Osm has the lead in the majority of places. I used a random number generator to generate coordinates (latitude between the southernmost and northernmost point of land, I forgot the numbers, and -180 to +180 degrees longitude). Any point that was not water, I'd zoom in to about street level by default and zoom out until I saw a reasonable amount of features on at least one of the maps. The maps I compared were Google, OpenStreetMap, Bing, and TomTom. I wanted to also compare Apple but their maps are only for customers it seems. To determine the best map, I considered mainly which had more features (a more complete map), but between near-ties I also looked at satellite imagery to determine which one was more accurate.
OpenStreetMap beats everything else.
I still want to write this up in a blog post with visual examples, and I'd like to redo the test (for comparison) with random coordinates biased by population centres (does anyone know how to generate those?). I'm curious if that paints a different picture. But purely geographically, this de facto standard isn't so great if you look beyond the western world.
Interesting! Yep, a lot depends on the distribution you use for your random sample.
But also, any attempt at objective weighting like this is going to pick a single winner, but it's not the case that everyone is better off using the same mapping service, since you mostly want coverage for the places you travel to and that's different for each person.
Right now, that's true: some maps are better in certain regions than others. But I don't think it's a given that "it's not the case that everyone is better off using the same mapping service". Being an OpenStreetMap contributor, I like to think that one day, the map might be complete enough that this is the common basis for all mapping services. The data is free to use, as long as you're willing to give credit and release any additions under the same license.
Hosting is not free and even open source apps like OsmAnd limit the number of downloads unless you buy the app. It's also still a big challenge to present the data nicely and make a nice user interface. There will definitely still be opportunities for competition and profit, even if all data is open. So hopefully, one day, everyone is better off by having a single, complete map available to use for anyone :)
OpenStreetMap beats everything else.
I still want to write this up in a blog post with visual examples, and I'd like to redo the test (for comparison) with random coordinates biased by population centres (does anyone know how to generate those?). I'm curious if that paints a different picture. But purely geographically, this de facto standard isn't so great if you look beyond the western world.