How to deal with routes that consist (mostly or entirely) of a mix of existing paths/ways in OSM that are infrequently signposted

Forking from another discussion:

I noticed something similar on Gaia GPS (which is a US centric product that incorporates OSM as well as other sources) near Moab UT. While there aren’t any nodes in OSM, they have pulled in some “overloading” information recently from other sources and I assume that this is where the following set of “route nodes” comes from:

While on the Gaia GPS example above the markers are just put down on set distances, I could see some way of working this into guideposts. at junctions, though some of those would have to be abstracted.

I’m not aware of any path/way that has more than two routes on it, so having labels on a route could get noisy from a visual standpoint in an area which has a lot of different ones. The cleanest visual way would be to have routes alongside the ways/paths ala Acadia National Park.

updated this original post to be more focused on the specific issue.

Benelux countries (and Germany, Croatia…) do something similar with their fietsknooppuntennetwerk / Knotenpunktbezogene Wegweisung numbered-node cycle networks, sometimes informally called “bike by number,” where the numbers represent a sequence of intersections (represented by numbered nodes) the cyclist must pass through. The system has won numerous awards, including one for functional design in 2013. These “routes” are sometimes known as “node networks” to distinguish them from “traditional” routes that consist of a collection of ways in a type=route relation.

Extending this to “largely pathless trails” (as these are described in a rural / wilderness environment) would be revolutionary. Even starting with the success (and simplicity) of the European system, it seems some additional development would be required for this sort of hiking, as especially if trails are faint or non-existent, orienteering skills much higher than “a cyclist must stay on a paved road until the next sign with the desired numbered destination” would be required. But, you may be onto something big here.

OSM has wiki for these, so does wikipedia, I’ll let you discover this rich world and see how it might be applied to trails / paths in OSM (and guideposts and much else) that would need to be incorporated into a “new system for routing hiking” as you describe. Exciting!

I don’t think a numbered node_network really works as well here - it’s not going from a nameless 4-62-89-24, but following this road, then that road, then a path here and a path there that (for the most part) are already named.

A good real world example of this would be trails in the Sierra Nevada which are part of the JMT or PCT.

In this segment, the name is “John Muir Trail”, with an alt_name of “Pacific Crest Trail” and a ref of PCT. That’s a bit messy, but does indicate the JMT came first and then the PCT is often (but not always) routed on top of it.

There are relations to routes which are the “JMT, PCT, and North Lake South Lake Loop”. The first two are real, official routes, the latter is a shorthand for a popular 4-5 night trip between the two named lakes - there are no signs for it, no official recognition of it, but it’s something instantly recognizable (along with the Rae Lakes Loop on the westside) to anyone that backpacks the range.

I think just having a route related to various ways and paths that has some signposts of itself would be appropriate and even more subject to ground truth / verifiability.

There could be some tag, not trailblazed, but something closer to guidepost which indicates that not every junction has a named signpost to guide people. relation:route doesn’t have anything regarding the amount of guideposts or signage, which seems like it could be appropriate.

@SomeoneElse it seems to me that if this route has some signs then it’s relatively “real”, even if it’s lacking signage at every junction that would make it easy to follow. Should there be something like route_signage_visibility, or route:visibility?

Amusingly on the PCT especially, most hikers have no idea where they actually are or what pass they just went over due to the length of the route, so whenever a trail marker at a junction doesn’t explicitly mention the PCT someone will scratch it onto the sign with an arrow indicating which way to go. :stuck_out_tongue:

Yes, it sounds to me like the existing route relations e.g. for hiking, cycling, etc. are a good fit for this.

What we are missing is a tag that we can put on such a route relation to say whether and how well it is signposted. The Wiki page pretends that all such routes that are on OSM are signposted, but this is not true. This leads to the same questions of verifiability and copyright that come up in the discussion about pathless hiking. If we assume for a moment that we have a route where this has been sorted out (e.g. it’s a very popular route that has been published and the publisher has allowed us to use it, in an area where the local community is fine with having it in OSM), what tag do we use to say it’s not signposted? I can also think of cycling routes that are signposted but so poorly that you constantly have to check a map to make sure you’re still on it.

This question is essentially independent of the questions that the tags trailblazed and trail_visibility deal with. It’s about knowing which path belongs to the route, not about knowing whether you’re still on the path. You could have a signposted hiking route that consists of a series of overgrown, poorly visible paths, or you could have a cycling route that only exists on the website of the charity that created it, but all the roads and paths that are part of it are clearly visible.

It could also be useful to record that a route doesn’t have an organisation that maintains it. The Wiki suggests using operator on routes… it seems natural to use something like informal=yes or operator:type=none when that doesn’t exist, so I’m surprised these options haven’t been used more often.
https://wiki.openstreetmap.org/wiki/Key:operator#When_there_is_no_operator

1 Like

Hey, “talking about it more” in a venue like this forum is a good indication we (OSM) are going in the right direction with this.

(So,) keep talking, everybody!

I’ve used name:signed=no on completely unsigned routes, but that obviously doesn’t apply here. I’ll probably make up some tag at some point in the future.

Yeah, I’ve been on those myself, and occasionally “made my own route” due to lack of signage heh.

It is independent, but I think a similar excellent, good, intermediate, bad, horrible, no could work as as inspiration. I could see dropping some of those values - at times it’s hard to create strong distinctions between them that are meaningful and unambiguous and there’s less variance here than with trail_visibility. Off the top of my head for values of a new route visibility, or route:guideposted perhaps.

excellent = every junction for the route is signed, possibly occasional signs along the route in between junctions to let you know you are still on it.
intermediate = most junctions are signed, but sometimes you need to check if you are still on the route
bad = only a few junctions are signed, a map needs to be referred to often
no = route is informal or unsigned and requires constant checking at junctions.

If name:signed seems appropriate but not complete enough, the values in my comment above could be added onto it. A skim of the wiki seems to indicate it’s just boolean at this point.

What would you use? Do you think there’s value in making one up now, starting to use it, documenting it in the Wiki, encouraging people to use it?

I’ve used name:absent=yes on similar “unnamed, unsigned” things. While our wiki doesn’t document this, the local/regional OSM community around me seems to find this tag both dataful and agreeable.