it completely depends on the context, a path in the middle of peaks and few places around (i.e. “in the mountains”) definitely will lead to different expectations what you might occur, compared to one in the centre of Aachen.
But I agree, if it is demanding alpine hiking you would better tag it explicitly and not rely on context alone 
informal=yes can mean lots of things; a dirt path cutting off a corner as a way out of a car park to an extreme scrambling route. You just cannot assume how a tag works in DACH applies worldwide.
It is far better to tag these things explicitly in such a way that the tagging works worldwide.
Thank you all for the comments so far. I deliberately coined this as RfC. Not just, as I wrote, if this makes sense, but also, if the terms are concise, if there is something missing, etc. So I recap, what a potential article on the Wiki might look like:
Sport=Mountaineering
Mountaineering combines hiking and climbing into an activity, where the goal is to reach points of interest in mountainous regions, be that summits, viewpoints on ridges, saddles or sometimes even huts. Typically, the terrain is neither welcoming to easy walking without any preparations nor are the climbs so spectacular as to merit mapping them as climbing=route.
For a start, there is only one OnTheGround feature, that supports exercising this sport, and that is “mountaineering=route”. This must not be understood as a route relation. It is used solely on linear ways. Those ways can have other attributes, e.g: highway=path, sac_scale=*alpine*, trail_visibility=*, assisted_trail=*, trail_blazed=*, climbing:grade:uiaa=*, and so on.
There are several things to consider: Most of what is currently tagged sac_scale=difficult|demanding_hiking in the OSM database will happily fit. The community did act in concert on this, as far as I observe. mountaineering=route should work as a substitute for highway=path in many cases, where trail=visibility=no, eg. on scree or bare_rock where trail_blazing=markers|cairns.
The other, how to word this, that it does not give a blank permit on just dropping gpx of random scramblings? And more?
Structurally, your proposed tagging seems to treat sport=mountaineering as the main tag that is then refined with mountaineering=* subtags. This is unusual because sport=* normally works the other way around: As a subtag for tags such as leisure=pitch. I feel it would be more standard to have a physical main tag (maybe leisure=mountaineering_route or something like that).
More fundamentally, I don’t really see the benefit of introducing such a tag if the ways are also going be tagged as highway=path. With the ongoing concern about unprepared people being led on dangerous routes, surely a proposal to reform the tagging of such features should make it “opt in”, i.e. tagged in such a way that a naive data consumer will not treat these the same as a stroll in the park.
IMO, a new proposal for data consumers that currently still ignore tags like sac_scale or trail_visibility that are in use for more than a decade will not help anyone.
Worse, it will be the source of additional nitpicking on their side.
I’d really go for highway=mountaineering as a primary tag to make it abundantly clear that this is not a normal path and never can be.
Though I would try to get Tag:highway=via_ferrata - OpenStreetMap Wiki widely accepted first: this one has nice part of being much easier to get clearly defined and solves even worse issue as far as routing goes.
highway=mountaineering has a big problem of what is the edge between this and highway=path/footway - especially bad in mountains on edge between this two, like Tatra mountains where many trails can be easily argued to classify as either.
Basically there is no “normal” path in a global context, paths are at most “normal” (meeting the expectation for a typical path) in a local context. Clearly in a mountain context, a “normal path” looks very different from one in an urban area.
I am also with Yves, we already have tags to distinguish them, like sac_scale and trail_visibility.
highway=via_ferrata is, as the name suggests, for via ferratas. This is something different than a difficult mountaineering path. For starters, it requires gear. It is a very bad idea to water down this tag and add any climbing or mountaineering paths to it.
Basically there is no “normal” path in a global context,
A normal path should be something that an average person can use without special gear, training or experience. That’s what I would expect from what is written in the wiki. A mountain path with SAC scale T4+ does not fall under this definition.
The good news about mountain paths is that we already have a widely used scale to define the threshold when it is not a path anymore. So it’s really easy to introduce a different highway type. We just have to agree if that starts at T3+ or T4+. (The wiki says for T3 (demanding_mountain_hiking) that hiking shoes are required, so that would mean that it cannot be used without gear.)
IMO, a new proposal for data consumers that currently still ignore tags like sac_scale or trail_visibility that are in use for more than a decade will not help anyone.
I’m all for educating data consumers about secondary tags but the situation with paths is not something anybody without significant mapping experience can be expected to understand. To get things right they have to know all about access modes, mountain paths (sac_scale, trail_visibility), winter sports (snowmobile, ice_road, winter_road, seasonal) and tides (tidal). And these are just the tags I’m aware of which could make a serious obstacle out of an innocent highway=path. I suspect there are more.
hiking shoes are “special gear”? FWIW, you can go anywhere where mountain shoes are suggested, even barefoot, and it will be less comfortable and you might move much slower, but it isn’t impossible (unlike for example climbing a wall for an untrained person without gear might be, or having to swim for a person that cannot swim). Usually the problem that people have with paths in the mountains is that you might fall down, often there is only a narrow space to walk and just besides the path there may be a cliff or similar situation, so it is not the path itself but the surroundings.
Winter roads are not particularly related to hiking, these are for example roads in Siberia and similar, where you can only drive in the winter because in the summer it is a swamp.
I am not proposing breaking definition of that tag.
Just mentioning that via ferratas remain unsolved subproblem of this issue and getting highway=via_ferrata accepted would be easier to achieve than getting highway=mountaineering accepted. So it may be better to start with that.
You know how this works: we end up with two competing tagging schemes, not the best way to simplify data consumers’ life.
That’s the whole point of suggesting highway=mountaineering instead of sport=mountaineering. It cannot be used in parallel with highway=path. All the other secondary tags will stay the same. If a data consumer targets mountain sports, they surely already look at sac_scale, so they only need to add a rule that says handle highway=mountaineering like highway=path. For everybody else, the ways simply disappear. So data consumers should have to make an active choice to include the path but once they do, it’s really simple to handle.
If you refer to the fact that there is going to be a vocal minority of mappers who insist that they could climb the Matterhorn barefoot and therefore will keep using highway=path no matter what, then yes, I’m sure there will be some. That cannot be helped. Still, even if not all ways are changed, the ones that are adapted already make the situation better.
My favourite so far, thus another try at camera ready documentation:
Highway=Mountaineering
Mountaineering combines hiking and climbing into an activity, where the goal is to reach points of interest in mountainous regions, be that summits, or viewpoints on ridges and saddles. Typically, the terrain is neither welcoming to hiking without first becoming knowledgeable of both the route and the area, nor are the climbs so spectacular as to merit mapping them as a climbing route. Mountaineering calls for a diverse set of skills and can require extensive preparations. Mountaineering is exercised in leisure time by a number of people, it can grow into a sport in its own right even.
The basic on-the-ground feature that supports mountaineering is the route. It can be mapped with the tag “
highway=mountaineering”. A route is similar to a “highway=path”, but it is lacking in some ways. Most eminently, the trail is not continuously visible on the ground, especially when it passes over bare rock or blocky scree for example. Yet, people commonly manage to follow along, by looking at markers, cairns, spotting the continuation a bit further, and so on. It takes a bit of training. Second most eminently, in contrast to a “highway=path” a “highway=mountaineering” may comprise non-walkable sections, where using ones hands is required to get along.Useful combinations are:
sac_scale=*alpine*, trail_visibility=*, assisted_trail=*, trail_blazed=*, climbing:grade:uiaa=*, and so on.When attributing difficulties, the most demanding section is to be used for the whole of the trail. Mountaineering trails shall not get split, where the difficulty changes, but only at places, where there is a goal to be reached.
I came to Openstreetmap for the richness of data for hiking. I am very certain, the smartphone app, that I use, when on the way, will support this tag.
I’d really like to make the tag applicable too, when the goal taken is not a high altitude point of interest, but the route chosen requires mountaineering skills, e.g. to the mid-route hut through a gorge. Will ponder a formulation… any help welcome.
We really need to be pro-active to get this new tag on osm-carto, 'cause Tagging something to make it disappear from The Map will not get much traction…
This is not really sustainable for ways, this make it looks like a tag for route relations.
I disagree that that should be necessary because osm-carto should not decide how to tag things and my impression is that they are quite reluctant to add new features to the map. But of course in practice things on carto will be mapped more frequently. Therefore it is still a good idea to ask carto to render it.
That’s my opinion as well. Now there are highway=path sections that can be accessed by highway=steps only and we only tag the stairs as stairs. This way data users can decide what to do: Show the actual situation of a section of way or use tags of the route relation to show the situation along all of a route.
IMHO it would be more effective to add a map layer that does show it, like for example opentopomap. Carto lacks nearly everything you’ll want to know for such ways like contour lines, sac_scale, trail_visibility, via ferrata, hiking routes, etc. so what’s the point to show the way without this information? You could get some of this with overlays, but it’s not like people are using carto to go mountaineering, there are plenty of applications which are much more suited for that.
The prompt, to not split when difficulty alone changes, directly mirrors the SAC guidelines. I personally find it perplexing, when the route to Praxmarerkarspitze over 2.3km from where it splits from the route to Kaskarspitze, changes from alpine_hiking to demanding_mountain_hiking, then back to alpine_hiking, then to simple mountain_hiking to become demanding_alpine_hiking in the last third. That reminds me of so-called Topos from via ferratas. I do not see much sense in such micro-mapping mountaineering routes, and it does not square with the spirit of sac_scale.
As to rendering, special maps of course will show the paths however blunt or unimposing as deemed fit for their purposes. Regarding OSM Carto: In issue 1500 a maintainer proposed to render some dots, like an ellipsis in typography, where the mountaineering route leaves the well trodden hiking trail. It won’t help anybody out there though, where cell network coverage often is poor, but might inspire app developers. It might be useful to map a short path segment, where the start of a mountaineering route is excellent in visibility, to aid hikers to follow verbal directions that count crossings.
Does the “not splitting” approach require a clear definition of what exactly defines a single “route”? I generally walk on relatively easy hiking trails, but sometimes when I come across a difficult section that requires use of hands, I tag that using sac_scale, only on the difficult section. I am not sure how I would apply it to the “whole of the trail” - how do I know where the trail starts and ends?
Same as Alan. Also, when micromapped that way, it is possible to to get the highest or average or whatever metric of the difficulty of an itinerary.