I fully support the OSM effort. Interestingly, through OSM and even Apple Maps you see just how difficult it is to build a mapping dataset. One story I heard w

I fully support the OSM effort. Interestingly, through OSM and even Apple Maps you see just how difficult it is to build a mapping dataset. One story I heard was that the Chinese government intentionally makes its published maps inaccurate by basically offsetting it by a few meters. I'm not sure if this is still current. I'm not sure of the reasoning. It's not like it would stop an invasion if the highway was a few meters from where you thoguht it was. Anyway, to counter that humans basically had to adjust the published maps by overlaying them onto satellite photos to remove the error. AI helps with this but it's still labor intensive. IN addition to being labor-intensive, there are so many disparate data sources that you need to match up. Much of that data conflicts. The likes of Yelp and even Foursquare exist primarily by selling location data. Even something like listing the opening hours of a business is incredibly labor intensive. Google has (had?) a system where an AI would call the business to ask their hours of operation. I think they had to modify it to say it was an automated call. Oh you want directions? I guess you need accurate direction data for roads. You probably need to know where bike paths are. Different locales can define a bike path differently. In some places bikes can share the road with cars. In others they can't. You want to integrate public transit? Well, every place is going to have a weird API. It's a good example because something that seems so simple involves a ton of labor and integrating thousands of data sources. Disclaimer: Xoogler.