Chicagoland Third Places

"Third places" is Ray Oldenburg's term for the informal public spots people gather outside of home and work: cafes, libraries, community centers, social facilities. This maps every one of them I could find on OpenStreetMap across 319 towns and cities in the six-county Chicagoland region, and compares them per capita instead of by raw count, since a raw count just tells you which town is biggest.

The angle I started with didn't hold up

I originally set out to build something bigger: pairing this with an analysis of how thoroughly each area has actually been mapped on OpenStreetMap, on the assumption that wealthier areas would be more completely mapped and worse-off areas would show up as under-mapped. The real numbers said the opposite of what I expected. Population density predicted mapping completeness far better than income did, and when I checked the same question at a different level of geography (towns instead of census tracts), the pattern flipped entirely. That's a real, documented problem in spatial statistics called the Modifiable Areal Unit Problem. I decided that was a more honest finding to write up than to build around, so I cut it from the interactive page and put my effort into making the third-places map itself as useful as I could.

Built to actually be used, not just looked at

Past the map itself, there's a ZIP code search so you can jump straight to your own area, and a second map showing all 2,645 individual locations plotted where they really are, not just aggregated by town. Both maps distinguish three categories: cafes, social facilities, and community/civic spaces (libraries and community centers combined).

What actually took the most work

Not the mapping logic, the click reliability. Individual map pins would intermittently fail to open their popup, worse the more you clicked, in a way that didn't show up in any browser console error. It turned out Leaflet's popups automatically nudge the map to keep themselves visible by default, which silently shifts every other pin's on-screen position by a few pixels each time you open one. I only found it by scripting a headless browser to click through the page automatically and compare it against my own manual testing, since a script clicks the exact same pixel every time and a real hand doesn't.