A pinned OpenAlex snapshot · all 482 partitions · 2000–2025
Don’t search for metaresearch. Search for Canada, then screen.
The usual design retrieves what looks like metaresearch, then asks whether it is Canadian. That makes the field boundary a property of your keyword list, and it is why such maps cannot be audited: you cannot measure what a lexicon never showed you. This frame is inverted. It is all Canadian research — an external, checkable criterion — so field membership becomes a classification over an enumerable universe, not a retrieval over the literature.
The boundary is not a line. It is a region.
Three frontier models screened the same 1,000 works against the same locked rubric, on the full eight-field payload the rubric always specified. Of the works any model called metaresearch, only 37% were called metaresearch by all three, and 47% 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.
So the deliverable is not a base rate. It is the disagreement dossier: the works that mark the empirical boundary of the field, and against which the inclusion criteria have to be written.
A boolean over a four-state space
OpenAlex records retraction as is_retracted, a boolean. The post-publication record has at least four states. Joined to Retraction Watch, this frame carries 1,052 works with a recorded notice, of which 52 are expressions of concern — a state OpenAlex has no field for at all, and silently reports as false, which reads as “fine”.