Geography Nerd

I work with people all over the world and like knowing where places actually sit in relation to each other. Is Boston north or south of Toronto? (South, but barely.) Is New York north or south of Marseille? (South, by a lot more than you would expect.)

The bet

I worked those two out years ago and was genuinely surprised, and the game exists because of a hunch: most people do not realize how far north Europe sits. Marseille is on the Mediterranean, it feels southern, and it is still well north of New York. That is the whole idea. Two cities, four arrows, and a mental map that turns out to be wrong more often than you would like.

The dataset is the top thousand cities or so from GeoNames. There are actually 1,045 in there and I no longer remember why.

Difficulty is distance

The one piece of design worth describing is how levels work. Difficulty is not about obscure cities, it is about how close together the two cities are along the axis you are being asked about. Early rounds put them far apart in the compared direction, so the answer is available to anyone. Later rounds squeeze them together until you actually have to know.

Country context leaks answers, which is unavoidable and mostly fine. If you know roughly where two countries are you can often skip the thinking. Cities within China are reliably the hardest, because a lot of people carry no internal map of the country at all.

The comparison itself is straightforward latitude and longitude, with east and west resolved by which half of the world each city falls in. After each guess an OpenStreetMap view, rendered through React Leaflet over CARTO tiles, drops a pin on both cities so you get the visual confirmation of where they actually sit. Seeing it is the point. Being told you were wrong teaches nothing.

What I took out

This started as a private repository over Thanksgiving weekend and it used to be a more serious thing. It had user accounts and stored high scores. Then I decided I was less interested in building a real game than in putting something fun somewhere people could play it, so I removed all of that and open sourced what was left.

What remains is free, has no accounts, no install, no analytics, and nothing to promote. It costs nothing to host. That is the entire business model, and for a project like this it is the right one: anything I added to make money would cost more attention than the game is worth.

There is one leftover in the repository worth admitting to. A data/ directory holds SQL files from when the geographic calculations were going to run in Postgres. They do not, because browser libraries handle it fine, so those files are a fossil of a plan that got simpler.

One last note on how it looks from the outside: the security policy, code of conduct, and contributing guide were written by Claude, which suggested them unprompted. I kept them. I want public projects to look like someone is behind them, and that is cheap to do properly.