(shortened to contain only tags relevant to the discussion)
Total relations analyzed: 857
| Tag | Count | Percent |
|---|---|---|
  route |
857 | 100.00% |
| type |
857 | 100.00% |
| piste:type |
846 | 98.72% |
| name |
813 | 94.87% |
 piste:grooming |
410 | 47.84% |
| piste:difficulty |
367 | 42.82% |
| piste:name |
82 | 9.57% |
| colour |
73 | 8.52% |
| piste:oneway |
70 | 8.17% |
| lit | 31 | 3.62% |
| piste:lit |
30 | 3.50% |
| piste:ref |
8 | 0.93% |
| piste:operator |
1 | 0.12% |
(shortened to contain only tags relevant to the discussion)
Total relations analyzed: 1249
| Tag | Count | Percent |
|---|---|---|
| route |
1249 | 100.00% |
| type |
1248 | 99.92% |
| piste:type |
1224 | 98.00% |
| name |
1189 | 95.20% |
| colour |
834 | 66.77% |
| piste:difficulty |
511 | 40.91% |
| piste:grooming |
405 | 32.43% |
| ref |
267 | 21.38% |
| piste:name |
77 | 6.16% |
| piste:oneway |
45 | 3.60% |
| piste:priority |
28 | 2.24% |
| piste:ref |
8 | 0.64% |
| lit | 8 | 0.64% |
| piste:lit |
5 | 0.40% |
| piste:abandoned |
2 | 0.16% |
| piste:type_1 |
1 | 0.08% |
| piste:colour |
1 | 0.08% |
I meant, signposted color on the ground, and resort maps ?
Not so, if a piste is not a route but the building block of a route. In the small resort that I am familiar with, that mostly consists of two routes, difficulty as advertised on the ground is NOT the same all over each single route, at least when route is understood as all pistes with the same name.
The routes are split into sections that get numbered refs which indeed have single difficulty. That way they can boast with a higher number of pistes 
I have been reading your comments repeatedly, trying to understand the reasons for the opposition to the proposal. I realize now that I may not have emphasized enough in my original post that the proposal focuses specifically and solely on piste:* keys on route=piste relations and its members. Its goal is to officially allow the piste:* keys on route=piste relations, as currently, only the piste:type key is explicitly permitted (see Tag:route=piste).
By doing so, tags such as piste:difficulty=* and piste:grooming=* would no longer need to be duplicated on every way that belongs to the relation. If certain ways require different values for piste:* keys, their individual tags would override the default values set in the relation.
I think it’s ok to document piste:difficulty on route as Lonvia put it:
For piste:grooming it’s complicated because it’s the grooming on one side, but also describe completely other practice, gear, etc… in conjonction with piste:type.
For the later meaning it’s ok to tag it on a route, for sure.
But by no means tags should be put on a route relation to avoid tagging every member ways: these tags on route relations do not have the same meaning than on ways.
What against redundancy? It helps spot errors! And really: When a route becomes harder, does that actually mean, all of its sections are equally affected?
I feel reminded of sac_scale: The key according to documentation not to be applied on relations, only on ways. Though I’d consider it fine on relations, the hardest part makes the class. @yvecai With pistes there is more distinction when applying difficulty – please explain.
PS: According to overpass turbo such routes (Tyrol as an example) mostly used for winter hikes and Nordic pistes – The Arlberg region suspiciously devoid of those. I also used them only for skitours.
PS: Is that a mapping error overpass turbo type=piste on a relation?
PPS: Here a downhill slopes map Pistenplan – The canonical routes are 1 and 3. So far nobody bothered to map them. I can sympathize.
I won’t quote @Lonvia a 3rd time, it’s just 2 messages up😁
Some mappers insist of mapping the 'official ’ piste difficulty on the relation. Also some others can’t remember the section by section difficulty after a 50km tour on nordic skis. Hence I was explained several times the need to tag an overall piste:difficulty on route relation, and I can hear that.
However , like with your sac_scale analogy, I think piste:difficulty is best mapped with higher granularity on ways, but in that case that’s obviously a way-by-way difficulty that is mapped.
There already are difficulty tags at the level of hiking relations, either determined from flaky information by operators (e.g. length of the route) or from “serious” analyses such as the French FFRP’s difficulty index.
But the good thing there is that sac_scale on ways and difficulty on routes are clearly different things, with different names. Why not do the same here?
Sure that would avoid confusion. But in tag pressets day, does it really matter ?
I will cease to matter when all tag names and values are UUIDs rather than words, and when nobody relies on the wiki to understand and debate what a tag means
Seriously, I don’t believe that presets will ever relieve us from the need to have an non-ambiguous and intuitive tag structure, because that’s what we keep referring to.
Given that we clearly don’t have that at the moment, how could that possibly matter?
More on topic: I can see some value in being able to choose routes based on the difficulty of segments, but on the other hand from a practical pov I doubt that anything else than overall ratings will get much use. I mean even on that 50km Nordic route, if there’s an icy downhill bit, people will do what they have always done, fall over and pile up, and similar for alpine skiing.
At least there’s a chance to adapt your speed 
.
More seriously, more than often in my area there’s warning signs in winter before such sections.
The 50km argument was made to me not for a 50km route, but rather a 50km run on various cross-country loops.
For reference I found back the discussion on actual validation rules for JOSM.
Let’s agree to disagree on that, and see if the difficulty tags can prove us both right
Same here. As implied by Sarah in a previous message there is a more or less solid rule that the difficulties of individual segments are inferior or equal to the global difficulty. Which brings us to the idea that local “hotspots” would probably be best served differently, so as to be shown on maps as danger signs for instance.
Maybe @Hungerburg will remember in which thread about sac_scale and path difficulty there was a discussion about tagging local difficulties? We could adopt a similar solution here, on ways first and maybe on points later.
That must have been one of the topics started by @erutan - I still consider the deliberations worthwhile, alas the talking community did not hop on, even tough hundreds of posts spent.
Today, I happened over this guidepost: The first few hundred metres of the routes posted all follow the same highway=residential paved with asphalt which, although steep, certainly nothing but “blue” in the local system.
From my understanding, all local systems only cover route difficulty, and that includes SAC® one. Only openstreetmap differs, in that difficulty is reported for the constituting ways, not the route itself.
PS: It may not be obvious from the photo, the top route is signed “black”. There is no “blue” route signed at all on this guidepost.
We really should have relied on the color= tag from the start.
Will abstain from commenting on the use of the colour=* key for difficulty rating. So much: The Austrian hiking scheme copied the downhill skiing piste scheme colours. If dissecting a route in Austria, one will find that the difficulty mostly based on sections thereof. Yet, some other schemes also account for route specific difficulties: length, fitness required, remoteness, possibilities for retreat, &c. But looking at the guidepost again tells me:
The openstreetmap scheme of mapping difficulty on ways makes sense, especially regarding how the openstreetmap data is organized. An intermediate route shares a segment with a hard route that curiously is deemed easy. No problem. Mappers can judge from what they see on the ground and do not even need to know, whether they are on a route or just somewhere off.
No idea how that applies to skiing. Downhill slope maps look like topos known from climbing. What is the route there? Single segments?
I am with that, if I am reading this correctly as a plea for the node (PoI). In hiking domain that might be ladders, rungs, short safety ropes and so on. In Nordic skiing might be a short icy downhill. Some though insist, that mappers must not generalize and only map ways and consumers must have all the smarts.
I have mapped thousands of nordic/cross country pistes in OSM. My thoughts:
piste:grooming=* is a physical attributes which belongs on the way, not the route. Grooming is often not the same for different sections of route, typically if there are multiple routes across a wide network of pistes.
piste:difficulty=* is widely used for downhill pistes and it seems to reasonably well defined. I never map nordic/cross country pistes with this tag because it is not well defined for cross country. However, resorts sometimes refer to routes as being “easy” or for “advanced” skiers, so perhaps the tag would make sense on the route relation, as a rough guideline for the entire route. I have never seen “difficulty” indications used on the ground for different sections of a cross country piste (in my region, the Nordic countries).
Routes in my region are often identified by colour, so that could make sense on the route relation.
Example of routes in a resort below:
