Good news: I’ve found much better data in TIGER 2014, which could be used as the base for cleaning this all up considerably.

(In general, access restrictions are pretty tricky around here: there are lots of different trail owners, and they all have their own policies. And I've seen cases where the sign at one of the trail allows bikes, and the sign at the other end forbids them. I don't think it was supposed to be a one-way restriction!)

I'm getting access restrictions from the posted signs, which I always try to photograph.

Checking osm.wiki/OSM_tags_for_routing/Access-Restrictions#United_States_of_America says that highway=footway defaults to bicycle=no and horse=no. So theoretically, I just need to add ski=yes to the trails, and they should be accurately tagged, assuming the wiki is to be trusted. :-)

Many thanks for your advice!

Thank you!

I know that highway=footway says: "For designated footpaths, i.e. mainly/exclusively for pedestrians." (If it had been an obviously multiuse path, I would have chosen highway=path.) Is it important to add restrictions like bicycle=no to highway=path, or is it implied?

I think these paths are actually foot=yes, ski=yes, and everything else =no. What access tags would you use in this case?

@mapsinE3: Thank you for the pointer! There was only one duplicate node in my area of the map, and that was left over from an old import. It's fixed, now.

@Nightdive: Thank you for the tip! My Garmin showed considerably less consistency between multiple tracks in the same area than the Nexus One. I bought the Garmin off a clearance rack several years ago, and it typically claimed an accuracy of 20 ft under good conditions—and repeatedly walking the same track tended to show errors of at least that size.

The Nexus One, on the other hand, seems to have no more than half the error of the Garmin under reasonably favorable conditions. It makes some sense: It's a much newer GPS chipset, and it has a good reputation with the geocachers who've tried it. So it's _possible_ that I might be seeing an actual improvement. It's also possible that I'm imagining things. :-)

I'll walk some tracks repeatedly and superimpose, and see what I get.

@mapsinE3: Thank you for the pointer! There was only one duplicate node in my area of the map, and that was left over from an old import. It's fixed, now.

@Nightdive: Thank you for the tip! My Garmin showed considerably less consistency between multiple tracks in the same area than the Nexus One. I bought the Garmin off a clearance rack several years ago, and it typically claimed an accuracy of 20 ft under good conditions—and repeatedly walking the same track tended to show errors of at least that size.

The Nexus One, on the other hand, seems to have no more than half the error of the Garmin under reasonably favorable conditions. It makes some sense: It's a much newer GPS chipset, and it has a good reputation with the geocachers who've tried it. So it's _possible_ that I might be seeing an actual improvement. It's also possible that I'm imagining things. :-)

I'll walk some tracks repeatedly and superimpose, and see what I get.