head/meta[name=author|citation_author]"},"assertionId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T11:48:07.263Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d","sourceUrl":"https://arxiv.org/abs/2611.16315","contentSha256":"31a7433689fd432835335a0643f2c33924f51932e9cbb2e41557d6d74843327d","textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observedAt":"2026-11-13T11:48:07.263Z"}}],"participationCoverage":{"total":2,"returned":2,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.16315","snapshotId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d","contentSha256":"31a7433689fd432835335a0643f2c33924f51932e9cbb2e41557d6d74843327d","textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observedAt":"2026-11-13T11:48:07.263Z","basis":"publisher-page-metadata","sourceLinks":41,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399/source"}},"canonicalUrl":"https://escapemeridian.com/readings/shared-item-level-error-limits-the-aggregation-of-small-f777f8a4","formats":{"html":"/readings/shared-item-level-error-limits-the-aggregation-of-small-f777f8a4","markdown":"/readings/shared-item-level-error-limits-the-aggregation-of-small-f777f8a4.md","json":"/readings/shared-item-level-error-limits-the-aggregation-of-small-f777f8a4.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":[{"id":"term-open-weights","name":"Open weights","explanation":"A model whose trained numbers are published so others can download, run and inspect it; the license and training data may still be restricted.","referenceUrl":"https://opensource.org/ai/open-weights"},{"id":"term-benchmark","name":"Benchmark","explanation":"A defined set of tests used to compare systems. Results depend on the tasks and measures selected.","referenceUrl":"https://arxiv.org/abs/2211.09110"}],"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"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399","revision":"0a1560c8-e133-4d95-80f0-68198afd0691","revisionKind":"publication-id","title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","url":"https://escapemeridian.com/readings/shared-item-level-error-limits-the-aggregation-of-small-f777f8a4"},"source":{"type":"source","id":"https://arxiv.org/abs/2611.16315","revision":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","revisionKind":"sha256-source-text","title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","url":"https://arxiv.org/abs/2611.16315"},"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":"f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399","title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","argument":"Authors Carolina Fortuna and Blaz Bertalanic report that, in the tested Fermi estimation tasks, combining answers from multiple language models delivered little improvement because the models shared item-level error: a common error on a particular question. The paper tests the usual averaging argument that independent, unbiased errors shrink as the number of answers grows. Across 2.51M answers from 15 open-weight models, teammates’ errors were uncorrelated given the item, but the item-level component accounted for a median 0.906 of error variance over items. Teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%. The authors report that three rounds of debate reduced within-team dispersion but left the item-level component unchanged; revision without peers achieved the same small gain. Mixing five models of similar accuracy reduced error relative to their homogeneous teams, but no mixture outperformed the most accurate model. The authors say the findings are conditional on the models, benchmarks, prompts and settings tested.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The reported findings are conditional on the models, benchmarks, prompts and settings tested; the evidence does not establish whether they generalize beyond those conditions.","speaker":{"name":"Fortuna, Carolina, Bertalanic, Blaz","roleAtSource":"Article author"},"source":{"title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2611.16315","publishedAt":"2026-11-12","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors report that teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%.","excerpt":"Teams of thirty therefore reduce a single agent's error by a median 1.7% (at most 5.4%), whereas draws that were also unbiased on each item would reduce it by 74-79%.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"In the tested answers, the item-level component accounted for a median 0.906 of error variance over items.","excerpt":"the item-level component accounts for a median 0.906 of the error variance over items.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"The authors report that debate reduced within-team dispersion but did not change the item-level component, and that revision without peers achieved the same small gain.","excerpt":"Three rounds of debate reduce within-team dispersion by 93% and leave the item-level component unchanged; revision without peers attains the same small gain.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"The authors report that mixing five models of similar accuracy reduced error relative to their homogeneous teams, but did not outperform the most accurate model.","excerpt":"Mixing five models of similar accuracy reduces error by 6.6% relative to their own homogeneous teams, but no mixture improves on the most accurate model.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"},{"text":"The paper’s findings are conditional on the evaluated models, benchmarks, prompts and settings.","excerpt":"The results are conditional on the models, benchmarks, prompts and settings tested.","url":"https://arxiv.org/abs/2611.16315","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2611.16315","checkedAt":"2026-11-13T11:48:07.263Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Do the reported limits on aggregation persist with models, benchmarks, prompts or settings beyond those tested?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","summary":"Authors Carolina Fortuna and Blaz Bertalanic report that, in the tested Fermi estimation tasks, combining answers from multiple language models delivered little improvement because the models shared item-level error: a common error on a particular question. The paper tests the usual averaging argument that independent, unbiased errors shrink as the number of answers grows. Across 2.51M answers from 15 open-weight models, teammates’ errors were uncorrelated given the item, but the item-level component accounted for a median 0.906 of error variance over items. Teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%. The authors report that three rounds of debate reduced within-team dispersion but left the item-level component unchanged; revision without peers achieved the same small gain. Mixing five models of similar accuracy reduced error relative to their homogeneous teams, but no mixture outperformed the most accurate model. The authors say the findings are conditional on the models, benchmarks, prompts and settings tested.","question":"Do the reported limits on aggregation persist with models, benchmarks, prompts or settings beyond those tested?","claims":[{"text":"The authors report that teams of thirty reduced a single agent’s error by a median 1.7%, while draws unbiased on each item would have reduced it by 74–79%.","excerpt":"Teams of thirty therefore reduce a single agent's error by a median 1.7% (at most 5.4%), whereas draws that were also unbiased on each item would reduce it by 74-79%.","support":"source-stated"},{"text":"In the tested answers, the item-level component accounted for a median 0.906 of error variance over items.","excerpt":"the item-level component accounts for a median 0.906 of the error variance over items.","support":"source-stated"},{"text":"The authors report that debate reduced within-team dispersion but did not change the item-level component, and that revision without peers achieved the same small gain.","excerpt":"Three rounds of debate reduce within-team dispersion by 93% and leave the item-level component unchanged; revision without peers attains the same small gain.","support":"source-stated"},{"text":"The authors report that mixing five models of similar accuracy reduced error relative to their homogeneous teams, but did not outperform the most accurate model.","excerpt":"Mixing five models of similar accuracy reduces error by 6.6% relative to their own homogeneous teams, but no mixture improves on the most accurate model.","support":"source-stated"},{"text":"The paper’s findings are conditional on the evaluated models, benchmarks, prompts and settings.","excerpt":"The results are conditional on the models, benchmarks, prompts and settings tested.","support":"source-stated"}],"limitations":["The reported findings are conditional on the models, benchmarks, prompts and settings tested; the evidence does not establish whether they generalize beyond those conditions."],"mediaId":"230e075913833340416682f4867155beaeb932dc78f75ec907c02e5c7fdb715b","correctionReason":null,"id":"f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399","publicationId":"0a1560c8-e133-4d95-80f0-68198afd0691","version":1,"sourceUrl":"https://arxiv.org/abs/2611.16315","sourceTitle":"Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-11-12","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-13T07:39:33.241Z","createdAt":"2026-11-13T11:32:02.541Z","historical":false,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"230e075913833340416682f4867155beaeb932dc78f75ec907c02e5c7fdb715b","kind":"article","url":"https://arxiv.org/abs/2611.16315","sourceUrl":"https://arxiv.org/abs/2611.16315","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-13T11:17:41.531Z","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/2611.16315","checkedAt":"2026-11-13T11:48:07.263Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v7","mediaId":"230e075913833340416682f4867155beaeb932dc78f75ec907c02e5c7fdb715b","confidence":0.68,"provider":"typesafe","model":"jev-1.13.0","inputHash":"68061d886c946962a94734fe2d76b97ca7035f729a6166c86f66e1bb52bbdca4","decidedAt":"2026-11-13T11:17:42.842Z"},"provenance":{"textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observationId":"106e022c543d79cebed3b6349890288e1456987702dac9ab025207b5a4e2e066","retrievedAt":"2026-11-13T11:17:41.478Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"b6183082-4973-42ef-8b16-b7d29f62f71c"},"reviewer":{"model":"gpt-6-luna","runId":"abfb01e5-d74a-4888-8d9e-3c1d4d7dd168"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-13T11:32:02.541Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Fortuna, Carolina","url":null,"kind":"unknown"},{"name":"Bertalanic, Blaz","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Fortuna, Carolina","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Fortuna, Carolina","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.16315","locator":"head/meta[name=author|citation_author]"},"assertionId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T11:48:07.263Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d","sourceUrl":"https://arxiv.org/abs/2611.16315","contentSha256":"31a7433689fd432835335a0643f2c33924f51932e9cbb2e41557d6d74843327d","textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observedAt":"2026-11-13T11:48:07.263Z"}},{"actor":{"name":"Bertalanic, Blaz","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Bertalanic, Blaz","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2611.16315","locator":"head/meta[name=author|citation_author]"},"assertionId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-13T11:48:07.263Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d","sourceUrl":"https://arxiv.org/abs/2611.16315","contentSha256":"31a7433689fd432835335a0643f2c33924f51932e9cbb2e41557d6d74843327d","textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observedAt":"2026-11-13T11:48:07.263Z"}}],"participationCoverage":{"total":2,"returned":2,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2611.16315","snapshotId":"e1ac7c045fe5b5966baae832275b58b1b3a1c18adda111fc3b125008f77d369d","contentSha256":"31a7433689fd432835335a0643f2c33924f51932e9cbb2e41557d6d74843327d","textHash":"e68fb33600a9d002c3d227bc98cb2f9b872f586da90e5f17a3bc20f9987a5630","observedAt":"2026-11-13T11:48:07.263Z","basis":"publisher-page-metadata","sourceLinks":41,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/f777f8a4d80e8e86e0d5f3e2d60e97583462c6be655d4fc0bb0b43452c078399/source"}}}],"audits":[{"publicationId":"0a1560c8-e133-4d95-80f0-68198afd0691","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading correctly credits the findings to Carolina Fortuna and Blaz Bertalanic and keeps results framed as what the authors report. The work is identified as an arXiv paper.","excerpt":"Authors:Carolina Fortuna, Blaz Bertalanic"},{"dimension":"summary","status":"uncertain","explanation":"Mostly faithful, but the central term Fermi estimation is never explained (rough order-of-magnitude guesses at quantities). Add a term-library tooltip or a short gloss. Open-weight also needs a tooltip. The lead, combining answers from multiple language models, blurs the design: teams were mainly repeated draws from one model, and the shared error is each model's mean error on an item. Smallest repair: say teams of answers from the same model, with mixed-model teams tested separately. The summary could also note that multi-agent systems are often justified by this averaging argument; that is what makes the result matter.","excerpt":"We decompose the log10 error of each answer into the mean error of its model on the item, common to all its draws on the item, and a residual"},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-11-12 matches the v1 submission date. The feed date of 2026-11-13 is correctly not used.","excerpt":"[Submitted on 12 Nov 2026]"},{"dimension":"support","status":"pass","explanation":"All figures (2.51M answers, 15 models, 0.906, 1.7%, 74-79%, debate, mixing) and the conditional limitation match the abstract verbatim. The question follows from the authors' own stated scope limit. No unsupported benefits or costs are implied.","excerpt":"The results are conditional on the models, benchmarks, prompts and settings tested."},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source.","excerpt":null}],"createdAt":"2026-11-13T11:32:02.541Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}