Every data and image source this site uses, the licence it is served under, and the credit each one asks for.
Last updated
This site is built almost entirely on other people's work, published under licences that allow it. This page names all of it. If you use our dataset, these are the credits you inherit along with it.
| Source | What we use it for | Licence |
|---|---|---|
| GeoNames | The place tree: countries, divisions to four levels, cities, names, local names, coordinates, population, and the stable id everything is keyed to | CC BY 4.0 |
| Wikidata | The independent cross-check on every capital relation, Wikidata ids, and the image chosen on a place's item | CC0 1.0 |
| Natural Earth | Country and first-level boundaries, and which countries share a land border | Public domain |
| geoBoundaries | First- and second-level boundaries where Natural Earth has no match | Varies by file, recorded per file; only attribution-only licences are used |
| OpenStreetMap | The basemap under the interactive map, served from a Protomaps archive | ODbL 1.0 |
| Wikimedia Commons | Photographs of places | Per file: public domain, CC0 or CC BY only |
| Wikipedia | The cited reference behind each country's curated flag colours | CC BY-SA 4.0 |
| Wikivoyage | Listed as a future source of banner photographs and of place descriptions. Not in use today | CC BY-SA 4.0 |
Two rows in that table need a word of explanation.
We read Wikipedia to decide the two or three colours in each country's flag, and we record the article we read. We do not republish any Wikipedia text, so nothing on this site is a derivative of it and its share-alike condition does not reach us. If that changes, and it will when place descriptions ship, the descriptions will carry their own source line and licence.
Wikivoyage is in the photograph pipeline's source order but no fetcher for it has been written. It is listed here so the list matches the plan rather than only today's code, and because its banner images are mostly share-alike, which our photograph rule below excludes.
Attribution under CC BY 4.0, which is what GeoNames asks, means naming the source, linking the licence, and saying that changes were made. Every page on this site names its sources in the footer and links here; every dataset download will carry the same notice in a file beside it.
OpenStreetMap asks for "(c) OpenStreetMap contributors", and the interactive map prints it whenever the basemap archive is in use.
geoBoundaries files carry the credit their upstream source requires, which differs per country: a government open licence often specifies wording that has to be passed on word for word. Every boundary file we publish carries that wording inside it, and every map that draws a boundary prints the credit for the file it drew.
A photograph on this site is only ever public domain, CC0, or CC BY at any version from 1.0 to 4.0. Every other licence is refused, including CC BY-SA at any version, the GNU Free Documentation Licence, the Free Art Licence, and anything with a non-commercial or no-derivatives condition.
The reason is mechanical rather than ideological. A picture on a page here is cropped to the page's shape, scaled to five widths and re-encoded as WebP. That makes a derivative work. A share-alike licence requires the derivative to be released under the same terms, which would pull the page around the photograph, and eventually the dataset, under a licence we have not chosen and cannot offer to the people who reuse our data. So we would rather show fewer pictures.
This costs more than it sounds like it should. On the first twelve places the pipeline looked at, 128 of 144 candidate photographs were rejected on licence alone. Abuja has no photograph at all for this reason. The decision is worth revisiting, and the obvious way to revisit it is a separate place on the page where a share-alike photograph is shown whole, uncropped, with its licence named beside it.
Where a photograph does appear, it carries the photographer's name and its licence on the image itself, linked to the file page. The database will not accept a photograph row without an author and a licence URL, so a picture cannot render here without a credit.
Photographs uploaded by people with accounts are published under CC BY-SA 4.0, which is the licence the upload form states. Those are shown whole and are not the page's main image, so the crop problem above does not arise.
Listed because the absence is deliberate, and because someone reusing the dataset should know why a country's boundaries are missing.
Share-alike boundary data is excluded for the same reason as share-alike photographs: ODbL and CC BY-SA files cannot be redistributed under CC BY 4.0. That removes Sweden's municipal boundaries and several national datasets.
Anything with a non-commercial or no-derivatives condition is excluded. Eurostat's GISCO data forbids commercial use, so the geoBoundaries files derived from it are excluded whatever licence the file itself declares. Spain's INE, the Singapore Open Data Licence and Statistics Canada's open licence are excluded because each adds a term a publisher would have to pass on, and CC BY 4.0 does not let us pass one on.
Anything whose licence we cannot read is excluded. Matching is exact, so a spelling the code has not been told about fails closed rather than being guessed at.
The reference dataset, including our own additions, will be published as a nightly download under CC BY 4.0, with attribution to All The Capitals and to GeoNames. Everything above is why that is possible.
Text and pictures contributed by people with accounts are published under CC BY-SA 4.0 and credited to the display name on the account. The terms set out the licence a contributor grants.
If you think something here is credited wrongly, or not credited at all, the contact page is the place to say so, and the takedown notice is the formal route for a copyright complaint.