It might be more minor in Thailand. I know it’s a big deal in Japan (comes up in discussions with delivery companies) and Vietnam (where plenty of roads are designed for motorcycles rather than cars anyways).
No, as I pointed out above, access=private would make no sense if access tagging were solely about physical accessibility. Let’s continue the discussion about narrow roads in the other thread; I’ve already finished making my point that physical accessibility matters.
I would not consider the snow, from what it looks, this is not designed to be locked any time or prevent access for people, so I think foot=yes would be suitable (but you know the situation better than me, I have to guess from the picture).
Tagging such would NOT map legal rights of way, but the physical ability.
From this topic I infer, foot=yes would not say anything about passibility. So I went with locked=no.
PS: The path the gate is on has sac_scale tagged, so foot=yes is implicit, on the path, and I suppose on all the hurdles.
What is the legal situation, may you cross this gate on foot or not?
I am not sure if there is consensus (yet) to start documenting on the Wiki.
Let’s look again at the example that led to this thread: bollards that block cars from driving on a cycleway but are far enough apart so mopeds fit through. The Dutch agree that even when mopeds are allowed on the cycleway, barrier=bollard by itself is sufficient, there’s no need to add moped=yes, the node should not prevent the moped from being routed along the cycleway. Elsewhere (e.g. Germany, UK) where mopeds are not allowed on cycleways, mappers also agree that a simple barrier=bollard node is fine: even though mopeds aren’t allowed, we don’t tag it moped=no (nor do we tag it as moped=yes just because mopeds fit past). So there is no real disagreement about the tagging, it’s just a question of how we document how we map on the Wiki.
Then there are cases where there are actually different tagging approaches. They are probably not that common but still common enough to think about. A landowner has illegally put up a locked gate across a public footpath, is that foot=no or is it foot=yes but locked=yes? A fence has appeared and I have absolutely no idea if it’s still my legal right to access the land behind it - I am not a lawyer - but what I do know is it prevents me from accessing it without climbing over it, is it OK to still tag foot=no? You have to carry your mountain bike over a stile but you can legally continue on the other side, should I tag the stile as bicycle=no?
How do we document this situation on the Wiki and explain it to new mappers and data consumers? Do we say
maxwidth:physical, locked, …)yes means allowed and possible, and no means either illegal or impossibleIf we don’t make a decision we end up with (5): the Wiki says completely contradictory things in different places 
I agree these are worth discussing. For clarity I think the examples should also include cases where signage might be illegal, either because somebody puts in place a sign where they have no legal right to do so, or because they destroy legal signage.
I can think of an example in Ireland where a dispute between a landowner and local authority about access to footpaths went all the way to the Supreme Court, a 58 day (!) trial, and several million euro of legal costs. During a period of around 5 years, it would have been impossible for a mapper to determine the legal access rights.
For these reasons I don’t see the distinction between legal and physical access as clearly as some people seem to. As a practical matter, don’t we generally have to assume that physical barriers reflect the legal situation, unless there is specific reason to think otherwise?
Note I am talking here about intentional physical restrictions, not (for example) a path where bicycle access is legal but the condition of the path makes it unsuitable.
I think so, but it seems like there’s some contention over whether mappers are also responsible for evaluating the effect of a barrier’s design. A closed gate is relatively straightforward – literally an open-and-shut case – compared to a bollard or other permeable barrier. As much as I hope that more routers adopt locked=* over time, I can easily imagine a long tail of other micromapping that routers would now need to understand if we handle physical passability with the same precise fervor as legal permission.
Incidentally, where was the concern about ambiguity when we decided that wheelchair=* would simultaneously describe physical passability (by virtue of values like limited) and legal permission (by virtue of its inclusion in the access key list)? Does this sign indicate a lack of permission or of physical access?
where is it? And what is the local context?
There’s a description on the Flickr page. I think it’s somewhere around here. Apparently the walkway goes up a slope that’s uncomfortably steep for most wheelchairs, so the college repurposed a sign for a disabled parking space to discourage wheelchair users.
Probably on a mailing list somewhere
Around about that time (2010) there was also use (in England at least) of “bicycle=no” on signed cycle routes that you may not want to take a regular road bike on. That usage has significantly decreased as people have discovered other tags to express the same issues.
The context of that sign makes clear it’s a lack of physical access in this case. Whether it’s a lack of legal access as well is likely dependent on local law around disabled access - at least in the UK and I believe the US (where this sign is), there’s explicit legislation that tries to ensure legal wheelchair access in a bunch of places where physical access might be difficult or impossible.
Huh, I’ve never thought of wheelchair as an access key, precisely because it’s so commonly used to mean “suitable for wheelchairs” and that is also the meaning documented in the Wiki. I didn’t know it was listed on the Key:access page. Should we just remove it from there and instead add a note that it works differently to the legal access tags?
There’s more where that came from. The page on mtb=*, which is also in the list, has a whole treatise on whether it accounts for legality. dog=* tends to be used in a less legalistic fashion on amenities and shops. 4wd_only=* is an oddball in multiple respects.
Best to avoid getting involved? A way with incline > 6% legally is wheelchair=no in the sense of “impassable” in the area of my local knowledge. I rather tag incline than wheelchair and leave it up to wheelchair-client to decide the other.
Unfortunately, I know of no such regulation over here in the U.S. Here, wheelchairs are considered personal mobility devices, not vehicles. It’s no accident that the photo I posted is a repurposed sign for a disabled parking space. The laws affecting wheelchairs are generally intended to enable or empower rather than restrict. So even in a wilderness area, you can legally take a wheelchair anywhere you can go as a pedestrian – even up a scramble, if you can manage. Thus, the ad hoc no-wheelchair sign on this trail is functionally equivalent to a wheelchair-related sign on an escalator in the subway. There might not be a law about it, but it’s “a difference without a distinction”.
wheelchair=no appears some 271,000 times in the U.S. Many of these occurrences are on features that seldom have signs of any kind, like sidewalks or curb ramps. In my experience, most of the time that Slack users mention wheelchair=no, it’s because of pedestrian infrastructure that is comically inaccessible, an all-too-familiar situation in this country. No wheelchair user would complain about an OSM data consumer leading them away from one of these walkways or curbs.
(drifting offtopic here, but) I’m actually surprised that the Americans with Disabilities Act doesn’t cover that, especially as there is apparently case law around it now.
(in case it isn’t obvious, I’m neither American nor a lawyer)
(I’m not an expert on this subject, but since you asked…) At most, in a built environment, the ADA would require the builder to ensure some access path but wouldn’t require that all paths be accessible. At the same time, it doesn’t ban wheelchairs from anywhere, because that wasn’t the law’s purpose. I was trying to make the point that, in the absence of such formal regulation, whatever someone posts as guidance to wheelchair users is tantamount to a taggable access restriction.
Needless to say, it would be quite impractical and undesirable for the Forest Service or National Park Service to level every trail for the use of conventional wheelchairs. The mountaineers among you would have no reason left to visit!
It should not be a surprise, see here, I started already doing so.
I did not touch the access page yet because I was doing some research on how access tags are used. This whole discussion could in my opinion use more of that, I get that some people do not like “access-keys” on barriers but that is quite academic for me unless there is an accepted alternative.
The research gave a nice find, emergency=fire_hydrant.
How are going to document that on Key:access#List of possible values? 
How do we document this situation on the Wiki and explain it to new mappers and data consumers?
I think there are only a few people that want to make a systematic difference between legal access, physical access and designated use. Yes, for roads and paths it should be done based on legal access but that is logical because doing it based on physical access does not make sense.
For barriers, entrances and fords most often there are no legal limitations/sign’s but there are physical limitations so the pragmatic choice is made to use that. If there is a sign that can/should be used instead.
Another nice find from the research is that if you look at the usage stats, highway=service (in way context) has highest access count but the second one is highway=bus_stop (in node context), so highway=bus_stop + bus=yes. That is neither legal nor physical…
So yes, the access page can use improvements but I see that a smaller then the alternative of redefining the current usage, documenting that, and, the largest effort, convincing current mappers that is a good idea.
Does it really matter to data consumers that something is not accessible because of legal restrictions of physical restrictions? For data consumers, better worry about something that is much more problematic in my opinion, highway=crossing + bicycle=no, see here.
Interesting you should mention this – some routing profiles have to specifically special-case highway=crossing nodes to ignore their access tags, because they already honor access tags on any other node, with or without barrier=*:
I did not touch the access page yet because I was doing some research on how access tags are used. This whole discussion could in my opinion use more of that, I get that some people do not like “access-keys” on barriers but that is quite academic for me unless there is an accepted alternative.
there are also other cases where access-like tags are used, e.g. amenity=parking motorcar=designated does not mean you may not walk there or ride a bike, it means parking is for cars.
Indeed, what makes a tag an acces-tag? Being mentioned on some list?
Personally, I never understood wheelchair=* an access tag, rather about accessibility than about access, kind of a reverse sac_scale – as if hiking=yes would say, no need to lift legs more than an inch, or so. Or the not-existing mountain_hiking=yes term along that line would say: You have to prepared to lift legs higher than your knees, etc.
Regarding the use on barriers - I often wonder, how people come to know? By being prompted by an editor preset?