In half the cities on the seed-oil-free map, every entry is a chain

A directory is only ever used locally. Nobody eats in the aggregate; they eat within about fifteen minutes of where they are. So the national headline — 5,887 venues reported to cook without seed oils — answers a question nobody asks. The question is what is on the map in your city.

We already counted this corpus once and found that four chains carry half of it. Reading it city by city says something sharper.

In 293 of 593 cities, there is no independent

Not few. None. In 49 percent of the cities with their own page, every single entry is a location of a chain you could have named without a directory. Widen it to “two or fewer independents” and you cover 499 of 593 cities — 84 percent.

This is not a flaw in anyone’s data collection. It is what happens structurally when a chain that has made one company-wide decision about frying oil contributes several hundred rows and a kitchen that made the same decision contributes one. The count measures locations; the interesting unit is decisions.

The independents cluster in about a dozen cities

The other half of the map is real, and it is concentrated. Ten cities hold a visible share of every independent in the corpus, and they are the cities you would guess — which is itself the finding, because it means the map thins out fast the moment you leave them.

What to do with this

If your city is on the map, check what kind of map it is. An all-chain city page tells you the chains near you have a national frying-oil policy. That is genuinely useful and it is also all it tells you — no local kitchen has been surveyed.

Absence is not evidence. A city with no independents listed is far more likely to be a place nobody has reported from than a place where nobody cooks in butter. A directory of venues that avoid seed oils cannot tell you anything about the venues it does not contain.

Ask, in the ~49% of cases where the map runs out. The four questions worth asking a restaurant are on every city page, and they work whether or not the place is listed.

Method

A city needs at least 3 venues to get its own page; the 2,445 venues in smaller places are rolled up to their state or country and are excluded from the per-city counts above, which would otherwise be flattered by a long list of one-venue towns. “Chain” is a flag in the source data, not a judgement about quality — several of the largest chains here made an earlier and more consequential fat decision than most independents did.

As with the first post, every entry reflects what a venue has been reported to cook with. Nobody inspected a kitchen, suppliers change, and a bought-in sauce can undo a tallow fryer. The map is a shortlist to call, not a verdict.