1,000 works · Opus 4.8 · GPT-5.6 (high) · Grok 4.5 · one locked rubric
The field’s boundary is not a line. It is a region.
Three frontier models screened the same 1,000 works, drawn from the real frame with known selection probabilities, against the same locked rubric on its full eight-field payload. Of the 51 works any model called metaresearch, only 19 (37.3%) were called metaresearch by all three, and 24 (47.1%) rest on a single model's opinion.
Two screeners can agree on a rate while finding almost entirely different works. At a ~1% base rate, the settled rejects buy 98% agreement for free — which is why an agreement statistic computed over the whole sample tells you nothing about the boundary, and why the table below, not a percentage, is the deliverable.
The three models are not interchangeable
How many of the 1,000 works each model called metaresearch (tier T1 or T2), on identical input. T3 is adjacent and does not count as in scope, which is why a model's count can never exceed the 51 works in the dossier.
Swap which model you call “the screener” and the size of the field moves. That spread — not the binomial confidence interval on any one model's labels — is the honest uncertainty on how big Canadian metaresearch is.
The disagreement dossier
All three verdicts, confidences and reasons side by side. These are the works against which the inclusion criteria have to be written — because these are the works on which reasonable screeners, given the same rubric and the same evidence, disagree.
| Work | Stratum | n_in | Opus 4.8 | GPT-5.6 | Grok 4.5 |
|---|---|---|---|---|---|
| The UTIAS multi-robot cooperative localization and mapping dataset 2011 · article · en · The International Journal of Robotics Research | aff_core w=5595 | 0/3 | OUTmedium infrastructure/announcement Descriptor for a multi-robot localization and mapping dataset; sharing a domain benchmark dataset is doing research, not studying data sharing, so it falls outside the metaresearch ecosystem despite being a resource paper. | OUThigh empirical The paper presents a robotics dataset for use in research but does not study research infrastructure itself. | OUThigh empirical Robotics multi-robot localization and SLAM dataset for domain algorithms, not research practice. |
| Model of cell signal transduction in a three-dimensional domain 2011 · article · en · Journal of Mathematical Biology | aff_core w=5595 | 0/3 | OUThigh empirical Mathematical model of cell signal transduction in a 3D domain; a mathematical-biology question. | OUThigh conceptual It presents a model of cellular signaling, not research practice. | OUThigh other Mathematical model of cell signal transduction; domain computational biology. |
| Recent star formation in the Lupus clouds as seen by<i>Herschel</i> 2012 · article · en · Astronomy and Astrophysics | aff_core w=5595 | 0/3 | OUThigh empirical Astrophysical study of star formation in the Lupus clouds. | OUThigh empirical The article studies star formation in molecular clouds, not the conduct of research. | OUThigh empirical Astrophysics of star formation in Lupus clouds, domain science. |
| Use of isolated digestive-gland cells in the study of biochemical and physiological processes in gastropod molluscs 2000 · article · en · Comparative Biochemistry and Physiology Part A Molecular & Integrative Physiology | aff_core w=5595 | 0/3 | OUThigh empirical Laboratory technique paper on isolated digestive-gland cells in molluscs; a bench method, not research methodology as an object of study. | OUThigh empirical This work uses isolated cells to study gastropod physiology, not research methods as an object. | OUThigh empirical Mollusc physiology methods paper using digestive-gland cells; domain biology despite 'study of' phrasing. |
| Why so repressed? Turning off transcription during plant growth and development 2009 · review · en · Current Opinion in Plant Biology | fund_new w=1679 | 0/3 | OUThigh conceptual Review of transcriptional repression in plant development; plant molecular biology. | OUThigh other This review concerns plant transcription biology, not the practice or system of research. | OUThigh conceptual Plant molecular biology review of transcriptional repression; title unambiguously domain research. |
| FES kinases are required for oncogenic FLT3 signaling 2010 · article · en · Leukemia | fund_new w=1679 | 0/3 | OUThigh empirical FES kinases in oncogenic FLT3 signaling; cancer cell biology. | OUTlow empirical The title indicates a cancer-biology study, not research itself, but the abstract is missing. | OUThigh empirical Biomedical paper on FES kinases in oncogenic FLT3 signaling; title is unambiguous domain biology. |
| Baroclinic Instability in an Euler Equations–Based Column Model: The Coexistence of a Deep Synoptic-Scale Mode and Shallow Subsynoptic-Scale Modes 2001 · article · en · Journal of the Atmospheric Sciences | aff_core w=5595 | 0/3 | OUThigh empirical Baroclinic instability column model in atmospheric dynamics; a geophysical fluid-dynamics question. | OUThigh empirical This analyzes atmospheric instability and meteorological models, not research practice. | OUThigh empirical Atmospheric dynamics analysis of baroclinic instability modes; domain geophysical fluid dynamics. |
| Flexural Performance and Moment Connection of Concrete-Filled GFRP Tube–Encased Steel I-Sections 2012 · article · en · Journal of Composites for Construction | aff_core w=5595 | 0/3 | OUThigh empirical Structural testing of concrete-filled FRP tube sections; civil engineering. | OUThigh empirical The study tests the structural performance of concrete-filled composite members. | OUThigh empirical Structural engineering tests of composite beam systems; domain construction research. |
| Model-based oscillometric blood pressure estimation 2014 · article · en | aff_core w=5595 | 0/3 | OUThigh empirical Biomedical engineering paper proposing a model-based algorithm for oscillometric blood pressure estimation; the object is a device measurement method. | OUThigh conceptual It reviews biomedical blood-pressure estimation algorithms, not research methodology. | OUThigh empirical Biomedical engineering of oscillometric blood-pressure algorithms; clinical measurement, not research practice. |
| Physical origin of nanograting formation on fused silica with femtosecond pulses 2014 · article · en · Applied Physics Letters | aff_core w=5595 | 0/3 | OUThigh empirical Laser physics study of the physical origin of nanograting formation in fused silica; the object is a material process. | OUThigh empirical This physics study explains nanograting formation in fused silica, not research methodology. | OUThigh empirical Laser physics study of nanograting formation on fused silica. |
| Proteomic approaches in brain research and neuropharmacology 2004 · review · en · European Journal of Pharmacology | aff_core w=5595 | 0/3 | OUTmedium other Technical review of proteomic laboratory approaches for brain research; a review of bench techniques in a domain, not of research practice, though abstract is absent. | OUTlow conceptual The title suggests a review of proteomic methods applied to brain research, but the missing abstract prevents a clear metaresearch determination. | OUTmedium other Title indicates domain proteomic methods applied to brain and neuropharmacology, not study of research practice. |
| A stochastic time series generator with adaptive software architecture 2010 · article · en | aff_core w=5595 | 0/3 | OUTmedium infrastructure/announcement Software implementation of a stochastic hydrologic time series generator; a domain modeling tool, not scholarly research infrastructure. | OUThigh empirical The paper presents a hydrologic time-series generator rather than studying research practice. | OUThigh empirical Hydrologic time-series software tool; uses computing for domain data, does not study research infrastructure as such. |
| Industrial Unionism and Its Bargaining Correlates in Nigeria Industrial Relations System 2013 · article · en · Journal of Management and Strategy | venue_new w=2684 | 0/3 | OUThigh conceptual Analysis of industrial unionism and collective bargaining in Nigeria; the object is an industrial relations system. | OUThigh conceptual The paper analyzes industrial unionism and collective bargaining in Nigeria, not research itself. | OUThigh empirical Industrial relations and unionism in Nigeria; labor studies domain. |
| Treffers-Daller, Jeanine and Willemyns, Ronald (eds.), <i>Language Contact at the Romance-Germanic Language Border</i>. Clevedon/Buffalo/Toronto/Sydney: Multilingual Matters, 2002, 149 pp. 1 85359 627 2. (Also available as Volume 23, Nos 1&2 of the <i>Journal of Multilingual and Multicultural Development</i>.) 2003 · article · en · Journal of French Language Studies | about_only w=3321 | 0/3 | OUThigh editorial/commentary Book review of a linguistics volume; criticism of one book, not a study of research. | OUThigh editorial/commentary The item is a bibliographic review of a language-contact book and does not study research itself. | OUThigh editorial/commentary Bibliographic book notice on Romance–Germanic language contact; linguistics, not study of research practice. |
| Dysfunctional HDL containing L159R ApoA-I leads to exacerbation of atherosclerosis in hyperlipidemic mice 2011 · article · en · Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids | fund_new w=1679 | 0/3 | OUThigh empirical Lipid biology of dysfunctional HDL in atherosclerosis; abstract absent but the title is unambiguous. | OUThigh empirical It studies lipid biology and atherosclerosis in hyperlipidemic mice. | OUThigh empirical Title is clear domain atherosclerosis/lipid research despite missing abstract. |
| Genetic variation and differentiation of populations within the <i>Quercus affinis</i> <i>Quercus laurina</i> (Fagaceae) complex analyzed with RAPD markers 2005 · article · en · Canadian Journal of Botany | venue_new w=2684 | 0/3 | OUThigh empirical Population genetics of hybridizing Mexican oaks using RAPD markers. | OUThigh empirical It studies genetic differentiation and hybridization in Mexican oak populations. | OUThigh empirical Population genetics of hybridizing oaks is domain evolutionary biology. |
| Extent of, and factors associated with, delay to hospital presentation in patients with acute coronary disease (the GRACE registry) 2002 · article · en · The American Journal of Cardiology | aff_core w=5595 | 0/3 | OUThigh empirical Registry study of delay to hospital presentation in acute coronary disease; abstract absent but the title is unambiguous. | OUThigh empirical It studies delays in hospital presentation among patients with acute coronary disease. | OUThigh empirical Title indicates clinical study of delay to hospital in ACS; missing abstract does not change object. |
| Influence of the Dispersed Particulate in Chocolate on Cocoa Butter Microstructure and Fat Crystal Growth during Storage 2008 · article · en · Food Biophysics | aff_core w=5595 | 0/3 | OUThigh empirical Food science on particulate effects on cocoa butter microstructure and fat crystal growth; the object is a food material. | OUThigh empirical The study investigates chocolate microstructure and fat crystal growth, not research practice. | OUTmedium empirical Title indicates food science on chocolate microstructure; not research-as-object. |
| Flat-top oscillons in an expanding universe 2010 · article · en · Physical review. D. Particles, fields, gravitation, and cosmology/Physical review. D, Particles, fields, gravitation, and cosmology | fund_new w=1679 | 0/3 | OUThigh conceptual Theoretical cosmology of oscillon field configurations. | OUThigh conceptual It studies cosmological field configurations, not research itself. | OUThigh empirical Theoretical cosmology of oscillon field configurations in an expanding universe; pure physics. |
| Donepezil in Vascular Dementia 2003 · article · en · Drugs & Aging | aff_core w=5595 | 0/3 | OUThigh other Drug review of donepezil in vascular dementia; abstract absent but the title is unambiguously a clinical therapeutics topic. | OUTlow conceptual The clinical treatment title does not indicate metaresearch, but the abstract is missing. | OUTmedium other Title indicates a clinical drug review of donepezil in vascular dementia; no abstract. |
| Policy rule Nash equilibria 2003 · article · fr · Canadian Journal of Economics/Revue canadienne d économique | french w=1554 | 0/3 | OUThigh empirical Economics paper on Nash equilibria of monetary policy rules. | OUThigh conceptual It develops mathematical models of monetary policy equilibria, not research itself. | OUThigh conceptual Monetary policy game-theory model of instrument rules; economics domain, not research systems. |
| Charles Bernstein's<i>Shadowtime</i>and faithful interpretation 2007 · article · en · Textual Practice | about_only w=3321 | 0/3 | OUThigh conceptual Literary and philosophical criticism of Charles Bernstein's libretto. | OUThigh conceptual The article provides literary and philosophical interpretation rather than metaresearch. | OUThigh conceptual Literary and philosophical reading of Bernstein, not research practice. |
| Health care in Romania: Fighting collapse 2010 · article · en · Canadian Medical Association Journal | venue_new w=2684 | 0/3 | OUThigh editorial/commentary News commentary on Romanian health care financing; the object is a health system, not research. | OUTlow other The title and available abstract concern Romanian health-care financing, not research itself. | OUThigh editorial/commentary News-style commentary on Romanian healthcare financing collapse, not research systems or practice. |
| How Does Brand-related User-generated Content Differ across YouTube, Facebook, and Twitter? 2012 · article · en · Journal of Interactive Marketing | aff_core w=5595 | 0/3 | OUThigh empirical Content analysis of brand-related user-generated content across social platforms; a marketing question. | OUThigh empirical The object is brand-related consumer communication on social media, not research itself. | OUThigh empirical Marketing study of brand-related social media content; object is consumer communication, not research. |
| The Case for the World Health Organization’s Commission on the Social Determinants of Health to Address Sexual Orientation 2012 · article · en · American Journal of Public Health | aff_core w=5595 | 0/3 | OUTmedium editorial/commentary Commentary arguing that WHO should treat sexual orientation as a social determinant of health; a health-policy argument rather than a study of research practice. | OUThigh conceptual It advocates a change in health-equity policy concerning sexual orientation, not research practice or governance. | OUThigh policy Public-health policy argument to include sexual orientation in WHO social determinants; not research-system policy. |
| Radiation-Induced High Toxicities and Local Recurrence in Stage 3 and 4 Oropharynx Cancer Patients 2014 · article · en · Journal of medical imaging and radiation sciences | aff_core w=5595 | 0/3 | OUThigh empirical Clinical study of radiation toxicity and local recurrence in advanced oropharynx cancer; a treatment outcome question. | OUThigh empirical The title describes clinical outcomes in oropharyngeal cancer rather than research itself. | OUTmedium empirical Clinical oncology title on radiation toxicities and recurrence; domain medicine with thin metadata. |
| Relationship between managerial values and hiring preferences in the context of the six decades of affirmative action in India 2012 · article · en · Equality Diversity and Inclusion An International Journal | aff_core w=5595 | 0/3 | OUThigh empirical Survey of managerial values and minority hiring preferences; the workplace, not the research workforce. | OUThigh empirical The study examines managerial values and hiring preferences in India. | OUThigh empirical Managerial values and hiring preferences under Indian affirmative action is HR/management research, not metaresearch. |
| The nature of care in light of Emmanuel Levinas 2006 · article · en · Nursing Philosophy | aff_core w=5595 | 0/3 | OUThigh conceptual Philosophical analysis of the concept of care in nursing via Levinas; disciplinary theory, not a study of research practice. | OUThigh conceptual The article develops a philosophical account of care in nursing. | OUThigh conceptual Philosophical account of care via Levinas for nursing; nursing ontology, not research about research. |
| An integrated continuous-wave terahertz biosensor 2007 · article · en · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE | aff_core w=5595 | 0/3 | OUThigh empirical Engineering design of a terahertz biosensor. | OUThigh empirical This develops a terahertz biosensor, with the object being an engineering device rather than research. | OUThigh empirical Engineering paper proposing a terahertz biosensor device design. |
| A concentration-dependent endocytic trap and sink mechanism converts Bmper from an activator to an inhibitor of Bmp signaling 2009 · article · en · The Journal of Cell Biology | aff_core w=5595 | 0/3 | OUThigh empirical Cell-biology study of Bmper endocytic regulation of Bmp signaling; the object is a molecular mechanism. | OUThigh empirical The study investigates molecular signaling mechanisms rather than how research is conducted or communicated. | OUThigh empirical Molecular cell-biology study of Bmper and Bmp signaling mechanisms. |
| Hypothalamic resistin induces hepatic insulin resistance 2007 · article · en · Journal of Clinical Investigation | aff_core w=5595 | 0/3 | OUThigh empirical Physiology study of hypothalamic resistin and hepatic insulin resistance; the object is a metabolic mechanism. | OUThigh empirical The study investigates hypothalamic resistin and insulin resistance, not research practice. | OUThigh empirical Physiology study of hypothalamic resistin and insulin resistance; biomedical domain. |
| CONSORT statement requires closer examination 2002 · letter · en · BMJ | aff_core w=5595 | 0/3 | T3high editorial/commentary Letter to the editor arguing about how CONSORT reporting guidance is applied; commentary about research reporting practice, not a study of it. | T3high editorial/commentary This letter comments on CONSORT reporting guidance and research reporting practice. | T3high editorial/commentary Letter discussing CONSORT trial-reporting guidance, commentary on research reporting rather than empirical metaresearch. |
| GIS-based numerical modelling of debris flow motion across three-dimensional terrain 2013 · article · en · Journal of Mountain Science | fund_new w=1679 | 0/3 | OUThigh empirical GIS-coupled numerical model of debris flow; a geohazard modeling contribution. | OUThigh empirical This study develops and applies a debris-flow simulation model to a geological case. | OUThigh empirical GIS numerical modelling of debris-flow motion; geohazard engineering, not research about research. |
| Some Observations on the Development of Superior Photocatalytic Systems for Application to Water Purification by the “Adsorb and Shuttle” or the Interphase Charge Transfer Mechanisms 2014 · article · en · Molecules | aff_core w=5595 | 0/3 | OUThigh empirical Photocatalysis experiments and review for water purification; chemistry. | OUThigh empirical This reviews and tests photocatalytic systems for water purification, not evidence-synthesis methodology. | OUThigh empirical Photocatalysis materials science for water purification. |
| Evaluating Spatio-temporal Complexities of Forest Management: An Integrated Agent-based Modeling and GIS Approach 2008 · article · en · Environmental Modeling & Assessment | aff_core w=5595 | 0/3 | OUThigh empirical Agent-based modeling and GIS approach to forest management complexities; the object is forest management. | OUThigh empirical The study models forest-management complexity, not research itself. | OUTmedium empirical Agent-based modeling of forest management; environmental modeling domain, not research practice. |
| Fatty acid profiles of polar and non-polar lipids of Pleurotus ostreatus and P. cornucopiae var. ‘citrino-pileatus’ grown at different temperatures 2007 · article · en · Mycological Research | aff_core w=5595 | 0/3 | OUThigh empirical Mycology study of fatty acid profiles in Pleurotus species; abstract absent but the title is unambiguous. | OUThigh empirical The title identifies an experimental study of fungal fatty acids, not research itself. | OUThigh empirical Title is fatty acid profiles of edible fungi; domain food/mycology chemistry. |
| Empirical Modeling of Butyl Acrylate/Vinyl Acetate/Acrylic Acid Emulsion‐Based Pressure‐Sensitive Adhesives 2004 · article · en · Macromolecular Materials and Engineering | aff_core w=5595 | 0/3 | OUThigh empirical Empirical modelling of polymer adhesive formulations; materials engineering. | OUThigh empirical This study models properties of pressure-sensitive adhesives, not research. | OUThigh empirical Polymer chemistry modelling of pressure-sensitive adhesives. |
| A Comparison of Methods of Adding Fibrolytic Enzymes to Lactating Cow Diets 2000 · article · en · Journal of Dairy Science | aff_core w=5595 | 0/3 | OUThigh empirical Feeding trial of fibrolytic enzymes in dairy cows; animal nutrition. | OUThigh empirical This experiment studies enzyme delivery in dairy cow diets rather than research methodology. | OUThigh empirical Animal-nutrition experiment adding fibrolytic enzymes to dairy cow diets. |
| Functional role of the endocannabinoid system and AMPA/kainate receptors in 5-HT2A receptor-mediated wet dog shakes 2005 · article · en · European Journal of Pharmacology | aff_core w=5595 | 0/3 | OUThigh empirical No abstract; pharmacology of endocannabinoid and receptor-mediated behavior in animals. | OUThigh empirical The study investigates neuropharmacological mechanisms of wet dog shakes. | OUTmedium empirical Neuropharmacology of endocannabinoid and AMPA systems; domain neuroscience (no abstract). |
| Analyse Expérimentale et Numérique du Comportement de Membranes Thermoplastiques en ABS et en HIPS dans le Procédé de Thermoformage 2008 · article · fr · The Canadian Journal of Chemical Engineering | french w=1554 | 0/3 | OUThigh empirical Experimental and finite element analysis of thermoplastic membranes in thermoforming; the object is material behavior. | OUThigh empirical The study models the thermoforming behavior of plastic membranes, not research practice. | OUThigh empirical Engineering characterization of thermoplastic membrane mechanical behaviour; domain research. |
| Use of a passive equilibration methodology to encapsulate cisplatin into preformed thermosensitive liposomes 2007 · article · en · International Journal of Pharmaceutics | aff_core w=5595 | 0/3 | OUThigh empirical Encapsulation of cisplatin into thermosensitive liposomes; pharmaceutics. | OUThigh empirical This pharmaceutical study develops a liposome drug-delivery method and does not study research itself. | OUTmedium empirical Pharmaceutics methods for encapsulating cisplatin in thermosensitive liposomes; drug-delivery science. |
| Controlled deposition of gold nanodots using non-contact atomic force microscopy 2005 · article · en · Nanotechnology | aff_core w=5595 | 0/3 | OUThigh empirical AFM-based deposition of gold nanodots; nanoscience technique. | OUThigh empirical This studies nanoscale gold-dot deposition using atomic force microscopy, not research practice. | OUThigh empirical Nanofabrication AFM technique; technical 'reproducible deposition' is assay-sense polysemy. |
| Converging Divergences in Formal and Informal Work: Longitudinal Evidence from Mexico 2015 · article · en · Global Labour Journal | venue_new w=2684 | 0/3 | OUThigh empirical Longitudinal study of formal and informal retail work in Mexico; the object is labor markets. | OUThigh empirical The paper studies formal and informal labor in Mexico, not research itself. | OUThigh empirical Labor sociology of formal and informal work in Mexico; not about research systems. |
| Private Sector Unions and Strategic Planning: A Research Agenda 2005 · article · en · Relations industrielles | venue_new w=2684 | 0/3 | OUTmedium conceptual Conceptual paper proposing hypotheses on union strategic planning; the 'research agenda' framing does not make research the object of study. | OUThigh conceptual The paper proposes strategic planning for private-sector unions rather than studying research practice. | OUThigh conceptual Labour-relations research agenda on private-sector unions’ strategic planning; object is unions, not research systems. |
| Lombalgies et réadaptation au travail 2005 · article · fr · Laennec | french w=1554 | 0/3 | OUThigh editorial/commentary Magazine feature on an interdisciplinary approach to low back pain and return-to-work rehabilitation; the object is clinical and occupational care. | OUThigh conceptual The work concerns clinical rehabilitation approaches for low back pain. | OUThigh editorial/commentary Magazine piece on low-back-pain work rehabilitation; clinical occupational health, not research about research. |
| Improving the Precision of Tree Counting by Combining Tree Detection with Crown Delineation and Classification on Homogeneity Guided Smoothed High Resolution (50 cm) Multispectral Airborne Digital Data 2012 · article · en · Remote Sensing | aff_core w=5595 | 0/3 | OUThigh empirical Remote sensing method for counting trees by species from airborne imagery. | OUThigh empirical It develops a tree-counting method for forest management, using a method to answer a domain question. | OUThigh empirical Remote-sensing tree-counting method answers a forestry inventory question. |
| Toughness Characterization of Fiber-Reinforced Concrete: Which Standard to Use? 2004 · article · en · Journal of Testing and Evaluation | aff_core w=5595 | 0/3 | OUTmedium empirical Comparison of concrete toughness test standards; measurement standards in materials engineering, and the 'reproducibility' language is assay-level, a polysemy trap. | OUThigh empirical This compares concrete testing standards and measurement results, not research methods as a social or scholarly practice. | OUThigh empirical Compares engineering materials-test standards for concrete toughness; object is construction materials testing, not research. |
| Detection of Multiple Potato Viruses in the Field Suggests Synergistic Interactions among Potato Viruses in Pakistan 2014 · article · en · The Plant Pathology Journal | fund_new w=1679 | 0/3 | OUThigh empirical Field survey of potato virus incidence and synergistic interactions in Pakistan. | OUThigh empirical It investigates potato virus infections and interactions, not research itself. | OUThigh empirical Plant pathology field survey of potato viruses and coinfection synergy. |
| Skew-resistant parallel in-memory spatial join 2014 · article · en | aff_core w=5595 | 0/3 | OUThigh empirical Computer-science paper on a skew-resistant parallel spatial join algorithm; the object is a database algorithm. | OUThigh empirical It develops a spatial-join algorithm for computing applications, not a study of research. | OUThigh empirical Parallel database algorithm for spatial joins; computer systems performance, not scholarly communication. |
| Résistance au changement de l’institution policière et criminalité évolutive : un paradoxe 2011 · article · fr · Champ pénal | french w=1554 | 0/3 | OUThigh conceptual Analysis of resistance to change in the police as an institution; organizational sociology of policing, not of research. | OUThigh conceptual It analyzes police institutional change and crime rather than research itself. | OUThigh conceptual Criminology analysis of police institutional resistance to reform, not science or research systems. |
What the screen actually found
Three frontier models (Opus 4.8, GPT-5.6 high, Grok 4.5) screened the same 1,000 works from the REAL 4.3M frame, on the rubric's FULL eight-field payload, with randomized manifest-logged chunks and harness-written labels: every defect D1, D2, D11 and finding 16 identified, repaired. Design-weighted base rates span 1.89% to 3.54% (1.9x). But the sets are the finding, as finding 16 predicted: of the 51 works ANY model called metaresearch, only 19 (37%) were called metaresearch by ALL THREE, and 24 (47%) rest on a single model's opinion. THE FIELD'S BOUNDARY IS NOT A LINE THE MODELS SHARE; IT IS A REGION THEY EACH CUT DIFFERENTLY. GPT-5.6 also violated the locked output schema on 18 of 1,000 records, writing genre values into the tier field, which the manifest validator caught. The deliverable is not the base rate: it is the disagreement dossier, the 51 works that mark the empirical boundary of the field and against which the inclusion criteria must actually be written.
Computed 2026-07-12T19:52:14Z · all 22 findings · this dossier as JSON