RfC: Have pathless hiking routes (no trail_visibility) be some manner of related nodes instead of paths

The discussion of node networks in this thread is very confusing. A node network is a very specific thing with very specific signage and should not be conflated with a pathless route in the wilderness. Very different situations. If what is meant is “an ordered list of nodes”, then lets say that to avoid confusion.


Agreed. For small gaps in a path I see no problem with that being mapped as a highway=path. My comments are specifically about routes that are entirely or almost entirely pathless.


Just to clarify another potential source of misunderstandings.

In the OSM database, a ‘way’ is a type of object that contains nodes. It can represent a highway=path but also a river, a wall, a ski lift. Just because a list of nodes is stored as a ‘way’ in the database doesn’t mean that maps need to draw a straight line between successive pairs of nodes. Straight lines drawn on maps is what we’re trying to avoid, but that’s not an argument to avoid the use of the OSM database object type ‘way’ in favour of something more complicated (a relation).

The real question for me is: does the OSM object that you want to invent - for the network as a whole - only contain nodes? Then it can be stored in the database using a ‘way’. Or does it also contain ways (such as highway=path) and/or areas? Then it would need to be a relation.

Edit: Also if you want to store information like ‘from A you can go to D either via B or via C’ then I think you would need a relation.

1 Like

A related / networked set of nodes is what I thinking of, similar to how climbing has a top, bottom, and bolt node but not connected visually with a way.

An ordered list of related nodes works for me. :slight_smile:

I’d be fine saying the only acceptable use of something with true trail_visibiilty=no is a ground truth hack to allow for routing of formal trails across small gaps. That sort of begs the question of whether trail_visibiilty=bad should exist, although it’s probably not reasonable to expect every gap in trail to be broken into a segment and labeled. I do find small segments on trail in the backcountry more useful and appropriate than paths with varying strength and visibility.

I illustrated an example here of a short patch of trail_visibiilty=good where IMO it’s not appropriate to create massively longer path to allow for routing. I also agree that these shouldn’t show up in standard routing so that kind of makes the point moot to me. :slight_smile:

I wasn’t leaving areas out on purpose, no, I forgot about them :smiley:
But I agree with @erutan that walkable areas in the mountains often don’t really have verifiable boundaries. Picture a U shaped valley, where precisely does the sac_scale=2 terrain end and sac_scale=3 terrain begin? The hiking just gets progressively more difficult the steeper the slope.

Thanks for that clarification.

I do fear that thinking of it as a way will lead to there being more nodes than is absolutely necessary or in the spirit of such pathless routes. That’s not necessarily a strong argument against it.

I feel like areas can be useful to show ambiguity - for some routes (say up to Dragon Pass/Peak) there is a heavily informally cairned area at the side of the lake below it which is a natural place to ascend and could be a set route_bottom. For others it really is just "head up somewhere, anywhere, just make sure you end up at the pass marked as route_top. It might not be worth having an area for the bottom as that could complicate things, instead just leaving it blank and up to map reading skills.

The route could also just be represented by area shapes that show “passable” terrain as was previously suggested by someone else earlier in the thread (Italy mod iirc).

For Valor Pass I know of people that hiked up the ridge on the south side, then dropped as that is a little simpler. We felt perfectly comfortable doing a more direct approach, but stayed above cliff bands (which doesn’t really show as well on opentopo, but with USGS or slope angle shading the map gets noisier for this purpose). This is a pretty vague representation based on my memory - and isn’t entirely accurate as there might be some 2-3m cliffs or ledges etc that could pose a navigational hazard in the middle of that area.

This tells the story pretty well, but is probably a bit overkill and prone to small errors in terrain. In this case just having a node at the top of the pass seems fine to me, the rest doesn’t have any needles that need to be threaded (aside from perhaps the chutes dropping from Valor lake to Ambition, but at some point you have to consider the route to the pass done).

(looking at this, the pass should be moved a little bit to align with the two non-technical approaches to gain the ridge)

Good summary. I think we should focus on these issues and leave 3) for later. (For the record I’m broadly in favour of @ezekielf’s position here).

On 1. I think we will always have some attempt to represent various hiking/climbing/scrambling routes as we do now. So I think the answer is broadly yes. What we don’t have is some clearer demarcation of bounds.

On 2. I think we need a bit more of a discovery phase. I can suggest locations in the Alps, Scottish Highlands and upland North Wales each with somewhat different characteristics, from those of the Western US. It would be interesting to have a perspective from NZ or Australia where missing the route seems rather hazardous.

We also should consider how this might apply to other areas: arid places such as deserts, tropical rainforests where routes get overgrown.



Deserts are relatively straightforward - sand will cover occasional foot traffic in many places within a year, which helps prevent informal paths from being created (though in some areas you have “biological crust” which should be avoided).

Paths can go onto slick rock as shown in the set of photos earlier in this thread which don’t hold any footprints at all but can be easily cairned. There are a lot of small cliffs in the rippling of sandstone which pose a navigational and safety hazard.

Generally long lines of sight, unless one is in a canyon, at which point things are pretty straightforward. Sedimentary cliffs tend to be problematic with different layers, having undercuts / overhangs along with strata that are very vertical are common, but finding routes that work up/down on fans caused by erosion is pretty easy to spot on topo using slope angle shading.

I’m personally mostly interested in the shorter routes up to a pass, side feature, etc rather than long multi-day routes. The longer ones, such as this previously mentioned 315km mix of paths and routes, I’m unsure about trying to represent in OSM.

Someone created and is selling a guide including “breadcrumbs” but said the following about it “My maps have breadcrumbs, not a line, in order to depict the route. Two reasons. First, there are oftentimes multiple routes to get you there. Second, I didn’t want to steal your fun by showing you the optimal route between points — you can figure that out.” in comments and then regarding a GPS track “No, there is not, intentionally. I would encourage you to ask yourself whether you need or want one. If you need one, the SHR is probably not the right route for you. If you want one, you probably know how to create one, or you can use this as an opportunity to learn.” Further comments showed that person found one online and it was useful for them. I haven’t bought the guide as I’ve done bits and pieces of nearly the entire SHR over the years just on my own trips.

I personally don’t want nodes in way so close together that they’d create a certain interpretation of a route, or create paths of that interpretation (though of course myself and others would be free to disregard it). That’s a pretty common feeling in the conversations I’ve had elsewhere was well. That said I am totally open to there being regional differences for the precision at which these routes are rendered. :slight_smile:

People die from being off route in the US fairly regularly (which is why people need to know how to routefind!. That said the terrain in NZ tends to be incredibly vertical (and in the alps from what I know), both ranges have significantly more glaciation than a lot of the ranges in the Western US and don’t have the nice hanging basins that I and others enjoy so much, which would limit terrain. A lot of their normal trails are SAC T3-4 lol, which surprised me, and there’s not as many pleasant options for off-trail travel given vertically, thick brush, and areas that are boggy.

There’s been a useful back and forth on the US Trails group channel on node_networks. Someone in Germany came up with a “basic network” that could be worth looking at for ideas.

I think the structure will be defined by how much precision is wanted. I prefer just having some key “digital cairns” so having a grouping of nodes being more cumbersome seems fine, and even helps promote that thoughtful minimalism.

If people want a more defined route to follow then the amount of nodes would lend itself to a way. The latter could be more appropriate for some of the formal “alpine routes” in the alps etc.

I’m not sure it’s a good idea, but there could be the option of a grouping of nodes for “looser routes” and a way for “tighter routes” to cover different use cases and community norms.

1 Like

Could we try some terminology exercise? Creating formal definitions for words like “OSM way”, “physical way”, path, route, etc ?

Perhaps it might help structure debates and reveal where we have different use cases in mind.

1 Like

I really raised these examples so we avoid just thinking about use-cases of wealthy people in the Northern Hemisphere. There are others we encounter though, as well, routes across tidal areas, and some wetland areas (examples below).

Looking from Little Eye to the two larger Hilbre islands (route is on OSM)

Top of Kinder Scout after traversing the peat, again a non-discernable route is on OSM. Amazingly this used to be part of a national trail. For some odd reason there are few photos in the centre, but this one is authentic. People regular lose their bearings which can become serious if darkness falls. Also about 2.5 million people live within 20 miles.


A story, terms in bold: Recently, I learned of a POI. It was not on OSM. I found it on aerial imagery. No way leading there was on OSM. I looked at aerial and DTM of the area and in my mind several routes formed, starting from the closest mapped way to the POI on top. I hiked to the bottom of where my preferred route started. To my surprise, there was an unmapped path there, right before my eyes. Turned out, the promise did not hold for long. I soon lost the path. Searching for the continuation, to my next surprise, I noticed dark green trail blazes on trees, telling me, that I am on track. Having the top marked on my hiking map app, I would have found my way there without path or trail, especially as the cliffs in the area were mapped in good detail. Having trail blazes certainly helped with staying in the space, that one could call a path and in addition enforcing belief, that this route was not only my own idea. All in all, a fun endeavour.

Update2: How to loose the path? Two things come to mind: 1) Follow something, that looks like a path, but is something else instead, e.g. erosion or a game trail, 2) the bootprints become so faint, that their traces indiscernibly blend into the surroundings.

1 Like

I’ve been using the following / have been educated on them:

path - something which has a visible surface, or human markers to follow and has some degree of verifiable ground truth. very OSM centric. if you took a lot of GPX recordings of different people on a path, they would not have significant variation.

trail - closely related to the above, used mainly for foot traffic but can include bicycles, horses, etc. generally formal / official, but not necessarily. may be pathless for short segments despite being shown as a path in OSM for routing purposes (getting to a destination via one or more paths/awys), but for the most part should have “ground truth” / can be visually verified.

route - while more general, in this thread I’ve been using it to describe pathless paths, or routes which either due to terrain do not show a surface to follow, and have not been marked via cairns, blazes, or poles to have a set path to follow. these are generally fairly low traffic, and require some degree of “routefinding” which is the US way of saying “orienteering” which is either an OSM or European phrase. if you took a lot of GPX recordings of different people on a path, there could be significant variation, though the variation would tend to contract and expand at times given on how passable terrain is.

node - a single coordinate / object on OSM.

way - a linear collection of nodes in OSM which generally form some sort of path, but which could be rendered in a way to not show a connection between nodes.

node_network - a specific way of relating numbered nodes (guideposts) at junctions of ways/paths in OSM to allow for routing (junction 32 to 11 to 789).

“group of related nodes” - my attempt at taking the idea of a node_network, but using nodes for key decision points within a route, rather than having them at the junction of paths. ideally less nodes than there would be in a way, but each one is significant.

This seems like a fun challenge for all of us: to come up with examples around the world of such routes or itineraries that are best described as a series of way points, where the connections between them cannot really be described as a path.

Thinking about concrete examples and whether and why they should be on OSM could also motivate us to think about what makes a route verifiable when it is not signposted or directly visible on the ground.

It could be that a route is verifiable because it is local knowledge, where if you live in an area then you just know that the best way to go from point A to point D is via B and C. Similar to how the name of a path might be acceptable to have in OSM even if it’s not on any signs.

Or maybe a route might be considered verifiable because it can be found in guidebooks? Copyright and database rights issues would need to be considered of course, and notability is an issue. But I imagine that there are some routes that have been printed and reprinted in many guidebooks for decades and cannot be considered the intellectual property of any one publisher.

Maybe past discussions mentioning trail visibility can help us identify candidates… though many of them seem to be about short connections (examples 1, 2). Another option is to check on Overpass for trail_visibility=no tags, or to reach out to communities in different countries.


Turns out that we have a 650 km unmarked hiking route on OSM in Scotland. Quite a few sections of the route are entirely pathless, such as this section of a variant. I found it because I have walked this, long before the Cape Wrath Trail was devised, and because there are quite a lot of Strava traces.

The “trail” has a paper guidebook, set of dedicated professional maps, and a webpage. Parts of the trail have been known for a long time as wilderness walks: my parents crossed the Rough Bounds of Knoydart, which forms part of the trail, before I was born.

This single superrelation probably offers quite a few examples in addition to the one I’ve linked to.

1 Like

I think there is something to “unmarked” routes created and generally known by locals and others familiar with the area. We could place a node at each of natural landmarks used to create the bounds of the otherwise unmarked route. These can be considered to virtual route marker.

(veering slightly offtopic from “pathless” routes into “verifiability if no signage”)

Unmarked trails will usually fail the verifiability test and on that basis should not be in OSM. The tricky bit is where do you draw the line? Often trails are “well known” (similar to what @IanH suggests) even without on the ground markings.

In the UK, the Lyke Wake Walk and Wainright’s Coast to Coast were historically largely unsigned. I’ve added guideposts et al to both OSM relations from survey and as you can see, they’re now very well signposted. Another “borderline” one is this one. There is some signage, but not much. Should it be in OSM?

As someone recalled me recently, most state borders in Europe are not verifiable on the ground but only from legitimate sources. In France we have chosen the same approach for routes because some do not have their own signing. The tricky part of course is to define what a legitimate source is, but it’s manageable.


I think most map users will appreciate being shown such “paths”, so I’m in favour of showing them.

This is pretty close to what I was envisioning. Say a few points that would illustrate how you would describe a trail in a brief spoken description like “drop down from the pass to the little pond, then keep to your right once you hit the pond and drop down a series of ledges until you reach a set of cliffs where you can find a crack down near a meadow” - you’re not going to be telling them GPS coordinates for a node every 5 meters, but a note for the pass, the pond, then the crack through the cliff band near the bottom etc.

There could be a use case for tighter / more accurate ways in some cases.

If I’m reading this correctly you’re asking if something with trail_visbility=intermediate, trailblazed=no, and no guideposts at intersections should be a path to be included in OSM?

Isn’t this one of the points of the informal tag? If there is a desire path created from human traffic that is visible it doesn’t need signs or guideposts to have ground truth.

An interesting consideration for this thread from the bottom of the verifiability page.

We have a general rule that says stuff must be verifiable. General rule doesn’t mean “no-exception-made set in stone”, but it means: You need a very good reason for mapping something not verifiable. Such a reason could be that the data is of a relatively high usefulness to relatively many people, or that the data is useful for mappers in their work.

I think in regards to guidebooks, something where the author “creates” a new route - like Skurka or Roper probably shouldn’t be included, as that would undermine their work (or at the very least it should be very vague, like a start/end). If a guidebook like Secor which has brief descriptions of hundreds of passes and peaks, then it seems fine to include things from it as they (with a few exceptions) existed before that and the guidebook is documenting them vs creating them a commercial product and/or service to the community.