{"id":"41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a","title":"CALM authors propose separating memory preservation from what an agent may use","argument":"The authors of CALM argue that a memory can be outdated for present-day reasoning yet still matter for reconstructing the past. They say relevance-based retrieval misses the distinction between semantic relevance, evidence-grounded validity, and query-specific eligibility. CALM is their proposed framework for managing memory over time: it preserves records while separately deciding which evidence to admit for a particular query. It converts records into evidence-grounded memory cells, updates their validity and lifecycle states using support, conflict, and dependency signals, then uses training-free or learned gating informed by temporal applicability. Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, the authors report that CALM outperforms state-of-the-art baselines in overall accuracy. They also report that query-conditioned admission improves evidence selection compared with simple filtering and reranking, and that preserving superseded records supports accurate reasoning about earlier states without compromising current-state answers. The design separates keeping information from deciding whether it applies to a query: retaining old records need not mean treating them as applicable to every query. The abstract reports these results but gives no numerical scores or detailed account of the experiments.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper’s abstract; it gives no numerical accuracy scores or detailed experimental methods. The reported results are attributed to the paper’s authors and are not independently assessed by the supplied evidence.","speaker":{"name":"Li, Chang, Ou, Xianhao, Ma, Xinyan, Li, Hao, Xu, Haoyuan, Liang, Jiafeng, Wang, Xiaohan, Chai, Jiajun, Yin, Guojun, Liu, Ming","roleAtSource":"Article author"},"source":{"title":"CALM: Conflict-Aware Lifecycle Memory Governance for Long-Horizon Language Agents","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.38573","publishedAt":"2026-10-28","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors argue that relevance-based retrieval misses the distinction between relevance, validity, and whether evidence is eligible for a particular query.","excerpt":"semantic relevance, evidence-grounded validity, and query-specific eligibility are distinct.","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"},{"text":"CALM preserves memory separately from its query-time admission, and updates memory cells using support, conflict, and dependency signals.","excerpt":"decouples persistent memory preservation from query-time admission. CALM converts heterogeneous records into evidence-grounded memory cells","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"},{"text":"The authors report that CALM achieves the best overall accuracy among state-of-the-art baselines in tests using LongMemEval-S and LoCoMo with Qwen and GPT backbones.","excerpt":"Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, CALM outperforms state-of-the-art baselines, achieving the best overall accuracy.","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"},{"text":"The authors report that retaining superseded records supports accurate reasoning about earlier states without reducing accuracy on current-state answers.","excerpt":"preserving superseded records supports accurate earlier-state reasoning without compromising current-state answers.","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38573","checkedAt":"2026-11-11T10:42:08.949Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What numerical accuracy results and experimental details support the authors’ reported comparisons across LongMemEval-S and LoCoMo?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"CALM authors propose separating memory preservation from what an agent may use","summary":"The authors of CALM argue that a memory can be outdated for present-day reasoning yet still matter for reconstructing the past. They say relevance-based retrieval misses the distinction between semantic relevance, evidence-grounded validity, and query-specific eligibility. CALM is their proposed framework for managing memory over time: it preserves records while separately deciding which evidence to admit for a particular query. It converts records into evidence-grounded memory cells, updates their validity and lifecycle states using support, conflict, and dependency signals, then uses training-free or learned gating informed by temporal applicability. Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, the authors report that CALM outperforms state-of-the-art baselines in overall accuracy. They also report that query-conditioned admission improves evidence selection compared with simple filtering and reranking, and that preserving superseded records supports accurate reasoning about earlier states without compromising current-state answers. The design separates keeping information from deciding whether it applies to a query: retaining old records need not mean treating them as applicable to every query. The abstract reports these results but gives no numerical scores or detailed account of the experiments.","question":"What numerical accuracy results and experimental details support the authors’ reported comparisons across LongMemEval-S and LoCoMo?","claims":[{"text":"The authors argue that relevance-based retrieval misses the distinction between relevance, validity, and whether evidence is eligible for a particular query.","excerpt":"semantic relevance, evidence-grounded validity, and query-specific eligibility are distinct.","support":"source-stated"},{"text":"CALM preserves memory separately from its query-time admission, and updates memory cells using support, conflict, and dependency signals.","excerpt":"decouples persistent memory preservation from query-time admission. CALM converts heterogeneous records into evidence-grounded memory cells","support":"source-stated"},{"text":"The authors report that CALM achieves the best overall accuracy among state-of-the-art baselines in tests using LongMemEval-S and LoCoMo with Qwen and GPT backbones.","excerpt":"Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, CALM outperforms state-of-the-art baselines, achieving the best overall accuracy.","support":"source-stated"},{"text":"The authors report that retaining superseded records supports accurate reasoning about earlier states without reducing accuracy on current-state answers.","excerpt":"preserving superseded records supports accurate earlier-state reasoning without compromising current-state answers.","support":"source-stated"}],"limitations":["The supplied evidence is the paper’s abstract; it gives no numerical accuracy scores or detailed experimental methods.","The reported results are attributed to the paper’s authors and are not independently assessed by the supplied evidence."],"mediaId":"0fccc4430f6f947bc9b7ab61310fd80b3cf2d2be9ad6c232ebb982260f8e608f","correctionReason":null,"id":"41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a","publicationId":"c049c6fc-7105-45e6-8d0b-5004b91b5b5c","version":1,"sourceUrl":"https://arxiv.org/abs/2610.38573","sourceTitle":"CALM: Conflict-Aware Lifecycle Memory Governance for Long-Horizon Language Agents","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-28","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-30T06:42:50.436Z","createdAt":"2026-11-11T09:39:22.039Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"0fccc4430f6f947bc9b7ab61310fd80b3cf2d2be9ad6c232ebb982260f8e608f","kind":"article","url":"https://arxiv.org/abs/2610.38573","sourceUrl":"https://arxiv.org/abs/2610.38573","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-31T05:47:43.640Z","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/2610.38573","checkedAt":"2026-11-11T10:42:08.949Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"0fccc4430f6f947bc9b7ab61310fd80b3cf2d2be9ad6c232ebb982260f8e608f","confidence":0.66,"provider":"typesafe","model":"jev-1.13.0","inputHash":"6fe385ce30bb362bf44924f0388e04623d7ed43e1bb9fe81b53d2b4df83eb8b3","decidedAt":"2026-10-31T05:47:44.949Z","redecided":{"from":"stream-admission-v5","under":"stream-admission-v5+confidence-0.5"}},"provenance":{"textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observationId":"68c168ade1a1189aae23693157b0f07d631072c3e8596c36ac16133540cf3f32","retrievedAt":"2026-10-31T05:47:43.481Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"92e9d135-2d70-45b2-8151-4d1b13b4fca7"},"reviewer":{"model":"gpt-6-luna","runId":"bf7cf004-352d-4ed0-887a-2f78deaf0d16"}},"audit":{"status":"passed","reviewedAt":"2026-11-11T09:39:22.039Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Li, Chang","url":null,"kind":"unknown"},{"name":"Ou, Xianhao","url":null,"kind":"unknown"},{"name":"Ma, Xinyan","url":null,"kind":"unknown"},{"name":"Li, Hao","url":null,"kind":"unknown"},{"name":"Xu, Haoyuan","url":null,"kind":"unknown"},{"name":"Liang, Jiafeng","url":null,"kind":"unknown"},{"name":"Wang, Xiaohan","url":null,"kind":"unknown"},{"name":"Chai, Jiajun","url":null,"kind":"unknown"},{"name":"Yin, Guojun","url":null,"kind":"unknown"},{"name":"Liu, Ming","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Li, Chang","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Chang","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Ou, Xianhao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ou, Xianhao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Ma, Xinyan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ma, Xinyan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Li, Hao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Hao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Xu, Haoyuan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Xu, Haoyuan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Liang, Jiafeng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liang, Jiafeng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Wang, Xiaohan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wang, Xiaohan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Chai, Jiajun","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chai, Jiajun","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Yin, Guojun","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yin, Guojun","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Liu, Ming","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liu, Ming","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:9","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}}],"participationCoverage":{"total":10,"returned":10,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.38573","snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z","basis":"publisher-page-metadata","sourceLinks":48,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a/source"}},"canonicalUrl":"https://escapemeridian.com/readings/calm-authors-propose-separating-memory-preservation-from-what-41624336","formats":{"html":"/readings/calm-authors-propose-separating-memory-preservation-from-what-41624336","markdown":"/readings/calm-authors-propose-separating-memory-preservation-from-what-41624336.md","json":"/readings/calm-authors-propose-separating-memory-preservation-from-what-41624336.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":"time","name":"TIME","kind":"organization","context":"TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.","referenceUrls":["https://time.com/"],"path":"/entities/time","json":"/entities/time.json","markdown":"/entities/time.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a","revision":"c049c6fc-7105-45e6-8d0b-5004b91b5b5c","revisionKind":"publication-id","title":"CALM authors propose separating memory preservation from what an agent may use","url":"https://escapemeridian.com/readings/calm-authors-propose-separating-memory-preservation-from-what-41624336"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.38573","revision":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","revisionKind":"sha256-source-text","title":"CALM: Conflict-Aware Lifecycle Memory Governance for Long-Horizon Language Agents","url":"https://arxiv.org/abs/2610.38573"},"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":"41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a","title":"CALM authors propose separating memory preservation from what an agent may use","argument":"The authors of CALM argue that a memory can be outdated for present-day reasoning yet still matter for reconstructing the past. They say relevance-based retrieval misses the distinction between semantic relevance, evidence-grounded validity, and query-specific eligibility. CALM is their proposed framework for managing memory over time: it preserves records while separately deciding which evidence to admit for a particular query. It converts records into evidence-grounded memory cells, updates their validity and lifecycle states using support, conflict, and dependency signals, then uses training-free or learned gating informed by temporal applicability. Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, the authors report that CALM outperforms state-of-the-art baselines in overall accuracy. They also report that query-conditioned admission improves evidence selection compared with simple filtering and reranking, and that preserving superseded records supports accurate reasoning about earlier states without compromising current-state answers. The design separates keeping information from deciding whether it applies to a query: retaining old records need not mean treating them as applicable to every query. The abstract reports these results but gives no numerical scores or detailed account of the experiments.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper’s abstract; it gives no numerical accuracy scores or detailed experimental methods. The reported results are attributed to the paper’s authors and are not independently assessed by the supplied evidence.","speaker":{"name":"Li, Chang, Ou, Xianhao, Ma, Xinyan, Li, Hao, Xu, Haoyuan, Liang, Jiafeng, Wang, Xiaohan, Chai, Jiajun, Yin, Guojun, Liu, Ming","roleAtSource":"Article author"},"source":{"title":"CALM: Conflict-Aware Lifecycle Memory Governance for Long-Horizon Language Agents","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.38573","publishedAt":"2026-10-28","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors argue that relevance-based retrieval misses the distinction between relevance, validity, and whether evidence is eligible for a particular query.","excerpt":"semantic relevance, evidence-grounded validity, and query-specific eligibility are distinct.","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"},{"text":"CALM preserves memory separately from its query-time admission, and updates memory cells using support, conflict, and dependency signals.","excerpt":"decouples persistent memory preservation from query-time admission. CALM converts heterogeneous records into evidence-grounded memory cells","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"},{"text":"The authors report that CALM achieves the best overall accuracy among state-of-the-art baselines in tests using LongMemEval-S and LoCoMo with Qwen and GPT backbones.","excerpt":"Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, CALM outperforms state-of-the-art baselines, achieving the best overall accuracy.","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"},{"text":"The authors report that retaining superseded records supports accurate reasoning about earlier states without reducing accuracy on current-state answers.","excerpt":"preserving superseded records supports accurate earlier-state reasoning without compromising current-state answers.","url":"https://arxiv.org/abs/2610.38573","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.38573","checkedAt":"2026-11-11T10:42:08.949Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What numerical accuracy results and experimental details support the authors’ reported comparisons across LongMemEval-S and LoCoMo?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"CALM authors propose separating memory preservation from what an agent may use","summary":"The authors of CALM argue that a memory can be outdated for present-day reasoning yet still matter for reconstructing the past. They say relevance-based retrieval misses the distinction between semantic relevance, evidence-grounded validity, and query-specific eligibility. CALM is their proposed framework for managing memory over time: it preserves records while separately deciding which evidence to admit for a particular query. It converts records into evidence-grounded memory cells, updates their validity and lifecycle states using support, conflict, and dependency signals, then uses training-free or learned gating informed by temporal applicability. Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, the authors report that CALM outperforms state-of-the-art baselines in overall accuracy. They also report that query-conditioned admission improves evidence selection compared with simple filtering and reranking, and that preserving superseded records supports accurate reasoning about earlier states without compromising current-state answers. The design separates keeping information from deciding whether it applies to a query: retaining old records need not mean treating them as applicable to every query. The abstract reports these results but gives no numerical scores or detailed account of the experiments.","question":"What numerical accuracy results and experimental details support the authors’ reported comparisons across LongMemEval-S and LoCoMo?","claims":[{"text":"The authors argue that relevance-based retrieval misses the distinction between relevance, validity, and whether evidence is eligible for a particular query.","excerpt":"semantic relevance, evidence-grounded validity, and query-specific eligibility are distinct.","support":"source-stated"},{"text":"CALM preserves memory separately from its query-time admission, and updates memory cells using support, conflict, and dependency signals.","excerpt":"decouples persistent memory preservation from query-time admission. CALM converts heterogeneous records into evidence-grounded memory cells","support":"source-stated"},{"text":"The authors report that CALM achieves the best overall accuracy among state-of-the-art baselines in tests using LongMemEval-S and LoCoMo with Qwen and GPT backbones.","excerpt":"Across LongMemEval-S and LoCoMo with Qwen and GPT backbones, CALM outperforms state-of-the-art baselines, achieving the best overall accuracy.","support":"source-stated"},{"text":"The authors report that retaining superseded records supports accurate reasoning about earlier states without reducing accuracy on current-state answers.","excerpt":"preserving superseded records supports accurate earlier-state reasoning without compromising current-state answers.","support":"source-stated"}],"limitations":["The supplied evidence is the paper’s abstract; it gives no numerical accuracy scores or detailed experimental methods.","The reported results are attributed to the paper’s authors and are not independently assessed by the supplied evidence."],"mediaId":"0fccc4430f6f947bc9b7ab61310fd80b3cf2d2be9ad6c232ebb982260f8e608f","correctionReason":null,"id":"41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a","publicationId":"c049c6fc-7105-45e6-8d0b-5004b91b5b5c","version":1,"sourceUrl":"https://arxiv.org/abs/2610.38573","sourceTitle":"CALM: Conflict-Aware Lifecycle Memory Governance for Long-Horizon Language Agents","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-28","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-30T06:42:50.436Z","createdAt":"2026-11-11T09:39:22.039Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"0fccc4430f6f947bc9b7ab61310fd80b3cf2d2be9ad6c232ebb982260f8e608f","kind":"article","url":"https://arxiv.org/abs/2610.38573","sourceUrl":"https://arxiv.org/abs/2610.38573","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-31T05:47:43.640Z","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/2610.38573","checkedAt":"2026-11-11T10:42:08.949Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"0fccc4430f6f947bc9b7ab61310fd80b3cf2d2be9ad6c232ebb982260f8e608f","confidence":0.66,"provider":"typesafe","model":"jev-1.13.0","inputHash":"6fe385ce30bb362bf44924f0388e04623d7ed43e1bb9fe81b53d2b4df83eb8b3","decidedAt":"2026-10-31T05:47:44.949Z","redecided":{"from":"stream-admission-v5","under":"stream-admission-v5+confidence-0.5"}},"provenance":{"textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observationId":"68c168ade1a1189aae23693157b0f07d631072c3e8596c36ac16133540cf3f32","retrievedAt":"2026-10-31T05:47:43.481Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"92e9d135-2d70-45b2-8151-4d1b13b4fca7"},"reviewer":{"model":"gpt-6-luna","runId":"bf7cf004-352d-4ed0-887a-2f78deaf0d16"}},"audit":{"status":"passed","reviewedAt":"2026-11-11T09:39:22.039Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Li, Chang","url":null,"kind":"unknown"},{"name":"Ou, Xianhao","url":null,"kind":"unknown"},{"name":"Ma, Xinyan","url":null,"kind":"unknown"},{"name":"Li, Hao","url":null,"kind":"unknown"},{"name":"Xu, Haoyuan","url":null,"kind":"unknown"},{"name":"Liang, Jiafeng","url":null,"kind":"unknown"},{"name":"Wang, Xiaohan","url":null,"kind":"unknown"},{"name":"Chai, Jiajun","url":null,"kind":"unknown"},{"name":"Yin, Guojun","url":null,"kind":"unknown"},{"name":"Liu, Ming","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Li, Chang","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Chang","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Ou, Xianhao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ou, Xianhao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Ma, Xinyan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ma, Xinyan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Li, Hao","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Hao","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Xu, Haoyuan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Xu, Haoyuan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Liang, Jiafeng","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liang, Jiafeng","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:5","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Wang, Xiaohan","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Wang, Xiaohan","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:6","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Chai, Jiajun","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Chai, Jiajun","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:7","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Yin, Guojun","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Yin, Guojun","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:8","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}},{"actor":{"name":"Liu, Ming","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Liu, Ming","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.38573","locator":"head/meta[name=author|citation_author]"},"assertionId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73:participation:9","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-11T10:42:08.949Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","sourceUrl":"https://arxiv.org/abs/2610.38573","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z"}}],"participationCoverage":{"total":10,"returned":10,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.38573","snapshotId":"49853f2b00a40454158988805a2ad111f8af1fa5f37eb98fbe4467cb2d15ba73","contentSha256":"c91afb61b1256681e1a012f8b9583756e65ed017fe1b07dccf1f96b5682bcb87","textHash":"554e1f579dbb93333e7413a2f5a4062614fff0cecaa6563e5d5ac84420179e4d","observedAt":"2026-11-11T10:42:08.949Z","basis":"publisher-page-metadata","sourceLinks":48,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/41624336630e52a602f8567ca40794ff6167d634fdf2f7d3b496302b2096044a/source"}}}],"audits":[{"publicationId":"c049c6fc-7105-45e6-8d0b-5004b91b5b5c","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The reading attributes the argument, framework and results to the paper's authors (Chang Li et al.) on arXiv. Results are framed as reported by the authors, not as established facts.","excerpt":"Authors:Chang Li, Xianhao Ou, Xinyan Ma, Hao Li, Haoyuan Xu, Jiafeng Liang, Xiaohan Wang, Jiajun Chai, Guojun Yin, Ming Liu"},{"dimension":"summary","status":"pass","explanation":"The reading leads with the authors' central tension and accurately restates the mechanism and reported findings using the exact technical terms. The plain-language sentence about keeping information versus deciding whether it applies makes the core idea accessible. Optional tooltip additions: LongMemEval-S and LoCoMo are evaluation benchmarks, Qwen and GPT backbones are the underlying language models, and gating means deciding what evidence is admitted. The question raises a real gap without leading the reader.","excerpt":"A memory can be outdated for present-day reasoning yet indispensable for reconstructing the past."},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-10-28 matches the v1 arXiv submission date. The feed date of Oct 31 is a later announcement and is not used.","excerpt":"[Submitted on 28 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"Each claim matches the abstract. The draft's 'without reducing accuracy' paraphrases 'without compromising' faithfully. The stated limitations correctly note the absence of numerical scores. No benefits, costs or forecasts are invented, and the omitted conflict-blocking and robustness claim is not a distortion.","excerpt":"Compared with simple filtering and reranking, query-conditioned admission further improves evidence selection."},{"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":"2026-11-11T09:39:22.039Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}