Where the data is from

Every puzzle uses real population data, not made-up numbers. The population figures come from WorldPop, a research initiative that produces gridded population estimates for the entire world. WorldPop combines official census and survey data with satellite imagery, things like nighttime lights, land cover, and settlement patterns, and machine learning to model how population is likely distributed within smaller areas, since real censuses are only reported at a broader administrative level, not down to individual towns or city blocks. This means rural areas count too, not just named towns and cities, so splitting a region "evenly" is really about splitting where people actually live.

It's a modelled estimate, not a headcount, so it's worth being upfront that it isn't perfect. Accuracy can vary in remote areas, places with outdated or unreliable census data, or regions where population has shifted quickly since the underlying data was collected. It's the best openly available option for a game like this, but treat it as a very good estimate rather than an exact figure.

Country and coastline outlines come from Natural Earth, a standard, publicly available world boundaries dataset. Place names (the cities and towns you might recognize on the map) come from a separate open dataset of populated places. For places with disputed or contested borders, GeoSlice follows whatever that dataset shows rather than picking a side, this isn't a political statement of any kind, just the most practical way to render consistent maps everywhere in the world.

Each day's puzzles are generated from this data using a seed based on the date, so everyone playing the same day sees the exact same three places, in the same order, no matter where they are or when they load the page that day.

Your score compares the split you draw against the real population split in that exact spot, calculated directly from the underlying data - not an approximation.