1 000 travaux · Opus 4.8 · GPT-5.6 (high) · Grok 4.5 · une seule grille verrouillée
La frontière du domaine n'est pas une ligne. C'est une région.
Trois modèles de pointe ont trié les mêmes 1 000 travaux, tirés de la vraie base avec des probabilités de sélection connues, selon la même grille verrouillée sur sa charge utile complète de huit champs. Parmi les 51 travaux qu'au moins un modèle a qualifiés de métarecherche, seulement 19 (37,3 %) l'ont été par les trois, et 24 (47,1 %) reposent sur l'avis d'un seul modèle.
Deux trieurs peuvent s'accorder sur un taux tout en repérant des travaux presque entièrement différents. À un taux de base d'environ 1 %, les rejets évidents offrent 98 % d'accord gratuitement ; c'est pourquoi une statistique d'accord calculée sur tout l'échantillon ne dit rien de la frontière, et pourquoi le livrable est le tableau ci-dessous, et non un pourcentage.
Les trois modèles ne sont pas interchangeables
Combien des 1 000 travaux chaque modèle a qualifiés de métarecherche (niveau T1 ou T2), sur des données identiques. T3 est adjacent et ne compte pas comme dans le champ ; c'est pourquoi le compte d'un modèle ne peut jamais dépasser les 51 travaux du dossier.
Changez le modèle que vous appelez « le trieur » et la taille du domaine bouge. Cet écart, et non l'intervalle de confiance binomial sur les étiquettes d'un seul modèle, est l'incertitude honnête sur la taille de la métarecherche canadienne.
Le dossier des désaccords
Les trois verdicts, confiances et motifs côte à côte. Ce sont les travaux sur lesquels les critères d'inclusion doivent être rédigés, parce que ce sont ceux sur lesquels des trieurs raisonnables, devant la même grille et les mêmes preuves, divergent.
| Travail | Strate | n_in | Opus 4.8 | GPT-5.6 | Grok 4.5 |
|---|---|---|---|---|---|
| International Business Education in the Age of Disruption 2025 · article · en · AIB Insights | fund_new w=1679 | 0/3 | OUTmedium conceptual Pedagogical framework for international business education; teaching practice rather than research practice. | OUThigh conceptual The article proposes a framework for international business education and course design. | OUThigh conceptual Pedagogy framework for international business education addresses teaching, not research methods, evaluation, or scholarly communication. |
| Towards understanding the Milky Way’s typicality: assessing the chemodynamics of M31’s bulge and bar, thick and thin discs 2025 · article · en · Monthly Notices of the Royal Astronomical Society | fund_new w=1679 | 0/3 | OUThigh empirical Chemodynamic modelling of M31's stellar populations; domain astrophysics. | OUThigh empirical This astronomy study models stellar populations in Andromeda, not research. | OUThigh empirical Astrophysics modelling of M31 stellar populations and galactic structure. |
| Canadians are losing their appetite for news — and trusting it less 2023 · preprint · en | aff_core w=5595 | 0/3 | OUThigh other Piece on Canadians consuming and trusting news less; the object is news media consumption and trust in journalism, not research or trust in science. | OUThigh empirical The title concerns public trust and appetite for news, not the research system. | OUTmedium editorial/commentary Title on declining news appetite and trust among Canadians; journalism/public opinion, not research about research. |
| Data hiding and extraction using pseudo-random generation and cover image replication 2025 · article · en · International Journal of Data and Network Science | venue_new w=2684 | 0/3 | OUThigh empirical New steganography algorithm for hiding data in images; computer security. | OUThigh empirical The study proposes and evaluates a steganographic data-hiding algorithm. | OUThigh empirical Steganography algorithm using cover-image replication; security engineering, not research reproducibility. |
| Unravelling symptom-specific polygenic effects on maternal mental health during the perinatal period and postpartum 2025 · article · en · Journal of Affective Disorders | fund_new w=1679 | 0/3 | OUThigh empirical Polygenic score analysis of maternal mental health symptoms; a substantive genetics question. | OUThigh empirical The study investigates genetic associations with maternal mental-health symptoms. | OUThigh empirical Genetic epidemiology of maternal mental health symptoms; clinical/psychiatric genetics object, not metaresearch. |
| Reengineering the Digital Manufacturing Workflow – Application to Wood Buildings Prefabrication 2025 · article · en · IFAC-PapersOnLine | aff_core w=5595 | 0/3 | OUThigh empirical Engineering paper reengineering a CAD-to-CAM digital manufacturing workflow for wood building prefabrication. | OUThigh conceptual It proposes a manufacturing workflow for wood prefabrication, not a method for conducting research. | OUThigh empirical Manufacturing CAD-to-CAM workflow engineering for wood prefabrication, not research-as-object. |
| P073 PREVALENCE AND PREDICTORS OF LP(A) MEASUREMENT IN YOUNGER PATIENTS FOLLOWING ST-ELEVATION MYOCARDIAL INFARCTION 2025 · article · en · Canadian Journal of Cardiology | venue_new w=2684 | 0/3 | OUThigh empirical Abstract on the prevalence and predictors of Lp(a) testing after myocardial infarction; clinical practice patterns, not research practice. | OUThigh empirical It concerns lipoprotein measurement after myocardial infarction, not research itself. | OUTlow empirical Title-only clinical cardiology abstract on Lp(a) measurement after STEMI; domain medicine. |
| Selecting a Standard of Review: What Does This Entail Post-Vavilov? 2025 · article · en · Alberta Law Review | venue_new w=2684 | 0/3 | OUThigh conceptual Administrative law analysis of the judicial standard of review post-Vavilov; 'standard of review' is judicial, not peer review. | OUThigh conceptual The article analyzes administrative-law doctrine, not the research system. | OUThigh conceptual Canadian administrative-law analysis of judicial review standards, not research evaluation or policy of science. |
| Precipitating factors and avoidability of 7-day readmissions after hospitalization for heart failure 2025 · preprint · en · medRxiv | about_only w=3321 | 0/3 | OUThigh empirical Health record review of precipitating factors and avoidability of 7-day heart failure readmissions; a quality-of-care question about clinical services, not about research. | OUThigh empirical This health-record audit studies avoidable heart-failure readmissions, not research methodology or governance. | OUThigh empirical Health-services audit of avoidable heart-failure readmissions; clinical care quality, not research practice. |
| The large-scale kinematics of young stars in the Milky Way disc: first results from SDSS-V 2025 · preprint · en · arXiv (Cornell University) | fund_new w=1679 | 0/3 | OUThigh empirical SDSS-V kinematic mapping of young OB stars in the Milky Way disc; the object is Galactic dynamics. | OUThigh empirical This astronomical survey analyzes stellar kinematics in the Milky Way, not research itself. | OUThigh empirical Galactic astronomy of young-star kinematics from SDSS-V; domain astrophysics. |
| Revealing the untold stories of sinkhole land subsidence over the remains of an underground river in tuff soil by Ground Penetrating Radar 2025 · article · en · GEOMATICA | venue_new w=2684 | 0/3 | OUThigh empirical Geophysical survey using ground penetrating radar to detect sinkhole subsidence. | OUThigh empirical This applies ground-penetrating radar to sinkhole detection, not research practice. | OUThigh empirical Geotechnical GPR study of sinkhole land subsidence; domain engineering. |
| The X-ray statistical properties of dust-obscured galaxies detected by eROSITA 2025 · article · en · Astronomy and Astrophysics | fund_new w=1679 | 0/3 | OUThigh empirical Astronomy study of the X-ray properties of dust-obscured galaxies in eROSITA data. | OUThigh empirical It studies the X-ray properties of dust-obscured galaxies, not research itself. | OUThigh empirical Astronomy study of X-ray properties of dust-obscured galaxies; domain astrophysics. |
| Student perception of gender biases and stereotypes present in their training as radiologists at a Chilean university: initial approach 2025 · article · es · Austral Journal of Imaging (English Edition) | about_only w=3321 | 0/3 | OUTlow empirical Survey of radiology students' perceptions of gender bias in their training at a Chilean university; medical-education equity rather than the research workforce. | OUThigh empirical It studies gender bias in radiology training, not the research workforce or research practice. | OUTmedium empirical Gender bias in clinical radiology training is medical education equity, not research workforce or research practice. |
| Correction: Current application status of non-invasive brain stimulation techniques in Alzheimer's disease: a bibliometric analysis 2025 · article · en · Frontiers in Aging Neuroscience | about_only w=3321 | 0/3 | T3medium other Correction notice attached to a bibliometric analysis of the brain-stimulation literature in Alzheimer disease; corrections are contextual, though the underlying work is a bibliometric study of a research literature. | T3high other This is a correction to a bibliometric article and is contextual rather than part of the analytic corpus. | T3high other Published correction notice; corrections are mapped as T3 even if the corrected paper was bibliometric. |
| From the predator to the prey: a case study of the vulnerability of <i>Harmonia axyridis</i> to aggressive competitors 2025 · article · en · Insect Science | aff_core w=5595 | 0/3 | OUThigh empirical Entomology experiment on intraguild predation and aggressiveness. | OUThigh empirical The study examines predator interactions involving an invasive beetle, not research itself. | OUThigh empirical Experimental ecology of ladybird intraguild predation; domain entomology. |
| Immune and metabolic disturbance as a function of genetic risk and phase of illness in major depression 2025 · article · en · Brain Behavior & Immunity - Health | fund_new w=1679 | 0/3 | OUThigh empirical Biobank study of inflammation and metabolic dysfunction in major depression. | OUThigh empirical The study examines biological and clinical features of depression rather than research itself. | OUThigh empirical Clinical epidemiology of inflammation and metabolism in depression, not research methods. |
| Deep reinforcement learning for joint dispatch of battery storage and gas turbines in renewable-powered microgrids 2025 · article · en · Energy and AI | aff_core w=5595 | 0/3 | OUThigh empirical Deep reinforcement learning for microgrid dispatch; engineering. | OUThigh empirical It develops and evaluates an energy-management algorithm, while code sharing is incidental to the scientific object. | OUThigh empirical Engineering application of deep reinforcement learning to microgrid energy dispatch. |
| ALLIED'S CANADIAN SUBSIDIARY STRIKES DEAL WITH TEAMSTERS. 2003 · article · en · Transport topics | about_only w=3321 | 0/3 | OUThigh other Trade news item on a trucking labour agreement; not research. | OUThigh other The title indicates a business news item about a labor agreement, not research itself. | OUThigh other Industry news about a labour deal; no research object. |
| Jacinthe Baribeau : Là où tu nous mènes 2001 · article · fr · Espace Sculpture | french w=1554 | 0/3 | OUTmedium other French-language art review of a sculptor's work; not research. | OUTlow editorial/commentary The title suggests an art-related work, but the abstract is missing. | OUTlow other Title-only arts item on Jacinthe Baribeau; no abstract and not about research. |
| Marc Lescarbot (v.1565-1642) : Premier historien de la Nouvelle-France 2006 · article · fr · Érudit (Université de Montréal) | french w=1554 | 0/3 | OUTmedium conceptual French-language historical piece on Marc Lescarbot as the first historian of New France; history of a seventeenth-century figure, not contemporary research practice. | OUThigh conceptual The title clearly identifies a historical study of Marc Lescarbot, not research practice. | OUTmedium other Biographical/historical piece on early modern chronicler Lescarbot; not contemporary research practice. |
| Contesting Bodies and Nation in Canadian History ed. by Patrizia Gentile and Jane Nicholas (review) 2014 · article · en · Canadian Historical Review | venue_new w=2684 | 0/3 | OUThigh editorial/commentary Book review of a Canadian history volume; criticism of one book, not a study of research. | OUThigh editorial/commentary This is a review of a history book, not a study of research itself. | OUThigh editorial/commentary Book review of Canadian history; Canadian nation history is not the Canadian research system. |
| Current Trends in Caucasian, East European and Inner Asian Linguistics 2003 · book · en · Amsterdam studies in the theory and history of linguistic science. Series 4, Current issues in linguistic theory | about_only w=3321 | 0/3 | OUThigh other Edited linguistics volume on Caucasian, East European, and Inner Asian languages; the object is language structure and history. | OUThigh other This collected volume concerns historical and synchronic linguistics, not research itself. | OUThigh other Linguistics collection on Caucasian and Inner Asian languages; domain scholarship, not metaresearch. |
| Cellular Automaton Simulation of Vehicle Dynamics in Park-and-Ride Facilities 2007 · article · en · Transportation Research Board 86th Annual MeetingTransportation Research Board | about_only w=3321 | 0/3 | OUThigh empirical Cellular-automaton simulation of vehicle dynamics in park-and-ride lots; a transportation-engineering question. | OUThigh empirical This models vehicle movement in park-and-ride facilities, not research practice. | OUThigh empirical Transportation simulation of park-and-ride vehicle dynamics; traffic engineering, not research. |
| Weed management in white beans with postemergence herbicide tankmixes 2013 · article · en · BioOne Complete (BioOne) | about_only w=3321 | 0/3 | OUThigh empirical Agronomy field trials of postemergence herbicide tankmixes in white beans; the object is weed control. | OUThigh empirical The study evaluates herbicide treatments for white beans rather than research practice. | OUThigh empirical Agronomic field trials of postemergence herbicides in white bean; crop science. |
| Occurrence Download 2020 · dataset · en · Global Biodiversity Information Facility | about_only w=3321 | 0/3 | OUThigh other GBIF occurrence download of bird records for Canada; a biodiversity data export. | OUThigh other The record is a biodiversity occurrence dataset, not a study of research. | OUThigh other Large GBIF bird occurrence download; biodiversity data product, not metaresearch. |
| Recommendations for Public Health Workforce Training and Practice Competencies for Digital Technologies in Public Health: A rapid review and synthesis 2022 · article · en · OSF Preprints (OSF Preprints) | about_only w=3321 | 0/3 | OUTmedium empirical Rapid review of training and practice competencies for digital technologies in the public health workforce; the object is professional practitioner training and curricula rather than the research workforce or research practice. | OUThigh empirical The review addresses public health workforce training competencies rather than evidence synthesis or research practice. | OUTmedium empirical Public health practice workforce digital competencies, not research workforce or metaresearch methods. |
| Image_4_Greenway interventions effectively enhance physical activity levels—A systematic review with meta-analysis.JPEG 2023 · other · en · Figshare | about_only w=3321 | 0/3 | OUThigh other Supplementary figure from a systematic review and meta-analysis of greenway interventions; the parent synthesis answers a public health question. | OUThigh empirical It uses systematic review and meta-analysis to estimate the health effects of greenways. | OUThigh empirical Systematic review/meta-analysis answering whether greenways raise physical activity is domain synthesis. |
| Antibody Characterization Report for Casein kinase II subunit alpha (CSNK2A1) 2024 · article · en · Zenodo (CERN European Organization for Nuclear Research) | aff_core w=5595 | 0/3 | OUTlow empirical Antibody characterization report benchmarking commercial antibodies against CSNK2A1 with knockout-controlled protocols; the closest rubric anchor (Western blot protocol reproducibility) puts reagent-level validation OUT, yet such standardized characterization campaigns are explicitly reagent-reproducibility infrastructure, so this sits on the boundary. | OUThigh empirical The study evaluates antibody performance in laboratory assays rather than research practice itself. | OUTmedium empirical Lab reagent antibody characterization and assay performance; polysemy of validation, not study of research practice. |
| sony f828 manual pdf 2024 · other · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Spam deposit consisting of camera manual download text; no research content. | OUThigh other This record is a camera manual and not a work about research. | OUThigh other Spam-like camera manual download page; not scholarly metaresearch content. |
| Here's Live World Men's U16 Water Polo Championships Malta 2024 Live Streams Free on TV 2024 · dataset · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Spam sports-streaming post deposited as a dataset record; not research. | OUThigh other This record is promotional sports event content and does not study research. | OUThigh other Sports live-stream spam listing, not scholarly metaresearch. |
| Iso 31030 pdf 2024 · other · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Spam PDF-download posting deposited in a repository; not research. | OUThigh other The record is a spam-like document-download entry about a travel-risk standard. | OUThigh other Spam download landing page for an ISO PDF, not a scholarly work about research. |
| PM_116086_B_Sint_Maria_Horebeke 2019 · other · nl · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Deposited heritage photograph of a Belgian parish church interior; not research. | OUThigh other It is a cultural heritage photograph record, not research itself. | OUThigh other Heritage photograph metadata for a Belgian church pulpit; not scholarly research. |
| FIG. 1 in Composition of filamentous fungi associated with bats (Mammalia: Chiroptera) with varied feeding habits in an Atlantic Forest remnant, Brazil 2025 · other · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Zenodo deposit of a single figure (study-area map) from a mycology paper on fungi associated with bats; no metaresearch content. | OUThigh other This is a figure record describing a biological study area and does not study research itself. | OUTmedium other Figure caption on bat-associated filamentous fungi; domain biodiversity content. |
| Hard Kill (2020) streaming vf film complet en ligne 2020 · article · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Spam movie streaming posting deposited in a repository; no research content. | OUThigh other The record is a spam-like online movie listing and does not study research. | OUThigh other Piracy/streaming spam text, not a research work on research. |
| Additional file 6 of Pulmonary and systemic responses to aerosolized lysate of Staphylococcus aureus and Escherichia coli in calves 2020 · article · en · Figshare | aff_core w=5595 | 0/3 | OUThigh other Supplementary data file of mass spectrometry results from a veterinary experiment; domain data. | OUThigh other The record is an additional mass-spectrometry data file from a calf experiment. | OUThigh other Supplementary mass-spectrometry data file for a veterinary immunology experiment; domain research artifact. |
| Additional file 1 of Genome and transcriptome analysis of rock-dissolving Pseudomonas sp. NLX-4 strain 2022 · article · en · Open MIND | aff_core w=5595 | 0/3 | OUThigh other Supplementary file of a bacterial genome and transcriptome paper; a domain data artifact. | OUThigh other This supplementary file documents genomic and transcriptomic experiments on bacteria rather than studying research practice. | OUThigh other Supplementary figures and tables for a Pseudomonas genomics paper; domain biology supporting material. |
| 20 real photographs of Canadian Rocky Mountains 2018 · other · en · cIRcle (University of British Columbia) | about_only w=3321 | 0/3 | OUThigh other A set of 20 photo postcards of the Canadian Rockies; not research. | OUThigh other It is a collection of photographs, not research about research. | OUThigh other Envelope of tourist photo postcards; not scholarly research about research. |
| Maps of North America displaying the geographic location of isolates of (a) <i>B</i>. <i>dermatitidis</i> and (b) <i>B</i>. <i>gilchristii</i> in relation to the major freshwater drainage basins. 2016 · other · en · Figshare | about_only w=3321 | 0/3 | OUThigh other Deposited figure showing fungal isolate locations; a data artifact of a domain study. | OUThigh other The record is a figure showing geographic locations of fungal isolates. | OUThigh other Figure caption mapping fungal isolates; ancillary material, not a study of research. |
| Transitions in the number of physicians in British Columbia, December 1973 - December 1974 2014 · article · en · cIRcle (University of British Columbia) | about_only w=3321 | 0/3 | OUThigh other Statistical tables on the physician workforce in British Columbia; a health workforce count, not the research workforce. | OUThigh empirical This reports changes in the number of physicians in British Columbia, not the research workforce. | OUThigh other Tabular health-workforce statistics on BC physicians; healthcare labour supply, not research workforce. |
| UBC Institute of Fisheries Field Record B.C. 72-37 2013 · other · en · Open Collections | about_only w=3321 | 0/3 | OUThigh other Fisheries field collection record; a primary data artifact. | OUThigh other The record is a fisheries field specimen record. | OUThigh other Fisheries field specimen record; archival data note, not a study of research. |
| Canadian Pacific Railway menu from 1927 2017 · article · en · Open Collections | about_only w=3321 | 0/3 | OUThigh other Archival railway menu from 1927; no research content. | OUThigh other This is an archival railway menu and image record, not a study of research. | OUThigh other Historical Canadian Pacific Railway menu artifact; not research. |
| Analyzing Powers in $\pi^\pm P$ (Polarized) Elastic Scattering From $T (\pi$) = 98-{MeV} to 263-{MeV} 2002 · other · en · HEPData | aff_core w=5595 | 0/3 | OUThigh empirical Particle physics dataset of pion-proton elastic scattering analyzing powers; the object is nuclear physics. | OUThigh other The dataset contains measurements of pion-proton scattering, not a study of research. | OUThigh empirical Particle-physics measurement of pion-proton analyzing powers. |
| Highest level of education by Indigenous identity, Indigenous geography and labour force status: Canada, provinces and territories 2022 · dataset · en · Statistics Canada Dissemination | about_only w=3321 | 0/3 | OUThigh other Statistics Canada data table on educational attainment and labour force status by Indigenous identity; official statistics about the population, not about the research system. | OUThigh other This dataset reports Canadian Indigenous education and labor statistics, not the Canadian research system. | OUThigh other Statistics Canada labour/education data table on Indigenous populations; not a study of research practice. |
| Stihl ms 025 ersatzteilliste pdf 2024 · other · de · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Spam/junk deposit (chainsaw parts list text spliced with multiple sclerosis filler); no research content of any kind. | OUThigh other The record is a spam-like technical download entry and does not study research. | OUThigh other Spam or junk download page about chainsaw parts, not scholarly metaresearch. |
| Keto Breeze ACV Gummies Canada Reviews 2024: (Fake or Legit) What Customers Have To Say? 2024 · other · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh other Spam product-review deposit for keto ACV gummies; not research. | OUThigh other It is promotional supplement content rather than research. | OUThigh other Commercial supplement marketing spam; not scholarly metaresearch. |
| Lasioglossum (Dialictus) coactus 2023 · article · en · Zenodo (CERN European Organization for Nuclear Research) | aff_core w=5595 | 0/3 | OUThigh empirical Taxonomic treatment of a sweat bee species; insect systematics. | OUThigh empirical It is a taxonomic species description, not a study of research itself. | OUThigh other Taxonomic species account for a bee; natural history nomenclature, not metaresearch. |
| Scatopsciara (Scatopsciara) neglecta Menzel & Mohrig 1998 2016 · article · en · Zenodo (CERN European Organization for Nuclear Research) | about_only w=3321 | 0/3 | OUThigh empirical Taxonomic species description of a sciarid fly; the object is insect systematics. | OUThigh empirical The work provides a taxonomic description of a fly species. | OUThigh empirical Taxonomic/species occurrence note for a sciarid fly. |
| Genre, cultures et sexualités 2014 · other · fr · OpenEdition (OpenEdition) | french w=1554 | 0/3 | OUTmedium infrastructure/announcement Announcement of a gender, cultures and sexualities seminar series on the exclusion and invisibilization of women creators; the object is cultural recognition and canon formation, not the research system. | OUThigh infrastructure/announcement This announces a seminar about gender, culture, and sexuality rather than studying research practice. | T3high other Seminar series announcement on gender, cultures, and sexualities; not an empirical or conceptual study of research. |
| SARS‐CoV‐2 Omicron variants in Bangladesh: Pandemic to endemic 2023 · article · en · Greater South Information System | about_only w=3321 | 0/3 | OUThigh empirical Genomic surveillance and clinical comparison of SARS-CoV-2 Omicron sublineages; domain epidemiology. | OUThigh empirical This epidemiological study examines Omicron variants and clinical outcomes in Bangladesh, not research. | OUThigh empirical Clinical and genomic surveillance of Omicron sublineages in Bangladesh. |
| Trend 1993 - 2012. Statistics Canada. CANSIM: Business, Consumer and Property Services - Accommodation and Food | Country: Canada | Table: Sales of alcoholic beverages of liquor authorities, wineries and breweries, by value and volume, fiscal years ended March 31 | Variable: Total beverages, Discounts and rebates included in total Canadian and imported beverages, Value in dollars | Units: , 1993-2012. Data-Planet™ Statistical Ready Reference by Conquest Systems, Inc. Dataset-ID: 075-001-009. 2015 · other · en · Data Planet | about_only w=3321 | 0/3 | OUThigh other Statistics Canada CANSIM data table on alcoholic beverage sales; official economic statistics, not a study of the research system. | OUThigh other The record is a Canadian economic dataset, not a study of the research system. | OUThigh other Statistics Canada alcohol-sales dataset; economic statistics, not the research system. |
Ce que le tri a réellement trouvé
Trois modèles de pointe (Opus 4.8, GPT-5.6 high, Grok 4.5) ont trié les mêmes 1 000 travaux de la VRAIE base de 4,3 M, sur la charge utile COMPLÈTE de huit champs de la grille, avec des lots randomisés consignés au manifeste et des étiquettes écrites par le dispositif : chaque défaut relevé par D1, D2, D11 et le constat 16, réparé. Les taux de base pondérés par le plan de sondage s'étendent de 1,89 % à 3,54 % (1,9 fois). Mais ce sont les ensembles qui font le constat, comme le constat 16 le prédisait : des 51 travaux qu'AU MOINS UN modèle a qualifiés de métarecherche, seulement 19 (37 %) l'ont été par LES TROIS, et 24 (47 %) reposent sur l'avis d'un seul modèle. LA FRONTIÈRE DU DOMAINE N'EST PAS UNE LIGNE QUE LES MODÈLES PARTAGENT ; C'EST UNE RÉGION QUE CHACUN DÉCOUPE À SA FAÇON. GPT-5.6 a de plus enfreint le schéma de sortie verrouillé sur 18 des 1 000 notices, en écrivant des valeurs de genre dans le champ du niveau, ce que le validateur de manifeste a détecté. Le livrable n'est pas le taux de base : c'est le dossier des désaccords, les 51 travaux qui marquent la frontière empirique du domaine et sur lesquels les critères d'inclusion doivent réellement être rédigés.
Calculé le 2026-07-12T19:52:14Z · les 22 constats · ce dossier en JSON