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 |
|---|---|---|---|---|---|
| 10.1037/e527112010-001 2000 · article · en · Time to knit | venue_new w=2684 | 0/3 | OUTlow other Placeholder DOI record with no attached metadata content. | OUThigh other It is a DOI metadata notice rather than a study of research. | OUThigh other Crossref error/orphan DOI notice; not a research work about research. |
| SALEMA (Maria-José), KAHN (Gisèle), TEIXEIRA (Luis Filipe) (dir.) – Documents pour l’histoire du français langue étrangère ou seconde. L’enseignement de la langue et de la littérature françaises dans la seconde moitié du xixe siècle 2004 · article · fr · Histoire de l éducation | french w=1554 | 0/3 | OUThigh editorial/commentary Book review of an edited volume on the history of teaching French as a foreign language; the object is language teaching history. | OUTlow other The work concerns the history of French-language education, not research itself. | OUTlow editorial/commentary Book notice on 19th-century French language teaching history; empty abstract and no object of research practice. |
| Parutions récentes 2016 · article · fr · Bulletin d histoire politique | french w=1554 | 0/3 | OUTlow other 'Parutions recentes' book-notices column in a political history bulletin with no abstract; a bibliographic listing rather than a study of research practice, but the payload is too thin to judge confidently. | OUTlow other The title indicates a recent-publications item, but there is insufficient information to identify a metaresearch object. | T3medium other Title is a recent-publications listing with no abstract; announcement-style item, not analytic metaresearch. |
| Scalable Blockchain-assisted Log Storage System for Cloud-generated Logs 2021 · preprint · en | aff_core w=5595 | 0/3 | OUThigh empirical Computer science preprint proposing a blockchain-assisted cloud log storage system; the object is a software architecture. | OUThigh conceptual The thesis proposes a blockchain system for securing cloud logs, not research infrastructure or practice. | OUThigh empirical CS systems thesis building blockchain log storage; tool-building, not studying research. |
| Céline MORIN (2017), Les héroïnes de séries américaines de Ma sorcière bien-aimée à The Good Wife 2018 · article · fr · Communication | french w=1554 | 0/3 | OUThigh editorial/commentary Book review on heroines of American TV series; media and gender studies commentary. | OUThigh conceptual This analyzes female protagonists in American television series, not the research system. | OUThigh editorial/commentary Book review of American television heroines; media/cultural studies, not research systems. |
| Self-Assessment 2015 · article · en · Dermatitis | venue_new w=2684 | 0/3 | OUTlow other A dermatology journal self-assessment CME feature; the abstract is only a journal blurb, with no metaresearch content. | OUTlow other The title and journal overview are insufficient to identify a research-system object. | T3medium infrastructure/announcement Journal self-description/promotional overview of Dermatitis; contextual scholarly-communication material, not analytic metaresearch. |
| <i>Peach latent mosaic viroid</i> . [Distribution map]. 2006 · article · en · Distribution Maps of Plant Diseases | about_only w=3321 | 0/3 | OUThigh other Plant pathology distribution map for Peach latent mosaic viroid; a geographic data note about a pathogen. | OUThigh empirical The work maps the geographic distribution of a plant viroid rather than studying research practice. | OUThigh other Plant disease distribution map; object is a plant pathogen, not research practice. |
| Complete Superior Sagittal Sinus Thrombosis Along With Subarachnoid Hemorrhage During Early Postpartum Period, Presented by Left Hemiparesis: A Rare Case Report 2017 · article · en · Journal of Neurology Research | venue_new w=2684 | 0/3 | OUThigh empirical Clinical case report of postpartum sinus thrombosis. | OUThigh empirical This is a clinical case report about cerebral venous thrombosis, not research practice. | OUThigh empirical Clinical case report of postpartum cerebral venous thrombosis; medical case study. |
| Hutton, Maurice, (1856–5 April 1940), Queen’s University, Kingston; McGill University, Montreal, and Macmaster’s University, Toronto; Professor of Greek, 1887, and Principal 1901–28, of University College, Toronto; Acting President of the University, 1906–07 and 1925; retired, 1928; Member of the Royal Society of Canada (English section); President of this Section, 1918–19; President of the League of the Empire (Canadian Branch); on the Executive of the League of Nations (Toronto Branch) 2007 · reference-entry · en · Who Was Who | about_only w=3321 | 0/3 | T3low other Biographical reference entry (obituary-type record) for a Canadian university professor and principal; contextual academic record rather than a study of research. | OUThigh other This is a biographical reference entry about a university professor. | T3high other Who Was Who biographical reference entry/obituary, not a study of research. |
| Canadian Thoracic Society Pulmonary Vascular Diseases Committee 2004 · article · en · Canadian Respiratory Journal | about_only w=3321 | 0/3 | OUTmedium other Item associated with the Canadian Thoracic Society Pulmonary Vascular Diseases Committee with no abstract; the content is clinical society business in respirology, not research practice, but the item type cannot be determined. | OUTlow other The title identifies a professional committee but provides too little information to establish a metaresearch object. | T3medium other Committee masthead/listing title only; not a study of research, mapped as non-analytic item. |
| Multiplication of Vector and Matrix 2014 · book-chapter · en | aff_core w=5595 | 0/3 | OUThigh other Textbook chapter on vector and matrix multiplication; a mathematics/computing topic. | OUThigh other The title identifies a mathematical textbook-style topic unrelated to research itself. | OUThigh other Linear algebra textbook chapter on vector and matrix multiplication; unambiguous domain content. |
| Corrigendum 2019 · erratum · en · Journal of Nutrition Education and Behavior | venue_new w=2684 | 0/3 | OUTlow other Bare corrigendum in a nutrition journal with no metaresearch content; a correction notice attached to domain work rather than to the metaresearch ecosystem. | T3high other A corrigendum is contextual scholarly communication material and is not part of the analytic corpus. | T3high other Corrigendum is a correction notice; contextual bibliographic item, never pooled. |
| 10.31030/9598470 2000 · article · en · Time to knit | venue_new w=2684 | 0/3 | OUThigh other Empty/erroneous DOI record whose 'abstract' is Crossref boilerplate; no research content. | T3high infrastructure/announcement This is a DOI metadata and persistence notice concerning scholarly infrastructure. | OUThigh other Orphan DOI metadata stub with no scholarly content about research. |
| Ed Bd, Masthead, Comm List, Membership List and TOC 2018 · article · en · Journal of Bone and Mineral Research | about_only w=3321 | 0/3 | OUThigh other Journal front matter: masthead, editorial board and table of contents. | T3high infrastructure/announcement The record is journal front matter and contextual scholarly-publishing metadata. | OUThigh other Journal masthead, membership list, and TOC administrative front matter, not a study of research. |
| Behaviour of bridges under spatially non-uniform earthquake motions 2020 · article · en · Proceedings of the World Congress on Civil, Structural, and Environmental Engineering | venue_new w=2684 | 0/3 | OUThigh empirical Structural engineering study of bridge response to non-uniform seismic motion. | OUThigh empirical The work studies bridge responses to earthquakes rather than research practice. | OUThigh empirical Structural engineering study of bridge seismic response; domain engineering, not metaresearch. |
| The clinically important impact of preschool oral immunotherapy on parental quality of life 2021 · preprint · en | aff_core w=5595 | 0/3 | OUThigh empirical Clinical study of preschool oral immunotherapy and parental quality of life; a patient/family outcome question. | OUThigh empirical The preprint studies clinical effects of oral immunotherapy on parents rather than research practice. | OUTmedium empirical Clinical impact of preschool oral immunotherapy on parental quality of life. |
| On the System Algebra Foundations for Granular Computing 2012 · book-chapter · en · IGI Global eBooks | aff_core w=5595 | 0/3 | OUThigh conceptual Theoretical computer science on system algebra foundations for granular computing. | OUThigh conceptual The paper develops mathematical foundations for granular computing rather than studying research practice. | OUThigh conceptual Computer-science theory of granular computing and system algebra, not metaresearch. |
| Public Administration after "New Public Management" 2010 · book · en · Value for money in government | about_only w=3321 | 0/3 | OUThigh policy OECD study of public administration reform after New Public Management; government management, not the research system. | OUThigh policy This report analyzes public-administration reforms rather than the research system. | OUThigh policy OECD public-administration reform and value-for-money government; public management, not research policy as object. |
| LNG Imports Into North America and Implications 2008 · article · en · Proceedings of Offshore Technology Conference | about_only w=3321 | 0/3 | OUThigh other Industry conference paper on LNG imports and gas markets; an energy-economics question. | OUThigh empirical The work analyzes LNG markets, supply, demand, and pricing rather than research practice. | OUThigh other Industry energy paper on LNG imports into North America; commercial energy supply, not research. |
| Diversity of cultured photosynthetic flagellates in the North East Pacific and Arctic Oceans in summer 2012 · preprint · en | aff_core w=5595 | 0/3 | OUThigh empirical Taxonomy and diversity of cultured marine photosynthetic flagellates. | OUThigh empirical It studies the diversity and phylogeny of marine flagellates. | OUThigh empirical Phytoplankton strain isolation and diversity is domain marine microbiology. |
| 10.5840/croatjphil20044221 2000 · article · en · Time to knit | venue_new w=2684 | 0/3 | OUTlow other Stub record whose 'abstract' is Crossref boilerplate about erroneously created DOIs; no research content. | T3high infrastructure/announcement The record is a contextual notice about DOI metadata and deletion procedures. | OUTlow other Unattached DOI error record with no real scholarly object to classify. |
| 01 DOSE ESCALATION WITH ORIGINATOR INFLIXIMAB IS MORE COMMON THAN STANDARD DOSING IN PEDIATRIC IBD – THE DEVELOP EXPERIENCE 2019 · article · en · Inflammatory Bowel Diseases | about_only w=3321 | 0/3 | OUThigh empirical Registry analysis of infliximab dose escalation in pediatric IBD; clinical question. | OUThigh empirical The registry study examines infliximab dosing in pediatric inflammatory bowel disease. | OUThigh empirical Observational registry of infliximab dosing in pediatric IBD; clinical therapeutics. |
| 10.1037/e551712010-001 2000 · article · en · Time to knit | venue_new w=2684 | 0/3 | OUTlow other Stub record whose 'abstract' is Crossref boilerplate about erroneously created DOIs; no research content despite the identifier vocabulary. | T3high infrastructure/announcement The record is a contextual notice about DOI metadata and deletion procedures. | OUTlow other DOI-error placeholder with no usable bibliographic content for inclusion decisions. |
| The Language of Nature 2016 · book · en · University of Minnesota Press eBooks | about_only w=3321 | 0/3 | OUTlow conceptual History and philosophy of the seventeenth-century mathematization of nature; HPS, but the object is not contemporary research practice as the rubric requires. | OUThigh conceptual The book examines the mathematization of nature in early modern science, not contemporary research practice. | OUTmedium conceptual History of philosophy of the 17th-century mathematization of nature; HPS counts as T2 only when object is contemporary research practice. |
| How Can Entrepreneurs Motivate Crowdsourcing Participants? 2013 · article · en · Technology Innovation Management Review | aff_core w=5595 | 0/3 | OUThigh conceptual Practitioner article on how technology entrepreneurs can motivate crowdsourcing participants; the object is business practice, not research. | OUTlow conceptual The work concerns motivating crowdsourcing participants for entrepreneurship, not research itself, and the abstract is incomplete. | OUThigh conceptual Entrepreneurship piece on motivating crowdsourcing participants; not about research practice. |
| The topology of restricted partition posets 2011 · article · fr · Discrete Mathematics & Theoretical Computer Science | french w=1554 | 0/3 | OUThigh conceptual Pure mathematics on the topology of restricted partition posets; the object is a combinatorial structure. | OUThigh empirical This mathematics paper studies the topology of partition posets, not research itself. | OUThigh conceptual Pure combinatorial topology of restricted partition posets; mathematics domain. |
| Speaking with Authority: Reading Catherine of Siena in the Times of Vittoria Colonna 2022 · article · en · Renaissance and Reformation | venue_new w=2684 | 0/3 | OUThigh conceptual Literary and religious analysis of Vittoria Colonna and Catherine of Siena. | OUThigh conceptual This is a literary and religious analysis of Catherine of Siena and Vittoria Colonna, not research practice. | OUThigh conceptual Literary-historical reading of Vittoria Colonna and Catherine of Siena; humanities domain. |
| Leib Malach’s Montreal Travelogue, 1930 2022 · article · en · Canadian Jewish Studies / Études juives canadiennes | venue_new w=2684 | 0/3 | OUThigh conceptual Historical essay on a 1930 Yiddish Montreal travelogue; cultural history, not the research system. | OUThigh conceptual This analyzes a historical Yiddish travelogue and cultural conditions in Montreal, not the research system. | OUThigh conceptual Historical/cultural essay on a 1930 Yiddish travelogue about Montreal Jewish life, not research practice. |
| In the Spotlight—Established Researcher 2022 · article · en · Journal of Experimental Zoology Part B Molecular and Developmental Evolution | about_only w=3321 | 0/3 | T3low other Journal spotlight interview profiling an established researcher's training, career path and editorial roles; a contextual career/news piece, not a study of the research workforce. | T3high editorial/commentary This is a researcher profile rather than a study of research practice. | T3high other Researcher profile interview/spotlight piece, not a study of research practice. |
| Phase I trial of sargramostim/pelareorep therapy in pediatric patients with recurrent or refractory high-grade brain tumors 2022 · article · en · Neuro-Oncology Advances | aff_core w=5595 | 0/3 | OUThigh empirical Phase I clinical trial of an oncolytic virus therapy in pediatric brain tumors. | OUThigh empirical This is a phase I clinical trial of a brain-tumor therapy, not metaresearch. | OUThigh empirical Phase I clinical trial of oncolytic therapy in pediatric brain tumours. |
| Après le Déluge: Periodizing Canada’s Episode of Light 2022 · book-chapter · fr · The MIT Press eBooks | french w=1554 | 0/3 | OUTlow other No abstract; title indicates a Canadian cultural/historical periodization essay, with nothing suggesting research as its object. | OUThigh other This book chapter concerns Canadian intellectual history, not the Canadian research system. | OUTmedium conceptual Canadian cultural/historical periodization chapter; missing abstract but title is not about research practice. |
| TIAS: Two-level Information-Agnostic Job Scheduling in GPU Clusters 2021 · article · en | aff_core w=5595 | 0/3 | OUThigh empirical GPU cluster job scheduling algorithm for deep learning workloads; a computer systems contribution, not a study of research. | OUThigh empirical The work studies GPU job scheduling for machine-learning workloads, not research as a system or practice. | OUThigh empirical Systems paper on GPU-cluster job scheduling for deep learning workloads, not study of research practice. |
| Rechargeable Metal–Hydrogen Peroxide Battery, A Solution to Improve the Metal–Air Battery Performance 2022 · article · en · ACS Energy Letters | aff_core w=5595 | 0/3 | OUThigh empirical Electrochemistry study of a rechargeable metal-hydrogen peroxide battery. | OUThigh empirical The study develops a rechargeable battery technology rather than examining research. | OUThigh empirical Battery electrochemistry proposal for metal–hydrogen peroxide systems; domain energy materials. |
| Foraging on anthropogenic food predicts problem-solving skills in a seabird 2022 · article · en · The Science of The Total Environment | aff_core w=5595 | 0/3 | OUThigh empirical Behavioural ecology study linking anthropogenic foraging to problem solving in seabirds. | OUThigh empirical This ecological study examines seabird foraging and problem-solving, not research. | OUThigh empirical Seabird behavioural ecology; title unambiguous domain science despite missing abstract. |
| Abstracts ‐ AAPA Presentations 2013 · article · en · American Journal of Physical Anthropology | aff_core w=5595 | 0/3 | OUTlow other Container record of conference presentation abstracts in physical anthropology; a collection of domain research reports, not a study of research. | T3high other It is a collection of conference presentation abstracts, relevant as contextual research output but not for the analytic corpus. | OUTmedium other Conference abstracts container/journal blurb; not a study of research practice. |
| Notices bio-bibliographiques 2004 · article · fr · Intermédialités Histoire et théorie des arts des lettres et des techniques | french w=1554 | 0/3 | OUTlow other Bio-bibliographical notices front matter with no abstract; journal contributor notes, not a study. | OUTlow other The title indicates biobibliographical notices but provides too little information to establish a research object. | OUTlow other Bio-bibliographical notices with no abstract; title indicates contributor notes rather than study of research. |
| Amino acid transporter gene TaATLa1 from Triticum aestivum L. improves growth under nitrogen sufficiency and is down regulated under nitrogen deficiency 2022 · article · en · Planta | aff_core w=5595 | 0/3 | OUThigh empirical Plant biology study of a wheat amino acid transporter gene under nitrogen conditions. | OUThigh empirical It studies a plant transporter gene and nitrogen-related growth, not research itself. | OUThigh empirical Wheat amino-acid transporter gene under nitrogen regimes; plant science. |
| Evaluation of Sleep M Biocrystal Pilow/Topper/Mattress in Clinical Trials using 20 Subjects to Test for Pain Reduction, Sleep Improvement, Stress Reduction and Improvement in Circulation for Clients with Diabetes over 3 Months 2020 · article · en · Open Access Journal of Gynecology | about_only w=3321 | 0/3 | OUThigh empirical Trial of a consumer sleep product; the object is the product's health effects. | OUThigh empirical It reports a small clinical intervention study of a sleep product. | OUThigh empirical Consumer sleep-product biofeedback trial is domain product evaluation, not research methods. |
| Mobile Charging Station Placements in Internet of Electric Vehicles: A Federated Learning Approach 2022 · article · en · IEEE Transactions on Intelligent Transportation Systems | aff_core w=5595 | 0/3 | OUThigh empirical Federated learning method for placing mobile EV charging stations; an engineering optimization question. | OUThigh empirical The study proposes a federated-learning method for electric-vehicle charging placement. | OUThigh empirical Federated-learning placement of mobile EV charging stations is engineering optimization, not study of research. |
| Poster: AML-484 First Results of a Phase II Study (STIMULUS-AML1) Investigating Sabatolimab + Azacitidine + Venetoclax in Patients With Newly Diagnosed Acute Myeloid Leukemia (ND AML) 2022 · article · en · Clinical Lymphoma Myeloma & Leukemia | aff_core w=5595 | 0/3 | OUThigh empirical Conference poster reporting first results of a phase II AML treatment study; a clinical trial report. | OUThigh empirical It reports a leukemia treatment study rather than studying research practice. | OUThigh other AML clinical trial poster of a drug study; domain oncology trial, not research-on-research. |
| Eptifibatide use in ischemic stroke patients undergoing endovascular thrombectomy: A matched cohort analysis 2022 · article · en · Frontiers in Neurology | about_only w=3321 | 0/3 | OUThigh empirical Matched cohort analysis of drug safety during thrombectomy; clinical question. | OUThigh empirical This cohort analysis evaluates a stroke treatment, not research practice. | OUThigh empirical Clinical stroke cohort on eptifibatide during thrombectomy. |
| THE RESOLVE SURVEY ATOMIC GAS CENSUS AND ENVIRONMENTAL INFLUENCES ON GALAXY GAS RESERVOIRS 2020 · article · en · UNC Libraries | fund_new w=1679 | 0/3 | OUThigh empirical Astronomy survey paper on the atomic gas census of RESOLVE galaxies. | OUThigh empirical It analyzes galaxy gas reservoirs and environments, not research itself. | OUThigh empirical Astronomy census of galaxy HI gas and environment; observational astrophysics. |
| Neurotoxic Consequences of Antiretroviral Therapies 2014 · book-chapter · en | aff_core w=5595 | 0/3 | OUThigh other Book chapter on neurotoxic consequences of antiretroviral therapy; a clinical pharmacology topic. | OUThigh other The chapter concerns neurotoxic effects of antiretroviral therapies, not research itself. | OUTlow other Title indicates clinical/neurotoxic domain content on antiretroviral therapies; no abstract. |
| Challenges for Diadromous Fishes in a Dynamic Global Environment 2009 · book-chapter · en · American Fisheries Society eBooks | about_only w=3321 | 0/3 | OUThigh empirical Population genetics of smelt colonization; evolutionary biology. | OUThigh empirical This investigates fish evolution and population structure, not research practice. | OUThigh empirical Fish population genetics and phenotype evolution of smelt; domain biology. |
| Frontmatter 2022 · book-chapter · en · University of Toronto Press eBooks | fund_new w=1679 | 0/3 | OUTlow other Book front matter with no content; nothing to indicate the volume studies research itself. | OUThigh other Frontmatter is not a study of research itself. | OUTlow other Book frontmatter with empty abstract; no scholarly object of study to classify as metaresearch. |
| Fungal infection and chronic lung allograft dysfunction: A dangerous combination 2022 · editorial · en · Transplant Infectious Disease | aff_core w=5595 | 0/3 | OUThigh editorial/commentary Clinical editorial on fungal infection and lung allograft dysfunction; the object is patient care. | OUThigh editorial/commentary This editorial concerns fungal infection and transplantation, not research practice. | OUThigh editorial/commentary Clinical editorial on fungal infection and allograft dysfunction, not commentary on research practice. |
| The Colour of Horror 2022 · article · en | aff_core w=5595 | 0/3 | OUThigh empirical Computational extraction of color palettes from film trailers; a media analysis question. | OUThigh empirical The work uses computational methods to analyze color palettes in films. | OUThigh empirical Computational colour-palette analysis of film trailers is media/CS application, not metaresearch. |
| Effects of Carotid Artery Stiffness on Cerebral Small‐Vessel Disease and Cognition 2022 · article · en · Journal of the American Heart Association | about_only w=3321 | 0/3 | OUThigh empirical Clinical cohort study of carotid artery stiffness, small-vessel disease and cognition. | OUThigh empirical It examines carotid stiffness and dementia outcomes in patients, not research practice. | OUThigh empirical Clinical cohort study linking carotid stiffness to cognition and dementia, not research practice. |
| Evaluation of feed restriction and abomasal infusion of resistant starch as models to induce intestinal barrier dysfunction in healthy lactating cows 2022 · article · en · Journal of Dairy Science | aff_core w=5595 | 0/3 | OUTmedium empirical Evaluation of experimental models for inducing intestinal barrier dysfunction in cows; bench/animal model development, not metaresearch. | OUThigh empirical The experiment evaluates models of intestinal dysfunction in cows, not research itself. | OUThigh empirical Dairy-cow experimental models of intestinal barrier dysfunction; animal science. |
| Experimental and Numerical Investigations on the Effect of a Novel Internal Surface Micro Grooving Towards Improving Convective Heat Transfer Performance of a Tube Heat Exchanger 2022 · article · en · SSRN Electronic Journal | aff_core w=5595 | 0/3 | OUThigh empirical No abstract; experimental and numerical study of heat exchanger micro-grooving; thermal engineering. | OUThigh empirical The study tests a heat-transfer modification for tube heat exchangers. | OUTmedium empirical Heat-exchanger surface engineering for heat transfer; mechanical engineering (no abstract). |
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