The data field guide

Behind the atlas.

Real country shapes. Real population research. Here are the sources, the credits, and how we turn them into a game.

The people behind the numbers.

CC BY 4.0

Our population weights come from WorldPop, School of Geography and Environmental Science, University of Southampton: Global2, release R2025A v1, using the 2025 constrained, UN-adjusted country population rasters.

The original grid is 30 arc-seconds, approximately 1 km at the equator. WorldPop combines census and other demographic sources with geographic modeling, and aligns national totals with UN population estimates. These are modeled estimates for 2025, with uncertainty that varies between countries and places. They do not describe a live population count or exact household locations.

The data is available under the Creative Commons Attribution 4.0 International license (CC BY 4.0). Slice Atlas adapts it by projecting, aggregating and rounding the grid as described below.

Credit: WorldPop, University of Southampton. Global2 constrained population estimates, R2025A v1 (2025), 30 arc-second resolution. Bondarenko M. and colleagues. DOI: 10.5258/SOTON/WP00840.

How the game adapts the data.

We place each country and its population cells on the same local equal-area map projection. We then sum the original population cells into a 128 × 128 scoring grid. Each occupied bin uses a population-weighted location. Rounding preserves the original national population estimate to the nearest person.

Across the twelve countries, 12,293,085 positive source cells become 41,620 weighted bins. A cut assigns each bin's population to the sector containing its weighted location. This keeps scoring fast and identical on phones, tablets, computers and the server.

This aggregation is an approximation: a line crossing a bin can produce a different split from one calculated against every original 1 km cell. The source data is also a model, so a high game score measures balance within this catalogue rather than exact real-world population boundaries.

The rendered outlines are simplified for clarity. WorldPop and Natural Earth maintain their borders independently, so some coastal or border cells can differ slightly from the displayed outline. We retain population weights to preserve each country's total.

The shapes of the world.

PUBLIC DOMAIN

Country outlines are adapted from Natural Earth version 5.1.2, the 1:50 million Admin 0 countries dataset. They retain islands and interior rings, with a local projection and modest simplification for the game.

Made with Natural Earth. Its raster and vector map data are in the public domain. Displayed boundaries follow that source's cartographic choices.

One shared atlas. One fixed revision.

A daily game and its friend challenges use identical country data and scoring rules. This catalogue stays fixed after publication. Changes receive a new revision so a shared challenge keeps its original meaning.

Geographic catalogueworldpop2025-r2025a-ne5.1.2-grid128-v1
Data and scoring rulesworldpop2025-r2025a-ne5.1.2-grid128-v1-sectors-v1
Countries and reproducibility details

Türkiye, Germany, United Kingdom, Italy, Japan, South Korea, Egypt, South Africa, Brazil, Australia, Mexico and India. Our provenance manifest records each original raster URL, source resolution, source population total, projection center and SHA-256 file digest. The importer rejects changed source files or output under the same revision.