On Mon, Jan 7, 2013 at 11:40 PM, Jeff Meyer <j...@gwhat.org> wrote: > All of the rules about observability and verifiability apply to country and > state borders, as well, as Mike states, but we include them and somehow > improve them.
Have we improved them? Being the last user to touch about 35% of all county boundary relations in the US, I've seen a lot of breaking too. One exception: some of the Oregon counties that were improved by Trimet with local data. There were a lot of untouched ways from the original import which were overlapping/duplicated closed ways. A lot of them had been operated on by a duplicate node removal bot that joined nodes with TIGER roads wherever the two intersected which made them a pain to edit. A number of them that had been converted to relations had either broken geometry or inconsistent tagging - and still do. I've been meaning to bring this up at some point. The entire western border of Colorado was once moved a few hundred meters east by accident without the user even noticing that they had touched a state border. I think it would be great to make more tools support more external data sets as opposed to dumping *everything* into OSM. You want county borders on your garmin? Check a box while creating the file and mkgmap downloads the most recent county borders from some source that isn't OSM and includes them. Now, building this functionality into every tool that uses OSM data may not be practical. But I can definitely see a place for a parallel project that hosts all such boundary data (maybe even parcel data) from official sources in a common format and can be easily mixed with OSM data before being fed to existing tools. I think this was the idea behind CommonMap although I see this particular implementation hasn't fared particularly well as the domain seems to have expired... But the idea may warrant another look. Toby _______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-us