brought up many valid points regarding highway=path (and adjacent tags).
My takeaways:
There is a wide agreement that current situation of highway=path is not great (if dissent exists to that general sentiment, it is rare). It applies to things that are very distinct from each other.
Some people think that there does not really need to be any proposal for new or different tags and the only things that need to be done is to do actual mapping, ie. add secondarytags to existing highway=paths without them.
A subset of such people (or all of them?) would like to agree on a default set of values of tags so that only outliers would need to be specifically tagged. I personally would agree with this but I cannot again gauge if such a proposal would get enough support. It would probably in its effects consist of modifying the wiki and maybe trying to affect the consumers to such an effect. I am undecided if it equals to redefining the tag.
Another group thinks this solution is non-workable (people and data consumers are lazy, there will always be ways just with tag highway=path, additional tags are difficult to apply and a lot of work that people will not do, it is unnecessarily complicated for everybody). It seems to me that this is a stronger camp, but it is hard to gauge without some voting/polling process, especially with my confirmation bias of falling into this camp :-).
Highway=path cannot be just redefined, it is too varied and too used and it would be too much of a mess. Enough people agree on this to make it a rock-solid conclusion.
There is not much agreement on how path relates to cycleway/footpath/bridleway. Some people think they are a subclas of path, some people think they are completely orthogonal. I personally think they are a subclass (i.e. highway=path with additional tags can express them, which some people think is a terrible state of affair and I agree).
A popular opinion seems to be that whereas there are many different primary tags for ways for cars, there is much less of them for ways that are not for cars, causing the current mess. A solution to that seems to be to devise new primary tags that would address this issue.
Some or most people agree that the current classification of highways mixes functional and descriptive characteristics. Some people think only one of these aspects should be used.
Typology of ways not used or designed by cars
Roads that look like roads but are too narrow for them, aka possibly alleyways, usually in urban settings (they oppose footways): Proposal:Alleyway - OpenStreetMap Wiki
Ways that are either closed to cars legally or not used by them because it is physically impossible, but that are otherwise very easy to ride a motorcycle on and usually also open for bikes and pedestrians etc. motorcycleway - as of now, these are usually (?) highway=path with access and designation tags.
Shared_use open and suitable for all sorts of non-motorized traffic (but allowing e-bikes etc.). Especially in Germany but possibly worldwide, these are now usually tagged highway=path + foot=designated + bicycle=designated. (Previous suggestions were two_wheel or mup as in multipurpose).
singletrack as wikipedia knows it â ways meant for bikes, usually downhil/uphill, not entirely safe for pedestrians and not terribly suitable for walking, usually somewhat build
snowmobile â I am told there are ways only usable by snowmobiles in the winter when snow falls but not in summer with no snow (though could also be pathless and snowmobile=yes, I lean to have a separate tag)
trails â what most people would recognize as hiking trails (or similar in parts of the world when hiking is not a thing but getting from place A to place B is), in SAC scale from hiking to demanding_mountain_hiking or alpine_hiking typically
pathless - routes where people regularly walk but do not manage to leave traces because of the ocean/too rocky terrain/wetlands (not to be used for links over squares made for routing purposes, this takes no position on Tag:highway=footway - OpenStreetMap Wiki)
scramble as was proposed, though relation with pathless would need to be discussed
climbing/mountaneering a discussion wherther it could be covered by pathless or scramble orif it is suitable at all would need to be conducted.
unknown pathways whose exact nature is unknown, would work like highway=road now (for imports, armchair mapping), also see here
I am sure I am missing something. The definitions are rudimentary and would need to be specified in a much greater detail.
New primary tag path/pathway=
All the bold words today could be values to new key pathway/path= (I prefer pathway).
Relation to highway=path
They could live together with highway=path and other highways. The goal would be to eventually switchover at some future date.
Transition
This would be a multiyear (possibly a decade long?) effort. At first, if voted into reality, just the new tags would be introduced and incorporated into the wiki as alternative that can (or should?) be used alongside the old way of doing things. After some time, when usage would start growing, wiki could be modified to officially recomend them and discourage the old ones from being used. Software like JOSM/iD/Vespucci et al. would be asked to incorporate them. Consumers would be notified of them. At some point, some automatic edits could be discussed, so for example all highway=footway would also become pathway=footway. Years from that point, highway=footway could be removed from ways that would be tagged as pathway=footway (and similarly for other tags).
What is important, this would be a long process. It would need a lot of work. The scope could be less - we can keep some highway tags that are well estabilished (though I think the new pathway namespace would make things easier for everybody in the long run). The main thing is that for some time, usage of both tags would be encouraged and welcome. That should make the transition much smoother. I think that the eventual goal should be retiring highway=path - possible some time after 2030. But even if that does not happen, the new pathway=* could start bringing some sense into the current mess.
Any proposal should include a timeline for the transition, as this could not happen all at once.
I think the biggest issue here is that not enough outdoor renderers support tags like access, sac_scale, and trail_visibility. If they did, a lot of real-world problems like trespassing and unnecessary rescues could be avoided.
But if end consumers stay lazy, just inventing new high-level tags might lead to the same old results. On the flip side, introducing new highway tags could push things forward since mappers tend to be lazy too, and theyâll likely pressure renderers to display them differently.
Another option that hasnât been mentioned is to slowly break down niche aspects of highway=path into new tags, like urban pathways, mountaineering/scrambling routes, or golf cart paths, and leave highway=path for trails.
That is covered by my point six. That would have been my initial approach, but my understanding of the previous discussion is that it is a non-starter.
Getting a bit lost in all the discussion on the same topic.
Iâd add a reference to what sounds more-or-less the same thing, perhaps a bit broader:
While Iâm a fan of scrambling, I have an issue with it as a separate category. It covers both hiking scale T4-T6 and the climbing scale I-III. Technically, it is already possible to locate it by tags.
Another practical issue is that it will necessarily overlap the actual climbing routes of lower grades. Iâve climbed the last sections of a multi-pitch climb without a rope because it was only II+, for example. Is it a climbing or a scrambling route? Or both?
Which leads us to climbing. The climbing routes may be represented by pathless but that is also incomplete and misleading. Most trad and sport routes are actually not pathless. Well, depends on the definitions on which we canât really agree, it seems. The routes are usually very well known. They are published in guide books and lately on web sites (the crag, bergsteigen, etc.). Sometimes they have a name plate or the name is just written with a paint. In Austria they are usually bolted and have anchors at the end of each pitch. This really depends on a country as bolting is illegal in some countries but the main routes are still well known.
Definitely not pathless but neither a path. It is something specific - a climbing route. And Iâd love to see and map more of them on OSM.
Currently, people use all kind of tags to add them (i.e. Paths with T6 sac_scale for a grade VII climbing route) and I think they deserve a separate classification.
As far as I know, quite a few of the real-world problems have been with general purpose maps (and not just ones based on OSM), rather than specialist outdoor maps. And the problems donât necessarily arise in barely-charted wilderness where you might think nobody would use a generic map - many big cities have dangerous trails (if you donât know what you are doing) very close to places where the general public goes for a Sunday stroll. That has sometimes been a main motivation for proposing a new main tag for more difficult paths.
It might be too early(!) in this discussion to try that kind of summary, but Iâll try just in case: so far we have mentioned several strategies for going forward:
eating popcorn (ie doing nothing)
extending the set of values of highway=*
creating a new ânamespaceâ besides highway=*
re-root the highway class, making it a subclass of route and creating other subclasses (e.g. climbing routes and other pathless routes)
BTW, climbing and mountaineering are not directly comparable. Most of the climbing routes I know are not even in the mountains!
I assume mountaineering (or alpinism?) would include glacier travel. Other than that, it includes walking in pathless terrain, hiking on marked paths, via ferratas, climbing with and without ropes (so, different grades), walking and climbing glaciers (different inclination, equipment, and technique), climbing and descending snow fields, ice, and firn.
At this moment, Iâm not even sure how to tag all these. Well, hiking and ferratas are safe, at least.
One I particularly had in mind was a discussion on this forum a couple of years back about Table Mountain, close to the city of South Africa.
Another example is this one not far from Vancouver. While this might be a bit closer to âwildernessâ, it seems that the hikers originally followed a reasonably easy and popular hiking trail but ended up on a highly technical trail that looked the same on their map.
Also from Vancouver, and apparently close to the northern suburbs of the city, is this incident. The fact this was blamed on Google Maps rather than OSM doesnât affect my point about the dangers of general-purpose maps.
In brief: a youngster going to a party just outside the city of Grenoble was led by Google Maps to think he could follow hiking signs with his bike, and ended up spending the night clinging to grass stems for dear life.
I heard that in Grenoble, the helicopters are on call not on the ground but in the air, so to be there quicker on the site of accident. Grenoble is partner-city of my own.
I remember a mountain rescue accident likely caused by: âShow the known way up to the closest point to the PoI targeted and then draw an arc to there.â Unfortunately, the arc coincided with a hunters path, not even mapped in OSM. Unfortunately, OSM data consumers follow that guessing, even though they should know better. I observed in places, they go down mapped cliffs as if nothing there.
This reinforce the idea that we need new highway tag(s) for scrambling/climbing/mountaineering routes especially those requiring specialized equipments.
Even this would require data-consumers to treat them the way âyouâ want them to treat them. Nothing would stop OSMcarto or routers,⊠to just add those values to their existing list highway=path/footway/cycleway At the same time everyone is able to treat ways different based on already existing tags.
Adding plenty of high-level-tags doesnât solve the problem either and if every issue is solved in that way we will end up at some point with highway=excelent_asphalt_painted_red_2m_wide_to_be_used_from_cyclist_and_pedestrians
Not likely! These tags donât exist yet, so theyâd need to be added from scratch, making it harder for them to just get lumped in with others. Yeah, a renderer might ignore it and make it look like something else, but itâs way easier for them to fix one new tag than to mess with a bunch of sub-tags. Plus, if users complain, theyâd have to pay attention!
Letâs be real âhow many renderers actually support tags like trail_visibility or smoothness? Not many, if any, because itâs tough for developers to juggle the complexity of combining tags. Thatâs why high-level tags can be easier for them to implement and support.
ah this is what âyouâ are afraid of! I promise, just a few smart tags, not a novel. We just need to fix the gaps so we donât have to jam a triangle into a circle.
Yes⊠and how many support surface? So I will need a highway="unpaved_cycleway_with_bad_smoothness" and my friend likes gravel tracks, so he need highway="paved_cycleway_with_good_smoothness" because renderer show no difference, just a blue-dotted line. I donât see how you will manage to solve the issue of Renderer not using secondary tags with:
seems to summarize the position that most of those involved in the discussions would agree on.
Meaning, the introduction of a path=* classification in addition to the highway=path, instead of it being a new primary tag, would
satisfy the need to categorise the various types of paths,
not have many objections from those who donât need them
As it would be a secondary tag, I think path would be more appropriate than pathway.
Then the text about deprecating the highway=footway (and others) would stay as-is. All clearly identified versions of paths could be added separately, including climbing, pathless, cycleway, motorcycleway, snorkelling route, and all the ones listed in the first post.
It looks to me that this is the growing consensus at the moment. Is that a fair assessment?
Should we narrow down the potential proposal to this? Meaning, trimming down the original post to focus on the sub-tag option.
Then, after some weeks of being open for comments, proceed with it becoming an actual proposal?
At the same time everyone is able to treat ways different based on already existing tags.
ah this is what âyouâ are afraid of! I promise, just a few smart tags, not a novel. We just need to fix the gaps so we donât have to jam a triangle into a circle.