Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

First thought: "this looks like what we used at Uber."

Sure enough, it is. It works well IIRC and there's a lot of interesting math surrounding it. Some of our internal tools we had access to at the time (now, presumably, under lock and key) had maps laid out in hexagons.

E: Some of the visualizers linked here seem a bit weird. I could be mis-remembering things, but the hexagons were WAY smaller than some of the visualizers here. I remember looking at the SF map and there was a lot of granularity even in the city center. Like I said though, could be mis-remembering.



https://h3geo.org/docs/core-library/restable

The smallest granularity is . 0000009 km^2, so I guess 1/3 of an inch. Which seems to me a ridiculously and unnecessarily precise, but who knows


It wasn't that granular. Maybe a few blocks of the city in size, give or take. The visualizers I'm seeing have a single hexagon span the entirety of Austin, TX for reference. At least in Uber's usecase, that was less than useful.


It’s a hierarchy… so you can take your pick of granularity? Only thing that matters is min, max and step-size.

Unless we’re talking about different things, I’m not clear why we’re not assuming the tool you remember chose h10 or whatever level was actually useful to it, where the visualizations chose h5


I don't remember ours being a hierarchy but perhaps it was just the visualization.


Uber's internal stuff used resolution 9 i think...


That table shows H15 cells are on average 0.9 m^2, or a hexagon with edges 0.59m long.

(The average edge length is 0.5m, but that's not the same as the edge length of the average area hexagon)




Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: