Tag trail_visibility: Proposed Improvements for this Descriptive Tag

erutan · 2023-05-09T20:12:37.864Z

SomeoneElse:

Where are you seeing that?

It seems to have failed to trouble the scorers at taginfo.

Both the formal documentation and the new proposal:

https://wiki.openstreetmap.org/wiki/Talk:Key:trail_visibility#New_text_proposal

and

https://wiki.openstreetmap.org/wiki/Key:trail_visibility

Trail visibility of horrible is “often pathless” - what does that mean? Presumably less than 50%, given that it isn’t “mostly pathless”. If there doesn’t have to be any trail surface present (like there is in the example photo) then a few cairns here and there can constitute a horrible path.

Obvious and visible are not synonyms in terms of intermediate visibility, is something “invisible” or just “not obviously visible”. The idea of an invisible path is sort of amusing. I’d argue that “faint” or “hard to see at times” would be more appropriate than “invisible” for intermediate unless it’s for very short periods.

What is the difference between “Path mostly but not always obvious/visible” and “Path sometimes invisible, mapped way partly pathless”? They seem to be describing the same thing to me - if a path is sometimes not visible, then it is by definition invisible and partly pathless.

SomeoneElse · 2023-05-09T20:29:15.812Z

erutan:

Both the formal documentation and the new proposal:

No, the text “trail-” (i.e. any key starting with that text) does not appear on those pages.

erutan · 2023-05-09T20:37:23.037Z

Ah sorry, I meant trail_visibility there. I’ll edit the comment - I’m hopping between threads catching up on comments. The (unrefuted) argument still stands, but thanks for catching the typo.

If `trail_visibility=no’ can indicate a path with no surface just markers, then no and horrible would seem to map to the following on trailblazed:visibility:

horrible: trail markers exist but are rare or hard to locate
no: trail markers almost not exist and/or they are so rare that they are close to useless

So trail_visibility=no could be mostly pathless (with no visible signs of human passage on the ground) with trail markers that almost don’t exist and/or they are so rare that they are close to useless. How is that a trail/path in any realistic sense of “ground truth”?

This is more of an issue now - when the tag referred to the trail surface it’s visibility could be “no” while the path also had trailblazed:visibility=excellent (or good or intermediate) which would still clearly indicate a path.

If we are formally collapsing surface and markers together, then a path existing with no visible signs of human passage on the ground and with trail markers that almost don’t exist and/or they are so rare that they are close to useless doesn’t make sense to me.

IMO if something resembling a path exists for 5% of the mapped trail then that 5% should just be mapped - the rest would be a pathless route needing routefinding and having, say, a 2-3km trail when there’s 100m of path doesn’t seem appropriate.

erutan · 2023-05-09T21:29:50.651Z

The proposed text also conflicts with how this tag is represented in the OSM editor - no is considered “pathless” there (e.g. a route), not mostly pathless. That’s a significant difference.

Screenshot 2023-05-09 at 2.27.11 PM

Matija_Nalis · 2023-05-09T21:44:32.953Z

erutan:

The proposed text also conflicts with how this tag is represented in the OSM editor

Which OSM editor?

erutan:

IMO if something resembling a path exists for 5% of the mapped trail then that 5% should just be mapped - the rest would be a pathless route needing routefinding and having, say, a 2-3km trail when there’s 100m of path doesn’t seem appropriate.

So don’t use it in such case, eh?

Take the example closer to the other edge case - when there is 19km of trail, but “only” 18700m of visible path, then it seems more appropriate to tag that 300m which is pathless (e.g. due to weather washing away the visibility of the path/makers with mud etc. and not being renewed) as trail_visibility=no, to indicate that you can actually pass there (but only if you know where you’re going).

I think the idea is to help the routers (apps) – because, if there was 300m unmapped empty space between two paths, the routers will assume the terrain there is impassable, and would (instead of routing over “invisible”/unmapped 300m) happily send you on e.g. 30km detour around the mountain (i.e. on actually mapped highway=path way); which would be (for hopefully obvious reasons) not optimal solution.

erutan · 2023-05-09T21:45:01.738Z

I got around to adding an informal “trail fragment” that I’ve been thinking about for a while and felt would illustrate some aspect of this discussion.

OpenStreetMap

OpenStreetMap is a map of the world, created by people like you and free to use under an open license.

One could theoretically create a path all the way from Bearpaw Lake to either the peak or pass beyond (but, since things need to routed, one over Italy pass, white bear pass, then down into bear lakes basin, meeting this trail, over feather pass, then down to Miriam lake etc lol) but that’s all very open fairly obvious terrain with no signs of a path and the overall visibility would be horrible. There might be a subjectively placed informal cairn here or there, but nothing decisive.

There is a clearly worn rut through this sandy area that has ground truth for a path. It violates the rule that everything has to connect and has to be routable, but to me feels like the best way to approach this issue vs faking paths to cover all possible routes.

IMG_84481920×1440 522 KB

erutan · 2023-05-09T22:14:41.698Z

Matija_Nalis:

Take the example closer to the other edge case - when there is 19km of trail, but “only” 18700m of visible path, then it seems more appropriate to tag that 300m which is pathless (e.g. due to weather washing away the visibility of the path/makers with mud etc. and not being renewed) as trail_visibility=no, to indicate that you can actually pass there (but only if you know where you’re going).

In that case splitting the trail into multiple segments would tell a clearer story (the 99% which is good or excellent) and the 1% which is no rather than saying that the overall trail is intermediate or bad, though either of those would also apply to your use case above.

It’s an interesting disregard of ground truth / verifiability as a principle. I would personally put routing below ground truth for informal trails/paths as in the example above, as disregarding that just leads to a lot of fake paths. For formal paths disregarding ground truth for short segments for the sake of routing makes sense.

There are existing informal paths on OSM which are essentially a few kilometers of pathless terrain with a handful of informal cairns. They generally start from a path, but they aren’t just a short connector between paths to create a route but are technically valid. Having some disclaimer that pathless paths should only exist as commonly used connectors between actual paths would seem reasonable to me.

Matija_Nalis:

Which OSM editor?

iD.

Honestly having visibility “no” be entirely pathless makes more sense to me than “mostly pathless” from a semantic / intuitive standpoint. I’m not sure if the Wiki or iD editor is used more as a reference, but that is a noticeable inconsistency.

Matija_Nalis · 2023-05-09T22:42:52.367Z

erutan:

In that case splitting the trail into multiple segments would tell a clearer story (the 99% which is good or excellent) and the 1% which is no rather than saying that the overall trail is intermediate or bad, though either of those would also apply to your use case above.

Yes, absolutely. Sorry, I was thinking that was taken for granted. Like if the surface was 99% asphalt and 1% dirt, of course one would split that way and tag each one separately to indicate that!

erutan:

There are existing informal paths on OSM

Yup. There is informal=yes too (as well as note=* for notifying fellow mappers how to proceed when some situation is not obvious)

erutan:

It’s an interesting disregard of ground truth / verifiability as a principle. I would personally put routing below ground truth for informal trails/paths as in the example above, as disregarding that just leads to a lot of fake paths

It depends. In my experience, mappers will most often bother to add such “fake paths” only if there is no passage marked on OSM at all, and they want to indicate that it can if fact be traveled.
Once the first “fake path” that connects more discernible points “A” and “B” is mapped (usually, that seems to be the shortest traversable way between them), much more rarely (if at all) do new alternative “fake paths” pop up (although the mapped “fake path” might get moved few meters here or there from time to time)

This applies equally when the paths are not “fake” but actually visible on the ground, but numerous in short distances (e.g. unmarked forest paths). Especially with constrained/coarse GPS signal under canopy.

erutan:

Honestly having visibility “no” be entirely pathless makes more sense to me than “mostly pathless” from a semantic / intuitive standpoint. I’m not sure if the Wiki or iD editor is used more as a reference, but that is a noticeable inconsistency.

Yes, makes sense to me.

osmuser63783 · 2023-05-09T22:49:46.908Z

So hiking apps (e.g. Komoot) do allow you to plot “off-grid” routes. Useful for calculating total ascent/descent, to get a very rough estimate of travel time, etc. This should in theory reduce the need to add “fake” segments just for the router

erutan · 2023-05-09T22:56:54.225Z

Matija_Nalis:

It depends. In my experience, mappers will most often bother to add such “fake paths” only if there is no passage marked on OSM at all, and they want to indicate that it can if fact be traveled.
Once the first “fake path” that connects more discernible points “A” and “B” is mapped (usually, that seems to be the shortest traversable way between them), much more rarely (if at all) do new alternative “fake paths” pop up (although the mapped “fake path” might get moved few meters here or there from time to time)

Yes, I don’t generally see paths for all possible routes over terrain either once one is in an area.

It does create the issue where people will try and follow the path that shows up on OSM (I’ve done this myself, though I generally find out pretty quickly there’s nothing to it and just routefind on my own) where no path exists. This can create erosion, or destruction of terrain where having people spread out their impact if there isn’t a “path” to follow. This has happened a few times along Roper’s SHR (though you could argue that it’s also increased use) where passes that used to be pathless now have a social path on them after GPX files were shared online, which detracts from the wilderness experience and goes directly against the intent of the person who created that route.

I could create kilometers of pathless path to extend the feather pass/peak above, but I feel like it’s best as just a fragment (the impact is already there, on some level it shows people have made it that far).

Matija_Nalis:

Yup. There is informal=yes too (as well as note=* for notifying fellow mappers how to proceed when some situation is not obvious)

I’ve been tagging a lot of poor visibility social paths with this (where there is actually some ground truth). I’ve been deleting ones that are just “I did this” and there’s no sign of a path, after reaching out for advice or the original author if possible.

osmuser63783:

So hiking apps (e.g. Komoot) do allow you to plot “off-grid” routes. Useful for calculating total ascent/descent, to get a very rough estimate of travel time, etc. This should in theory reduce the need to add “fake” segments just for the router

Gaia GPS and Caltopo allow this in the US market.

I also feel like having to manually set off trail routes isn’t a bad thing, it keeps people from getting in over their head.

erutan · 2023-05-09T23:30:53.722Z

rhhs:

Can you please state in your posts if you can accept the new text, and if not, which text changes you propose?

After looking at the trailblazed:visibility page (which is a lot clearer wording in my opinion), the iD editor, and comments here I’d propose these for new visibility descriptions:

trail_visibility=excellent
Unambiguous well-constructed path everywhere. Easily followable when visibility is reduced.
None, orientation unproblematic, no map required

trail_visibility=good
A continuous path which is always visible but occasionally may have to be searched for.
Basic sense of direction, map recommended

trail_visibility=intermediate
Path mostly easily visible, but has short sections where it is hard to find.
Good sense of direction, map required

trail_visibility=bad
A path mostly exists, but the mapped way is partly pathless and you need to be able to find the path again.
Basic skills in orientation

trail_visibility=horrible
Often or mostly pathless, signs of a path exist but are often rare or hard to locate.
Advanced orientational skills

trail_visibility=no
Pathless (should be used for short segments between paths to help with routing).
Excellent orientational skills

The bright line around intermediate seems to be the hardest to define, I tried not to change the line too much, but I feel like this has more clarity to it. No is my opinion, and refers to pathless paths / routes which has it’s own topic.

The text in the iD editor needs some tweaking too, Bad doesn’t necessarily mean no markers.

There’s also the contradiction on whether this includes markers or not on the talk page, I also condensed down ground vs landscape wording, fixed references to “routing” etc for the text up top.

erutan:

The key trail_visibility=* is used as part of a classification scheme for hiking trails and paths. The key describes how easy it is to follow a mapped trail using all available visual aspects of a path. These aspects include the visibility of the path on the ground (differences in the trail surface vs it’s surroundings, a leveled or concave path surface, differences in vegetation such as a treeless corridor through a forest, etc.) as well as human-added markers such as trail blazes, poles and cairns.

It is based on the classification of the Swiss Alpine Club (SAC).

The key is not about routing visibility (how easy it is to choose the right trail at a trail junction), as that is covered by guideposts .

If a path does have markers they should be indicated in via trailblazed. See trailblazed:visibility=* for specific attributes regarding trail marking quality.

Hungerburg · 2023-05-10T22:12:18.203Z

Having fiddled with the wiki a lot, I am not very convinced that most of my changes did add much value. So I will not prepare my own version of what could be changed to the better.

All I can say, this OSM thing here is not really based on the scale by the SAC, and the proposed changes will make resemblance ever more remote. The scale by the SAC itself got changed in the meantime too. Best drop the statement of its basing, its mere name-dropping/argument-on-authority now.

If trail and path are not the same, it is inapplicable to OSM paths anyway. The scale developed by the SAC was meant to apply to routes, that may be pathless, not on paths.

osmuser63783 · 2023-05-10T23:19:53.584Z

I really like your descriptions of the individual values, I find them clearer and more logical than the ones in the Wiki and the ones in iD while still being consistent with the spirit of the tag. For no I would drop the “should be used” bit though, to me that sounds like encouragement to use it more.

I also agree with Hungerburg about dropping the remark about the SAC and with yvecai about keeping the intro concise. I would keep the bit from yvecai’s suggestion about how people take markers into account differently, as that is what much of this discussion has been about. Maybe the remarks about guideposts and the trailblazed tag could go into a “see also” section at the bottom of the page instead of the main definition?

rhhs · 2023-05-11T09:45:27.986Z

Hungerburg:

I oppose any change to the documentation, that will permit mappers to mark trail blazed (alpine) routes with trail_visibility=excellent|good|intermediate where on the ground there is no trodden path.

How would you tag this path? (no trodden path because it’s used only when there is snow cover)

image1200×900 142 KB

yvecai:

The more text you add in a definition, the more it is prone to bad interpretation.
In particular it does not help saying “It is not about …” for concision.

Taking that to the extreme, the best would then be to delete the whole wiki, wouldn’t it? No text, no confusion I think our aim should be to maximise consistent mapping, for which some description of how to map is necessary to bring mapping by different persons all over the world in line with each other. There is an optimum somewhere: a concise and clear text will help consistent mapping, but too much and unclear text will confuse more than help.

Hungerburg:

will the illustrating pictures still match the words

That’s why I deleted the second picture for trail_visibility=good.

Hungerburg:

Why mention mapped paths?

By “mapped trail” I mean the way as marked on the map (virtual), as opposed to the path in the real world. Do you have a better way to phrase that? “mapped way” maybe?

erutan:

If a trail is a “Continuous established path, always visible” then why is a basic sense of direction required? “Path always visible, but sometimes has to be searched for” makes more sense to me and is less likely to break current usage.

I’d prefer to remove “basic sense of direction required” than to have “always visible” and “search” in the same sentence (this is an obvious contradiction to me: if it’s always visible, it logically follows that it never has to be searched for).

erutan:

So trail_visibility=no could be mostly pathless (with no visible signs of human passage on the ground) with trail markers that almost don’t exist and/or they are so rare that they are close to useless. How is that a trail/path in any realistic sense of “ground truth”?

I propose to add that to add a path with trail_visibility=no to the map, it should have some form of Verifiability such as that there are markers (but they are so bad quality that they are useless and don’t help trail visibility) or that the path position has been published by an official authority (it applies to the path I mentioned in my post 35 above). One can also use it to tag small pathless sections of a longer path with generally better visibility as @erutan suggested.

I updated the “New text proposal” at the wiki talk page accordingly.

Hungerburg · 2023-05-11T10:18:06.852Z

A bit about SAC: The SAC Mountain and Alpine Hiking Scale 2020 does not have a “trail_visibility” gradient among the aspects considered. Nor do the SAC- Berg- und Alpinwanderskala 2012 or the Neue Berg- und Alpinwanderskala 2005 available at time of the introduction of this tag.

T1 and T2 routes both need a path that is visible throughout. The difference is the risk of slipping and the consequences therefrom; Or, as I came to conclude from the Austrian system, in mountain hiking (Bergwandern, red dot) you will have to lift your legs to get over roots or rocks, in hiking (Wandern, blue dot) you can stroll along not lifting your legs much apart from where there are steps.

Starting T3, the route may have pathless sections, the ratio of which increasing in T4 and T5 until in T6 most of the route is over pathless terrain.

I am with @rhhs, excellent needs a constructed path. The wording in 2020 may look like bad translation, “Bahnung” in German is the prime feature of what constitutes a Way. It need not be constructed, can also come from wear, but the last years have seen a shift in grading. There are good pictures at SAC-Wanderskala – Wikipedia that capture this. It will break the documentation, but who will not get furious, to learn that a path with gaps has visibility good?

Hungerburg · 2023-05-11T10:19:10.108Z

rhhs:

How would you tag this path?

That is an easy one This is a piste:type=hike

Hungerburg · 2023-05-11T10:30:14.114Z

erutan:

These aspects include

This starts an open taxonomy. An invitation to people to add their own pet features of what they encounter expressed on the ground.

osmuser63783 · 2023-05-11T10:36:50.606Z

Hungerburg:

I am with @rhhs, excellent needs a constructed path.

Ah, I just spotted that. “excellent” has been used about 2000 times for paths with “informal=yes”. So if we change the definition of “excellent” to say this can only be used on constructed paths, then I don’t think this matches how people have been using trail_visibility

Hungerburg · 2023-05-11T10:57:44.312Z

I ran that in my local area and got three hits: One is a bridge in neighbouring South Tyrol, one is a scramble (that I created under the old regime, I will update tagging), the third is in neighbouring Bavaria and most certainly is not informal, as it connects some alpine pastures and even has a ref number as used by the DAV. So - false positives only here.

Running this on even lower zooms I see more of them, preferably in urban surroundings

osmuser63783 · 2023-05-11T11:37:41.206Z

Yeah, it looks like a lot of them are desire lines in urban paths and things like that… can’t really blame people for using trail_visibility like this, even if this isn’t what the key was originally meant for.

The most practical solution might just be to say “Unambiguous path everywhere”? It should not make a difference for mountain paths: well constructed paths are unambiguous, and conversely if a path is eroded and poorly maintained then it is probably no longer unambiguous!

(I do wonder how much of this is splitting hairs, and how closely people read the Wiki…)