I wrote a simple question, as I thought, but it has already generated over 20! comments. Showing again how ambiguous many of the “rules” in osm are.
I map all combined cycle and foot ways (blue sign with a bicycle and pedestrian) with: highway=cycleway + foot=yes. Which I think is correct, and it’s used very often here in Sweden.
many cycleways in Nederland do have sidewalks, yes.
And on the path wiki there is a picture of a german cycleway with a sidewalk.
Yeah, simplicity is relative. Much depends on how people conceptualize “ways”. Local laws and customs inform people on these conceptualizations, and those differ between places.
In Homer’s Odyssey, Odysseus must atone Poseideon by walking with his oar held high above his head. He must walk so long inland that the people around him no longer wonder why he is holding an oar above his head, but instead wonder out loud why there’s a dude walking around with a spade or shovel (ἀθηρηλοιγός in the original Greek) held on top of his head.
Think cycleway vs path & designated vs yes, and oar vs shovel 
.
<german_mode>
No. 
But everything else is true.
</german_mode>
Are you referring to:
https://wiki.openstreetmap.org/wiki/File:Canal_Forsyth_Streets_bikeway_jeh.JPG
(but it is in the US, not Germany)
Standing at a crossroads, would a designation a.k.a. “Benutzungspflicht” be able to force me to go somewhere that I do not intend to go?
no it would not, and relevant as well, it is not an absolute requirement, it can be inverted when you are bringing large objects with you, in this case you will have to use the road as a pedestrian even when there is a sidewalk.
53 posts were split to a new topic: Difference between access=designated and access=yes
A path can become a cycleway though. Simply by putting up the cycleway sign, without altering any physical characteristics.
It has little to do with OSM and far more with that the real world rules and practices vary, sometimes widely, from country to country if not from state to state. It is unreasonable to assume that every legal and practical nuance can be modelled easily or even should be modelled in OSM.
I agree @Minh_Nguyen , and on top of that there seem to be two different understandings of “yes” at play here.
Some people seem to use “yes” to mean “allowed”. Others use it to mean “there’s a legal right to use this highway”.
There’s a difference between the two. If I have a legal right to use a footpath and someone places an obstruction on the footpath, such as a fence or gate, then the law gives me the means to have the obstruction removed, even if it was placed by the landowner themselves. There are regularly cases where landowners are fined for obstructing a public right of way (example). This is, from what I understand, what “public right of way” means in English law, and so it is what the yes tag means on footpaths in the British countryside. The yes here contrasts with permissive where the landowner is allowed to revoke permission at any time. It’s hard to imagine a stronger designation than yes. Did the German and American mappers who proposed =designated as an access tag know this? (As far as I know, neither Germany nor the US have this system of public rights of way)
The situation is quite different when you think of “yes” as merely meaning “legally allowed”. In Germany there’s a sign combination that means, on this footway considerate cycling is permitted. I can understand if there’s a reluctance to tag that as a cycleway. So it’s tagged highway=footway bicycle=yes instead. The yes here contrasts with designated, it seems to say “you’re merely allowed here (but don’t make yourselves at home!)”, when designated means “this is your path”.
The British use designation=* for this, which is very clever. access=yes alone doesn’t say anything about the right of way.
Indeed - but I was replying to a post about tagging in England. Had that been about walking somewhere a few miles north, in Scotland, where the law is similar to that in the Nordic countries, it would have been less relevant to ask “what is the legal right of access to walk there”, because the laws there are different to England.
Those are OSM tags - I’m guessing they mean that you saw “there is some sort of sign that says that it is currently OK to walk or cycle there”. Depending on the signage that might be e.g. bicycle=yes (or designated if you prefer) or it might just be bicycle=permissive.
The challenge with highway=path (see other threads in this forum ad nauseam) is that foot=designated is saying just “you’re allowed to walk here currently” but is not saying whether that’s an inalienable legal right or just based on the current whim of the landowner. That’s not a problem in jurisdictions with sensible access laws (like Scotland), but it is in England (and Wales).
Outside of England and Wales this same issue also affects “privately owned public space” in lots of other places, such as “modern” (i.e. private) urban redevelopment not by the city but by a private company - but that’s another discussion for another topic.
With no freedom to roam, the U.S. notion of public easements is probably the closest thing we have to the English system of public rights of way, but it’s much, much fuzzier. Have you ever noticed how, every time we import from a local U.S. government agency, we have to reproduce their disclaimer of warranty? That’s because no government GIS agency is willing to be the definitive source of easements and other property lines. With easements, everyone just hopes it doesn’t get to the point of needing to involve the courts. Needless to say, we stick to the signs when we can, even if they’re unstandardized and sometimes wrong.
I don’t think there ever was an “American proposal” for *=designated. In 2019, iD added a dedicated Cycle & Foot Path preset for shared use paths. This was partly a solution to a major routing problem: in regions where pedestrians are generally allowed to walk along bike paths, mappers often applied the Bike Path preset to shared use paths, based on the path’s commonly understood identity, without considering the allowed modes of transportation, but routers were unwilling to infer foot=yes from highway=cycleway.
Since the U.S. was one of the affected countries, iD’s American developers (and I myself) were acutely aware of the problem. However, the highway=*way *=designated combination was by no means an American tagging style. By 2019, there were already 341,000 occurrences of this combination worldwide, without the help of iD. Only about 11% of those ways were in the U.S.
I think this was more or less organic usage. As long as designated is positioned as the highest form of access, then mappers will gravitate toward it and find a use for it, even if they can’t comprehend the old highway=path proposal. German mappers used designated to overload the access keys with orthogonal traffic rules based on signs 240 and 241, and American mappers used it to overload the same keys with orthogonal design details. The common denominator is that designated means no other mode of transportation is as integral to the way.
I kind of see where they’re coming from. I mostly ignore the problem too. After all, in the absence of any mimicry, a feature’s design strongly influences its identity. Ideally, the primary feature tags should be enough to express that identity. However, a shared use path forces us to choose from multiple identities or find some workaround to avoid that choice.
Shared spaces defy neat categorization into primary feature types. Consider the café that doubles as a community bicycle repair workshop, or the town hall that doubles as a community center, or the pawnbroker that also repairs watches, or the money lender that also cashes cheques and sends remittances and packages overseas. Besides these extremely common examples, there’s no shortage of one-off edge cases, especially in small towns, such as the combination laundry–car wash–tanning salon–fitness center just blocks from the combination grocery–flooring–mattress–appliance–tool shed–candy store (Facebook page), a county away from a drive-through deli–liquor–live fishing bait store.
Common workarounds for these POIs include mapping multiple features next to each other (ignoring the one feature principle), falling back to a more generic primary feature tag like shop=yes, using a semicolon in the primary feature tag (knowing that data consumers will tend to drop all but the first value), or picking the one that best describes the place and adding secondary keys like mattress=yes alcohol=yes (ditto). Each of these workarounds is suboptimal, but OSM seems to cope with the situation because the problems are largely theoretical.
We can think of the highway=path *=designated scheme as just another workaround, but unfortunately, it conflicts with the access restriction tagging scheme. I think this problem is less theoretical because routers depend so much on access tagging. Personally, I’m more or less satisfied with either choosing from among highway=footway etc. or leaving a trail’s identity more ambiguous with highway=path *=yes. But it’s clearly unsatisfying to some mappers for various reasons. I hope one of the proposals for iterative refinement with path=* will help to close that gap.
Thanks again to @Minh_Nguyen and @Peter_Elderson for the interesting and well argued posts! There’s one thing that I take (a slight) issue with, however.
Again, much depends on local laws, but at least in the FtR world, it is arguably somewhat troublesome that we should have two different ways to tag one and the same entity, namely a shared-use way that’s not obviously useable for all types of bicycles (and thus not a clear candidate for Duck Tagging as cycleway).
I can well understand that instead of asking mappers to randomly choose between highway=cycleway + foot=yes and highway=footway + bicycle=yes, it is more consistent to use highway=path and the access rules it directly implies (again, in the FtR world).
Regarding =path with =designated tags, notice that in the North we do have signs like these:
They are not traffic signs, and thus do not engender rights of ways. They do, however, convey the idea of a preferred route. Notice that that particular sign has both a pedestrian and a bicycle pictured (versions with only one or the other exists too). To me, that naturally suggests both foot=designated and bicycle=designated, even to a =path way in OSM.
Also, most forest or park paths with these signs in Finland are maintained, and snowploughed (although they are the least priority in winter maintenance). A sizeable portion of paths without these sign in parks and forests get turned into ski tracks in winters that have lots of snowfall (in an ad hoc fashion, so they are not part of a stable ski network, and thus not part of a piste relation in OSM). Thus the designated key does convey information not easily conveyable with other tags. At least here in the exotic North .
highway=footway implies foot=designated. If there is also permissive, by rule or by sign, you can add foot=permissive or customer, because that is the more restrictive access restriction. It would still be a footway designated to walkers, just with an extra restriction.
You don’t tag the rule, you tag the effect of the rule. If it’s a footway, tag highway=footway, which implies foot=designated. If it’s permissive, because of a signe or a rule, add access=permissive. If there is no path at all, just freedom to go anywhere you like, there is no path to tag, and you might leave it at that or find another object to attach this tag to: a zone, an area. But, from what I have seen, the PROW rule says something like “you have a right to cross private land, but you have to use the provisions made for that” which are paths and gates, and no unnecessary swerving! So you don’t just climb over any stone wall, fence or hedge. Correct?
The long ongoing public footpaths are routes I think, consisting of permissive footways and walkable sections of other ways. Route relation tagged as route=foot. I don’t know a specific tag to indicate the designation of the string of ways as an official public footpath, but you always have from=, to=, via=, description=.
I think this way the effects of the typical English PROW rules are translated into OSM-wide understood tags. Maybe some aspects are still not captured - but not everything needs to be captured, unless it seriously misses the effects they have on the ways people travel, and on the rendering and routing of the ways.
Nederland has such rules where it comes to different kind of signed cycleway (mandatory cycleways, mandatory cycleandmoped-ways, non-mandatory cycleways) and there are no definihg classification tags for that. So we just tag the impact of the signs on access for regular bicycles, mofa’s, mopeds, and electric mofa’s (designated, yes, or no) where applicable, and forget the mandatory aspect because there just isn’t a generally understood tag for that. But we can and do tag any alternatives as *=no or *=use_sidepath.
I don’t quite see why the FtR angle is relevant for this discussion, at least I can’t see what it has to do with the original question. FtR is about the freedom to use (on foot) ways that without it would be considered trespassing, it isn’t FtiTL (Freedom to ignore Traffic Laws).
Or put differently, even in FtR countries you are not free to go hiking on motorways.
Well, we tag both. highway=footway for the effect of the rule, designation=public_footpath for the rule.
This reflects the facts on the ground - a walkable path and a signpost saying “public footpath”. It also means that savvy renderers/routers can infer some of the lesser nuances without these having to be translated into OSM tags. (For example, bikes are allowed on public bridleways, but bike racing isn’t!)
According to the author of the proposal, designated means:
In general, “specially designated” means that there is a sign there saying something like “bicycles allowed”, or a bicycle icon, or something along those lines.
The reason highway=path introduced access=designated is, because by default “you are not forbidden to walk, bike or ride on a path”. Not forbidden → yes. Allowed → designated. Clearly something that would belong into access tags.
Do you mean that these shared use paths could be rugged paths only for hiking and mountain biking? I thought bicycle=* is essentially about bicycles in general, whereas one would need to tag mtb=* to indicate suitability for mountain bikes specifically.
This is sign number 645, “Direction sign for light traffic (pedestrian and cycle traffic)”, correct? It is a traffic sign, just not a mandatory or prohibitory sign.
The situation is essentially the same in the U.S.: the only standard signs that positively indicate suitability for cyclists and pedestrians are classified not as regulatory signs but as guide signs – that is, mere recommendations. The standard signs are extremely rare, but the more common nonstandard signs would lack regulatory effect as well. We use foot=designated and bicycle=designated primarily because the value exists, but not because it’s a special “yes and then some” level of access.
I understand that sometimes it isn’t possible to distinguish a cycleway from a footway from the signs alone. In some regions, we can use cues other than signs or laws to make that determination, such as names or cycle barriers. There certainly is a gray area in which neither mode of transportation clearly dominates, in which case highway=path may be appropriate. Perhaps that gray area is the norm in Finland.
If anything, I suspect mappers would find it more intuitive if we were to allow an only value (which could appear on multiple keys in the case of a shared use path). Data consumers might find it more intuitive too, judging from OSRM’s misinterpretation of designated on some access keys. But that might be head-spinning for folks who are already having their world turned upside-down by this discussion. 
I think this gets to the heart of the matter: in some regions like the U.S., this is a distinction without a difference. Any gray area between “not forbidden” and “allowed” is not the stuff of laws; rather, it’s where we start debating whether it’s safe to walk along a busy, poorly designed road or drive on a track of questionable quality.
Its “officially and verifiably assigned to this mode of travel”. Of course that always encompasses specific access, and usually it is clear by signage or markings, and usually a layout and design aspect, subject to local lwas, customs and signs.
How often does this exception (designated+some other access value for one transport mode, for highway=path) cause a tagging problem? I just described one case where it is solved acceptably.
