Now imagine how bad it is outside of the SF bubble. I searched for coffee shops and the closest result to me is 7 hours drive away.

Coffee shops gets me 3 results in my city, coffee shop gets me 4 results with the closest 45 miles away. Totally useless.

I don't really understand how it's this bad when OSM data alone can get you much better results.

There are only two coffee shops in Wellington (NZ) according to DDG/AppleMaps.

https://duckduckgo.com/?q=coffee+shop&t=h_&ia=recipes&iaxm=m...

Edit: searching for "cafe" gives better results but still not great

https://duckduckgo.com/?q=cafe&t=h_&ia=recipes&iaxm=places&s...

I just tried "coffee shop" and "coffee shops" near my location in eastern Pennsylvania (semi-rural) and the results are atrocious. "Supermarket" shows all the possible options so it's not all bad.

It's a fun example, imagine how garbage search in languages that have more than one case is is. :/

"coffee shop" on the other hand yields 20 results. I agree that some support for fuzzy search is needed, but there is a reasonable amount of data there.

I hate when searches parse like that. Ideally coffee shop, coffee shops, coffee, cafe, espresso, etc. should all give me the same exact results: 100% of the stores in the area that sell coffee.

I don't think there's a single piece of mapping software that doesn't suck hard in some way. It's pretty annoying how google maps shows you x number of results zoomed out, y zoomed in, and z with the map frame moved half a block to the left.

Just show me everything. Search the entire city. Flood my map. Let me do the vetting, that's what I came to maps to do anyway.

where the sets of y and z might not even overlap

This shows that they are probably not even doing stemming which shouldn't be that hard to implement and could have great benefit/cost

Weirdly, "coffee shops in san francisco" shows a lot.