The UX could be a bit more intuitive, and some of the questions could be better, but these are the outcome of a two-line prompt (!). One can always ask Claude to fine tune both the content and the interface.

And here is one for open-source software facts, using information from, inter alia, github and package managers.
claude.ai/public/artif...

Sample prompt: "Can you make a credence calibration game based on engineering projects? Use Baselight and pick questions about cost, time to build, weight, manpower, height, etc."
claude.ai/public/artif...

Another way to read this chart is "how many people can reach city X by train in less than 2h". But Brussels is on the list and no one in their right mind would travel in that direction, so I'll stick to the alternative interpretation.

Want to draw your own insights? Check the data for your country? Use a different metric? You can now fork and modify Baselight queries. Start here: baselight.app/u/jsoares/qu...

And then there's Paris. It's a curious one, as the only listed major metropolis with a reachability ratio of less than 1, i.e. the surrounding reachable population is smaller than that of the city itself. Lower ratios are expected for larger cities, but compare London.

How do we measure this? Turns out @ec.europa.eu collected data on the time it took to travel between rail-connected city pairs, and the data is on @baselightdb.bsky.social. We're restricting our ranking to those reachable in under 2h, as it's just a weekend after all.

Also, the usual shameless plug: @baselightdb.bsky.social is the place for all your data needs!

baselight.app/u/jsoares/qu...

We were all in different places at the time, and only managed to establish contact with everyone 6 hours later. Seriously considering getting everyone Thuraya satphones or Starlink Roam. Any suggestions, given sporadic emergency use only?

Told you this would happen!