Each one is a 3D model built from public records: roads, buildings, terrain, power lines, water systems, hospitals and bridges. We use them to try ideas before anyone spends money on them.
Each was chosen by a written rule to represent a different kind of rural America. Open one to fly over it, switch data layers on and off, and see where every layer came from.
Ten ideas, each run 20 times on simulated households, each judged against a success line written down before it ran. Failures are reported as plainly as wins.
A companion table covers 3,163 U.S. counties with population history, income, poverty and hospital counts, built from Census, USDA and CMS data. 719 counties have no hospital at all; 477 of those are rural.
These are screening results, not findings. The simulations are simplified models with assumed equipment numbers. They are good for deciding which ideas deserve a closer look, and not for making claims about real people.
Every household in the experiments is synthetic. No real person's records are used. Political and election data are excluded from every model.
Some layers are incomplete in ways we report rather than patch. OpenStreetMap misses many rural buildings, and federal water-system records have no locations, so some systems are counted but not placed on the map.