Always trying, sometimes failing. Another try, a collage of paths I encountered:
Bottom example T6/UIAA II terrain. If you do not see the path there, I can paint it in for you. Might that be a candidate for the header picture of highway=path?
I put it onto commons, in case there are takers, File:Openstreetmap Paths Collage.jpg - Wikimedia Commons
I think itâs a good idea to warn data consumers that highway=path isnât always what the average person would consider a path. A gallery could work well.
The main risk with this is that the Wiki serves both as documentation for data consumers and instructions for mappers on how to map things.
We all agree that some scrambling and mountaineering routes over pathless terrain are currently tagged highway=path, but we donât all agree that they should be tagged highway=path. If we document them as an example of highway=path, then it should be accompanied by a warning that a lot of people donât think this is a good idea, and that the community is still discussing what to do about this.
I was contemplating that issue much like you do: Do not entice mappers to add more of that. But the poll, whether pathless UIAA III routes are mappable as paths in openstreetmap currently sits at 18:24. So people indeed want to do it. I just wonder, how many of those that voted yes ever managed that. Might be another poll?
Is it really a consensus? The problem section of that page has been there since the 2nd revision, in August 2008. Was it ever discussed, was was the âpathâ line just added by couple of people from a country with allemansrätten-adjacent access rights?
I think using highway=path is so popular because in most of the world, access rights are not described in laws and signposted along ways as much as in the UK, so mapping a small way as a path is suitably vague because access rights are not regulated.
Iâd certainly say it isnât safe for a router to assume that a bare highway=path, without access tags, is usable by bikes and horses.
Hi,
This has been hashed over already before (and, again, thank you to all concerned, this is a really interesting and intriguing discussion), but:
I donât disagree with this statement, but would contend its implications. If a way-element has only a highway=cycleway tag, one can infer in a truth-preserving manner that: a) someone using OSM once decided that the way is at least minimally passable for some type of bicycle, and that b) there are no legal restrictions on bicycle traffic[1]. Thatâs it.
That means that there really isâin the real worldâa huge overlap between many cycleways and paths (and some weirdos even tag these as highway=footway + bicycle=yes like the little perverts that they are 
). Even traffic-signed cycleways can be narrow/unpaved/in a horrible condition due to lack of maintenanace. None of that is inferable from the highway tag alone. We need the extra tags for all (or at least most) ways.
I, for example, drive a touring bicycle that has no suspension and relatively narrow tyres. I often do tours for a few hundred kilometers on it, and I scout possible routes using OSM. I want to know whether a way is paved (preferably with asphalt), traffic-signed or signposted (prerefably for bicycles), and segregated (from pedestrians or cars), roughly in that order.
Whenever I take a shortcut through unpaved roads, Iâm not overly concerned about the specific highway-type. I want to know whether I can legally drive there, whether the surface is passable even for narrow tyres and whether it is wide enough that I can yield to other traffic.
Hence the need for the extra tags.
I joined OSM in 2008 when the map was mostly white background and we were elated for any ways with just one tag on the map. Perhaps weâre now reaping what we sowed back then. But weâve also come a long way since. It may very well take another decade and a half for all the ways and paths to get these extra tags. I think that itâs just in the nature of the beast.
[1] The absence of legal restrictions might mean little worldwide, but in Finland it means diddly-squat. Out Freedom to Roam laws guarantee bicycle access to very nearly every road and path.
I agree, but a. routers do assume that, and b. access tags are not about safety. The mapper will have to add suitable characteristics to convey usability information to the data user. Thatâs why I think clear and unambiguous documentation of what to expect and where to look for extra information is important, to improve the actual (unsafe) situation.
a. Document the range of variants
b. List of extra tags characterizing key variants
c. Define/advise/suggest implicit tags and default values.
All in accordance with current practices and current base of information, in order to minimize damage and risks.
I have personal opinions and preferences on a., b. and c., but I think it is more important to document a set, any set, of a., b. and c.
Of course, data users are free to apply their own risk assessment and policy.
Bad routers do, sure 
When I say âsafeâ I mean in the sense of âa safe assumptionâ, not âpersonal safetyâ. In other words, a router can try and send bikes over every single highway=path, but it will result in bad routes.
No, itâs more than that. highway=cycleway is duck tagging. highway=path is granular tagging.
In other places it will result in excellent recreational routes, though sometimes an atb would be recommended. It depends on the extra tags such as pavement, smoothness, width. I would think, if I map a mud path through brushwood in hilly terrain (yes, we got that in Nederland), I will always add the extraâs. Bikes are often allowed there, in fact a lot of paths like that have bikes as their main purpose, while allowing pedestrian use. And horses, too, but for some reason horse riders prefer other paths.
Iâm not sure about camels and elephants, I admit.
Yes, but Iâd argue that itâs only very slightly more than that. A Duck Tagged highway=cyclewayâin itself and as the only tag on a wayâtells me absolutely nothing e.g. about its surface (or the condition of the surface) or its width.
Then how to tag a path with no rules about who and what can use it? Maybe like when you see something flying in the air: youâre sure itâs a bird, but you canât say if it looks and quacks like a duck. I think we should apply bird tagging then 
When one option is a potentially slightly more boring route and the other is getting sent over a cliff, itâs not hard to guess which option most people would prefer and is more useful for a general purpose router. If people specifically want a more scenic trip, they could use a router that is tuned for that.
have you seen page introduction? These are not really defaults, and there is no real consensus to treat them as defaults.
no (as mentioned in earlier post I would be happy to discuss it in a separate thread)
In Scotland, the Outdoor Access rules mean that footpath/bikepath/bridleway etc are rarely valid - highway=path is the correct tagâŚ
So my approach has been to add sac_scale, smoothness, width, surface, incline, trail_visibilty and other appropriate tags.
A curious issue arises in that informal mtb paths are still legally accessible as footpaths in either direction. In this instance, incline and sac_scale are a good indicator that whilst legally accessible, practically another route may be appropriate.
In short, I donât have an answer other than get out there, survey, and improve the supporting tags.
That is the kind of introduction with excess warnings, that helps no-one, added recently by someone with an allergy for defaults. For the country-specific default tables, correct, they make mappers leave out/remove necessary tags. World wide defaults on the contrary, help mapping consistently, efficient and directed at world wide data use. But that, again is another discussion.
If highway=path in general, osm-wide, is legally accessible for foot, bicycle and horse, unless otherwise indicated, and not accessible for motor vehicles. This means that in countries where legislation excludes bicycles, highway=path should be accompanied by bicycle=no, but motor_vehicle=no is implicit and does not need tagging.
In countries where regulations allow other vehicles, an extra tag should tell the data user that.
My current understanding is that this would affect a minority of legislations.
Not true! It tells you that the mapper considered the way to be suitable for bicycles. From this you can extrapolate that the surface is smooth enough and firm enough for bicycle wheels to roll on (though best to assume itâs on the more rugged side just in case) and that it is at least wide enough for a bicycle to pass. It doesnât tell you the exact surface and the exact width, but it does narrow things down a bit.
And highway=cycleway tells you what it is not.
From what Iâve seen, highway=cycleway implies that it is paved.
âAbsolutely nothingâ cannot be true. I sense hyperbole. Nothing can be as meaningless as a bare âhighway=pathâ.
