Does anyone use highway=footway with Access Tags for generic urban paths?

General talk Tagging general discussion
1

[PS: Paths topic haters, please move on :wink: ]

Pathways are generally classified by both construction type AND primary usage, but highway=path is applied generically, regardless of construction, leading to inconsistencies.

For example, a trail designed for MTB downhill riders is never tagged as highway=cycleway; instead, it is always highway=path + bicycle=designated. Similarly, in most cases (with some country-specific exceptions), a hiking-only trail is tagged as highway=path + foot=designated, not highway=footway.

In Thailand, where motorcycles are widely used, it is often unclear whether a constructed urban pathway is intended exclusively for pedestrians. In the absence of signs or physical barriers (such as curbs, steps, or bollards), motorcycles often becoming the primary users.

However, tagging constructed urban paths as highway=path + motorcycle=yes, as we have opted to do in the Thailand community, raises concerns:

  • It confuses new mappers—why tag something that looks like a footway as a path?
  • To most end-users, it is rendered as an outdoor trail, which represents the vast majority of highway=path usage.

This issue is similar to the segregated urban footway/cycleway dilemma in Europe.

Alternative Tagging?

In the absence of a dedicated tag for generic constructed urban paths, would a preset like highway=footway + bicycle=yes + motorcycle=yes be a possible alternative?

Other countries in the region use:

  • highway=service + service=alley
  • highway=residential + motorcar=no

However, these tagging approaches are even more controversial. And not all narrow urban pathways are found between buildings or provide access to residences:

Screen Shot 2568-03-23 at 09.37.58
Screen Shot 2568-03-23 at 09.37.58446×526 249 KB

Screen Shot 2568-03-23 at 09.38.06
Screen Shot 2568-03-23 at 09.38.06844×690 56.9 KB

Screen Shot 2568-03-23 at 09.38.22
Screen Shot 2568-03-23 at 09.38.22500×482 366 KB

Screen Shot 2568-03-23 at 09.38.30
Screen Shot 2568-03-23 at 09.38.30496×512 368 KB

2 Likes
2

Interesting! (and noted that we have “paths topic haters”. :grin:) I’m a total newb to OSM but decided to get involved to map out my local pedestrian network – streets are fine, but sidewalks don’t show (and we can’t assume they exist).

Anyway, I’m VERY interested in such topics and having lived in other countries, understand that there are substantial differences in expectations about who can use a given path.

If these photos were for locations near me (Oregon USA), I’d say all but the one with the silver vehicle could possibly accommodate motorized vehicles (itself an increasingly complex topic now that we have electric bicycles and scooters competing for space) if the path is wide enough. However, for the one with the silver vehicle, I would say the local expectation is that this would primarily be a “sidewalk” for pedestrians (though bicycles would also be welcome). There is an additional wrinkle in ownership: paths on private property will potentially have limitations though this may also be true for public facilities. There are a number of sidewalks locally that are technically reserved for foot traffic only though bicyclists tend to ignore such restrictions.

1 Like
3

Not because the topic itself is risible, but rather because a new thread about paths for bicycles or motorcycles has come up every few weeks for several months now. None of the threads have arrived at a satisfactory conclusion, so we keep going in circles.

There is some contention about these tags in Europe, but I wasn’t aware of it having anything to do with motorcycles.

This seems to be a trailing thought: what is the controversy about? As you point out, other countries are using these tagging combinations. Who opposes this practice and why?

4 Likes
4

To be fair, many of these discussions started because of controversial edits to the global wiki. This time, I’m simply gathering global feedback (and, apparently, ongoing hostility) to bring back to the Thailand community for discussion on potential changes.

This isn’t about motorcycles—it’s about the use of highway=path for something constructed.

These cases are well-documented:

highway=service + service=alley

  • Minor roads, not pathways.
  • Doesn’t match the definition of highway=service.

highway=residential + motorcar=no

  • Do most of these pictures really show roads giving access to residences? :laughing:
  • How does rendering a 1m-wide pathway as a residential road help mappers or end-users?
  • Using unsigned legal access tags for physical restrictions is documented now as controversial.

According to Grab, there is no consensus in the region for pathways too narrow for motorcars, but the majority of SE Asia use highway=residential for narrow roads wide enough for motorcars. Thailand community has already voted against the use of service=alley for that purpose (after years of edit wars), but I am open to use that tag for narrow constructed generic pathway if our community prefers it to footway or path.

5

The original post could’ve be worded and titled more clearly. I don’t think highway=path is exclusive to hewn paths. As you’ve pointed out, it’s being used for shared use paths in Europe, and it’s also used very deliberately for golf cart paths. Conversely, bridle paths and towpaths are most often hewn, not constructed, but they get their own highway=bridleway and towpath=yes tags anyways.

I’d agree that such a broad definition of highway=path leaves editors and data consumers unable to infer much about the path that would help users intuit its characteristics. Ideally there would be some distinction between “urban” constructed paths and “rural” hewn paths.[1] Some have floated the idea of a new highway=trail tag for the latter, taking inspiration from North American terminology. But introducing a new highway=* value would cause quite a bit of disruption, so we might need to settle for some sort of iterative refinement like path=trail.

So this answers your question:

with a resounding no.

It isn’t hostility or hatred to explain that all of these points have already been made several times in recent months. I share your frustration that a simple solution has not been forthcoming. We seem to have boxed ourselves into a corner. Which definitions or principles can we be less dogmatic about in order to find a way forward?

  1. Quotation marks because hewn paths can certainly occur in a less regulated urban environment or an urban park, while constructed long-distance paths mainly cross rural areas. ↩

2 Likes
6

I was confused about this too, but @Minh_Nguyen’s question settled it (and agree that the first post is somewhat meandering).

Well, I’ll meander along too :slight_smile: . Also, my reactions are Eurocentric and possibly not generalizable outside it. But here goes:

Isn’t it actually a completely open question to whom such a way is “primarily” “intended” for? That highway=path as the top-level tag catches this ambiguity is in some sense completely appropriate, in my opinion.

Again, speaking for infrastructure rich parts of Europe (and particularly places with robust Freedom to Roam laws) ways that lack signage are legally allowed for all non-motorized forms of travel. It’s up to you and your hurry/choice of footwear/width of bicycle tyres/type or absence of suspension whether you want to walk/cycle on any particular non-signed path.

Couldn’t informal=yes be used to express this succinctly? That is, to pick out the “rural” (un?)hewn paths? Other secondary tags add further clues. A path that has a surface= tag in the =paved superset (e.g. =asphalt, =fine_gravel, etc.) has to be constructed, no? If it is neglected and falls into an advanced state of disrepair, the smoothness= tag can be used to express that fact. Otherwise a width= estimate tells you if you can physically fit e.g. a bicycle through it.

7

Again, speaking for infrastructure rich parts of Europe (and particularly places with robust Freedom to Roam laws) ways that lack signage are legally allowed for all non-motorized forms of travel.

IMHO this is part of the issue, probably that ship has sailed, but why would we imply restrictions for motorized vehicles when there are no signs at all, as a global standard?

2 Likes
8

Even without the freedom to roam, non-motorized transportation is usually less tightly regulated because there’s less of a safety concern. As long as you have the right to be on the property (especially public property), modes of transportation may be governed as much by norms as by laws.

I haven’t even gotten to informal=yes yet. Frankly I’m struggling to even come up with accurate words for this distinction other than path and trail, which apparently only North Americans recognize intuitively. Constructed versus hewn doesn’t really work either.

From my perspective, this is a stereotypical urban path. It was built in a park in the middle of a city. I’d tag it as highway=footway because it’s a footpath – designed primarily for pedestrians – even if children aren’t prevented from riding bicycles and tricycles on it.

This is a golf cart path built in a country setting, highway=path golf_cart=yes because we don’t have a more suitable top-level tag:

And this is a stereotypical rural trail. Except it isn’t rural because it’s also in a park in a city. This asphalt trail was laid, not so much built. It isn’t informal=yes because it has an operator=*.

Americans would also call this a trail, but we already have highway=cycleway for cases like this:

Ah, finally, primitive trails out in the wilderness – but still not informal=yes:

1 Like
9

I’ll fairly often use highway=footway, bicycle=yes for shared-use paths where bikes are permitted but the construction standards are those of a footway (narrower, usually without any particular accommodation for cyclists at junctions/bus stops/etc.).

3 Likes
10

Oh, this is absolutely true, and a point raised by @julcnx too for sure! My myopic point was that FtR laws specifically apply only to non-motorized travel (i.e., one cannot forbid such travel except in few special cases). There are also non-signed ways that do not permit motorized travel because every way leading to them has a sign forbidding motor vehicles, for instance.

Would you say, or intuit, that both ‘paths’ and ‘trails’ can be either informal=yes or informal=no? Or can only one of them be ‘informal’ in some sense? What would be the on-the-ground-verifiable differences between ‘paths’ and ‘trails’?

Oh, yeah! In the absence of explicit signage, things like curbs (which @julcnx also mentioned) in urban environments are also big clues for the intended primary users of the ways. If you need to cross a curb to get to a way, it’s probably not intended for bicyclists, for example.

However, in the absence of even clues like that, it’s really difficult to assess the intended users (which in my mind implies =path). But local laws differ and everybody’s mileage varies.

1 Like
11

If we were to accept a path/trail distinction based on ordinary American English, then it’s orthogonal; either can be formal or informal. A desire path is still a path because of the urban environment, despite its hewn quality.

https://www.instagram.com/p/DE24GKTNRAs/

“Trail” connotes something in a wilder, less sterile environment. It’s akin to the difference between a street and a road – a fun discussion to have somewhere else.

However, highway=path is not based on American English. It’s inspired by British English, which doesn’t distinguish trails at all. Then it took on a distinctly OSM meaning because of regrettable technical decisions that people made years ago, and we’re still continually renegotiating its true meaning.

The original post focused on urban paths that definitely aren’t trails and definitely aren’t informal. If highway=service or highway=residential would be an overstatement, then highway=path or highway=*way would be the other options. If @julcnx and the Thai community care about motorcyclability, they’ll need to either use motorcycle=yes based on practical suitability, not legal right, or accept that data consumers will have to infer this quality from a variety of physical characteristics like surface and width. Something has to give.

More than that, norms differ from one locality to another. How can I presume this footpath to be a highway=footway just by glancing at a photo without even bothering to look it up in OSM? It’s located in an urban park (leisure=park) and it looks like any ordinary path I’ve strolled along in a park.

In a different context, a path with the same surface and width running parallel to a street might be either a sidewalk (highway=footway) or a sidepath (which would be highway=cycleway locally but highway=path in some other countries for reasons). That determination can be based on a variety of factors like signs, aids, and connections to routes. In the absence of any of these clues, I’d conservatively fall back to highway=footway because pedestrians’ needs are clearly met while it’s unclear if cyclists’ needs are met.

In other words, it’s an art rather than a science. Call it vibe-tagging if you will. This is a bitter pill to swallow for anyone expecting to apply the on-the-ground rule very literally to every decision we make. Fortunately, that rule doesn’t say we need to ensure an alien can read the wiki and reliably make the same tagging decisions anywhere on Earth mere minutes after landing here.

3 Likes
12

Just to clarify, the controversial part in the access documentation concerns using legal access tags to indicate physical restrictions (e.g., access=no). Technically, if a way is not illegal and is known to be used by other means, I see no issue with using bicycle/motorcycle=yes for that purpose.

Great! Is the reason you avoid using highway=path because it feels unnatural for something constructed? Or does it need to be more signed and accommodating, like segregated foot/cycleways?

1 Like
13

Interesting, so then there are a number of keys that exclusively refer to a legality when set to no but also account for practicality when set to yes? This might be lost on people who learned that yes and no are opposites without such nuance.

But no matter. What do you think about the current definition of yes?

The public has an official, legally-enshrined right of access.

For better or worse, many mappers consider this to be a fairly strong statement, if not quite as strong as designated. When something is more of a gray area, neither explicitly prohibited nor explicitly permitted, would you instead set it to yes on the basis that it’s physically passable, irrespective of any tangible reference to a law?

14

Technically, if a way is not illegal and is known to be used by other means, I see no issue with using bicycle/motorcycle=yes for that purpose.

well, up to a certain degree. Adding motorcycle=yes to a highway=footway or highway=cycleway would raise the question what makes it a footway or cycleway nonetheless. Similarly with highway=path, the wiki suggests that it is a typology of way where motorized traffic is not expected, and overruling it with access tags, while technically possible, doesn’t feel like an appropriate solution if these are not some weird and rare exceptions but a typical road class in the country.

1 Like
15

highway=path is just a plain bad tag so I don’t really use it, other than for desire paths of indeterminate access.

1 Like
16

Interesting, so then there are a number of keys that exclusively refer to a legality when set to no but also account for practicality when set to yes? This might be lost on people who learned that yes and no are opposites without such nuance.

well, interestingly we also have a similar situation with motorcar, where motorcar=no means double tracked vehicles like cars, hgv, goods, etc. while motorcar=yes means just cars. This is based on common signage and its meaning (in countries who are part of the Vienna convention on road traffic)

2 Likes
17

Regarding that aspect, I believe the wiki currently reflects only the situation in developed nations. In developing countries where motorcycles are widely used, there are often no default legal restrictions on small motorized traffic on paths, including forest trails. See the default access restrictions for Thailand.

I agree, it can be confusing. I’ve proposed previously using impractical for “legal AND impractical”. This might not be a perfect solution, but I’m open to better ideas!

18

In developing countries where motorcycles are widely used, there are often no default legal restrictions on small motorized traffic on paths, including forest trails. See the default access restrictions for Thailand.

Thailand is the outlier though, it likely doesn’t stand for all “developing countries” because all the other documented situations exclude motorcycles on paths

19

Yes, on that wiki page, because most of the documented countries are developed or transitioning.

But fine, let’s say highway=path shouldn’t be used for motorcycle traffic—what’s the alternative for a way that doesn’t resemble a road, looks like a footway, is public, and is primarily used by motorcycles because the hot climate makes walking (or cycling) impractical?

20

Given the limitations (do not use path and do not invent a new type of highway=*), I would probably opt for a combination of highway=service (for urban and suburban paths), with appropriate access tags and highway=track (for rural ones, connecting farmlands and small hamlets/isolated farms).

While service is far from perfect, at least it describes the lowest road class still available to motorized vehicles (well, motorcycles), it renders rather intuitively as a capillary road, and is generally used by (motorcycle) routers (as a last-resort choice). And I’d stick motorcar=no to each and every one where it’s not physically possible to drive a car, regardless of the “access is only for legal permissions” dogma.

2 Likes