How the score works
The number on the card is the catalogue score. Rank is that number, high to low. It is a data index of Canadian artists we can show, not a judgment of talent and not a license to print.
What the engine measures
A separate engine can propose a score from fields we actually store. Missing signals are skipped. They are not filled with 50.
| Signal | Weight | Evidence |
|---|---|---|
| Gallery | 22% | Matched roster and gallery type. A name we cannot match is weak, not elite. |
| Market | 22% | Midpoint of a real price range. No invented $1,200. No auctions. |
| Social | 15% | Website and/or Instagram URL present. Not follower vanity. |
| Record | 18% | Verified flag plus filled city, province, medium, genre, site, artwork, portrait. |
| Exhibition | 10% | A count or exhibition text. Silence is omitted, not a zero career. |
| Legacy | 13% | Institutional text: museums, RCA, co-ops, biennales. This exists so no-Instagram masters are not buried. |
Confidence is how much of that weight was present. Low confidence is labeled on the card as an incomplete profile. It does not secretly pull the number toward 50.
Catalogue vs proposal
The live rank uses the CSV. The engine writes a proposal file. Applying it is a deliberate step, with a backup. Same engine for every artist. The formula will keep evolving as the record gets thicker. Thin records cannot sit at the top.
Bands on the card describe the record, not the career: dense, strong, partial, sparse.
How it gets updated
Recompute proposal, read the diff, apply to CSV only when we mean it, then rebuild the site. Nightshift does not silently rewrite rank.