Coverage
What the catalog contains and, more usefully, what it is missing. Everything here is computed from the corpus at build time.
The number that matters most, and the one we are worst at. Almost everything here was machine-imported and has never been read by a person.
0.5% of the catalog has been confirmed by a human. Verifying one record is the single most useful thing anyone can do here.
The thin categories are not thin because the work does not exist. Open hardware, manufacturing, and clinical guidelines mostly do not live in Git repositories, and repository search is where most of this corpus came from — so the shape of this chart is partly a picture of our own blind spot.
311 entries carry a licence we could not resolve to an SPDX identifier. Those are recorded as NOASSERTION or LicenseRef-* and are never claimed as OSI-approved.
All of this is in /v1/meta.json if you would rather compute your own.