Newest first. Releases (YYYY-MM) and patches (YYYY-MM.n) cover all three datasets and are frozen under /data/v1/<version>/. Entries before the first release are the version history of the researcher census (launched 18 March 2026); “Census refresh” rows are the monthly database updates on the 11th, which feed the release around the 15th. Corrections between releases are noted here and applied in the next release.

DateVersionDatasetWhat changed
2026-11-052026-11.4AllRelease 2026-11.4: 913 published companies (1,470 active), 55,865 researchers (calibrated estimate 51,946), 14 shops; 6 pathway top lists.
Patch 2026-11.4: the public census follows the high-permanence boundary. Soil Carbon is no longer published (it held 42,911 of 95,329 listed researchers, mostly agricultural soil science); General CDR counts a paper only when it carries no other pathway tag and is now labelled Cross-cutting CDR (policy, MRV, markets, systems); a researcher is listed when at least one paper carries a public tag, with the public pathway as the one with most papers. Listed researchers drop to 55,865. Afforestation companies are no longer published (not high permanence); published companies drop to 913. Internal data is unchanged; releases 2026-11 to 2026-11.3 stay frozen with the old scope.
2026-11-042026-11.3AllRelease 2026-11.3: 969 published companies (1,470 active), 95,328 researchers (calibrated estimate 88,178), 14 shops; 7 pathway top lists.
Patch 2026-11.3: two lists per pathway. Most active (top100_, top50_europe_) now ranks by an activity score: every CDR paper since 2021 weighted by age with a half-life of three years (a 2026 paper 1.0, 2023 0.5), ties by paper count; most cited (top100_cited_, top50_europe_cited_) ranks by citations to those papers. Every row carries its rank in the other list (rank_by_citations, rank_by_activity). Preprints and their journal versions are merged (408 preprints marked, among them 2026 pairs the July run had not seen); one duplicate researcher record merged (Minggang Xu). The counted-papers files carry the ranks in all four lists.
2026-11-042026-11.2AllRelease 2026-11.2: 969 published companies (1,470 active), 95,329 researchers (calibrated estimate 88,179), 14 shops; 7 pathway top lists.
Patch 2026-11.2: new files top100__papers (CSV and JSON), one row per person in the top lists and per paper that counted for the ranking (rank, paper id, year, title, citations, pathway tags, link), so every rank can be checked against the papers. The top-list pages link each paper count to the counted papers on the profile page.
2026-11-042026-11.1AllRelease 2026-11.1: 969 published companies (1,470 active), 95,329 researchers (calibrated estimate 88,179), 14 shops; 7 pathway top lists.
Patch 2026-11.1: institutions in the researcher top lists and profile pages now come from the papers’ own affiliations 2024-2026 (OpenAlex authorships; the affiliation text when OpenAlex has no institution match) instead of the author profile, which mixed in mis-attributed papers. 157 of 631 top-list institutions changed (Planavsky: Yale University, not UC Riverside; Stubbs: University of Glasgow, not Royal Bank of Scotland). The institution_source column says works, works_raw, orcid or openalex. Company and shop files are rebuilt from the same database.
2026-11-042026-11AllRelease 2026-11: 969 published companies (1,470 active), 95,329 researchers (calibrated estimate 88,179), 14 shops; 7 pathway top lists.
Public researcher data reduced to aggregates plus the top lists per pathway: profile pages only for the roughly 630 researchers in those lists, everyone else counted in the aggregates.
Per-company headcounts and follower numbers withdrawn (licensed data); EU SME size classes take their place.
2026-10-11Census refreshResearchersMonthly database refresh: 1,779 new papers discovered and classified, 2,669 new researchers (92,660 → 95,329); author metrics re-enriched.
2026-09-11Census refreshResearchersMonthly database refresh: 1,348 new papers discovered and classified, 2,091 new researchers (90,569 → 92,660); author metrics re-enriched.
2026-08-11Census refreshResearchersMonthly database refresh: 1,758 new papers discovered and classified, 1,861 new researchers (88,708 → 90,569); author metrics re-enriched.
2026-07-11Census v3.4ResearchersMembership tightened: researchers without a single classified-relevant CDR paper are no longer counted or exported. Researchers 147,377 → 88,691.
2026-07-11Census v3.3ResearchersDeterministic pathway tie-break: papers tagged with a specific pathway and General CDR assign the specific one. DAC 1,251 → 4,122 researchers; General CDR 18,066 → 11,123.
2026-07-11Census v3.2ResearchersTier-2 deduplication: 817 same-person author records merged on identical normalised names plus shared co-authors or institution, with common-name and ORCID-conflict guards. Researchers 148,194 → 147,377.
2026-07-11Census v3.1ResearchersDeduplication and primary-works scope: 561 ORCID-split author records merged, 1,441 upstream-deleted IDs retired, 6,959 historical merges repaired; datasets, supplements, peer-review comments and errata no longer count as papers (−2,002). Researchers 150,203 → 148,194; papers 27,816 → 25,814.
2026-07-09Census v3ResearchersMonthly refresh pipeline live. 8,328 new papers discovered since April, 3,067 classified CDR-relevant; all author metrics re-enriched; ORCID enrichment for 10,930 new researchers. Researchers 122,674 → 150,203; relevant papers 24,749 → 27,816.
2026-03-23Census v2.1ResearchersDirect air capture classifier recalibrated: 1,312 DAC papers recovered (30 → 1,342 papers, 48 → 3,951 authors).
2026-03-23Census v2ResearchersEvery paper reclassified by title and abstract with a language model; 12,384 non-CDR papers removed (39,278 → 24,749). 11,237 authors left without a CDR paper.
2026-03-18Census v1ResearchersInitial launch of the CDR Researcher Census: 122,674 authors, 39,278 papers, keyword-based pathway classification.