Should narrow roads that are too small for cars still be tagged as minor roads?

julcnx · 2024-08-14T01:43:09.868Z

ftrebien:

So it should come down to figuring out whether a particular street is primarily for pedestrians, primarily for motorcycles, or primarily for both with equal weight.

In Thailand, narrow pathways are now primarily used by motorcycles, as walking has become less common. Unfortunately, there’s no specific designation like highway=motorcycleway for these paths, unlike the available types for cycleways, bridleways, and footways.

ftrebien:

In July 2023, @dieterdreist replaced the original text “Too narrow” with “Illegal for motor vehicles” in the top right cell. The original text tried to contrast this case with “Narrow” in the top left and “Wide enough but illegal” in the bottom right. Would anyone oppose reverting this change? Or perhaps improving the description?

I agree that we should revert the conflicting change and improve it with something like the following:

Screen Shot 2567-08-14 at 08.38.081540×1080 107 KB

Notes:

Distinguish between 4-wheel vehicles rather than including all motor vehicles (which also cover narrower 2 and 3-wheel vehicles).
We could use a different photo for the section that indicates it’s illegal or combine it with the “Too Narrow” section, but I wanted to highlight the signage related to the illegal aspect.

adreamy · 2024-08-14T06:38:19.068Z

As a non-native English speaker, it’s hard to follow the discussion.
Therefore, I will try to briefly write down the important points of discussion.

I don’t think it matters whether a four-wheeled vehicle can get around.
In some transportation cultures, small vehicles or motorcycles are the primary means of transportation.
I think there should be one classification.
In OSM, we classify roads (highway key) primarily by their use.
Therefore, I think it is better to categorize by use and treat other classification factors as sub-attributes.
In particular, ‘alley’, or ‘골목’ in Korean, is typically a very narrow street, but you don’t call it an ‘골목’ just because it’s narrow. Changing it now would cause even more confusion, but at the very least it’s a very confusing phrase, and I think it’s best to avoid using it.
I think the criterion for ‘highway=path’ should be whether it can be legally traveled by a motorized vehicle, not whether it can be traveled by a four-wheeled vehicle (i.e., width should not be the criterion, but only whether it can be traveled by people - or horses or bicycles).
Also, when it comes to car traffic, even the same 4-wheeled vehicle can be restricted by the width of the road, etc. I think this should be expressed as a sub-property.

osmuser63783 · 2024-08-14T08:33:32.866Z

In the UK, as far as I’m aware, the category of “street/alleyway too narrow for cars but legal for motorcycles” doesn’t really exist, so we don’t have that problem. (But I don’t own a motorcycle - most people don’t - so I wouldn’t know for sure.) The example in the picture would just be a footway I think.

julcnx:

We could use a different photo for the section that indicates it’s illegal

Yes, @dieterdreist posted that picture as an example of something that would be tagged service=alley in Italy, with a description of “Quite narrow but legally accessible to motor vehicles”. Therefore it’s probably not a good idea to put it in the Wiki as an example of highway=footway.

@dieterdreist, I am guessing the picture is this way which was previously tagged as highway=footway and you changed it to service=alley, presumably because it’s legal for motorcycles?

Is this case common in Italy? I mean, is this example representative of service=alley? Or are most cases of service=alley e.g. here in Rome wide enough for cars?

@adreamy : does this type of street exist in Korea? How are they tagged?

julcnx · 2024-08-14T11:47:03.658Z

osmuser63783:

In the UK, as far as I’m aware, the category of “street/alleyway too narrow for cars but legal for motorcycles” doesn’t really exist, as far as I know, so we don’t have that problem.

In Thailand, like in many other developing countries, there are countless narrow paths primarily used by motorcycles in urban, rural, and remote areas. These paths rarely have access restrictions.

They may serve as shortcuts in cities, provide access to homes, or even connect villages. In urban areas, they can resemble concrete footways, while in rural or remote areas, they look more like hiking trails.

We tag these paths as highway=path + motorcycle=yes, following past recommendations like this one, which is also used in neighboring countries.

If we accept that an alley only accessible by motorcycle can be tagged as service=alley, then by the same logic, every highway=path in thailand would need to be converted to highway=track, highway=residential, or highway=unclassified, depending on its usage.

This would cause renderers to display these narrow motorcycle paths as roads, and without manually adding motorcar=no, cars could be routed through them. For these reasons, I’m against making this change to the global wiki.

Duja · 2024-08-14T12:31:40.323Z

The main issue people (myself included) have with alley is that it requires highway=service, and those narrow old streets are not service roads, quote:

The highway=service tag denotes ways used for vehicle access to a building, service station, industrial estate, business park, beach, campsite, etc. It is also used for access to parking, driveways, and alleys.

I suspect the highway=service; service=alley combination comes from ambiguity of word alley, which means different things in different countries. Thus, the current service=alley Wiki page is tagged as “controversial” because of the “old medieval streets” section (I plan to edit it so as to clearly distinguish controversial and accepted usage).

However, that’s a bit of tangential issue. For the time being, I cautiously favor Fernando’s highway=pedestrian + motor_vehicle=yes + maxwidth=1.7 (although I’d prefer footway over pedestrian).

SomeoneElse · 2024-08-14T12:32:33.220Z

osmuser63783:

In the UK, as far as I’m aware, the category of “street/alleyway too narrow for cars but legal for motorcycles” doesn’t really exist, so we don’t have that problem.

As ever, there are always edge cases. The note tags on some of those have very convoluted rules. See also here. Some of those may have motor_vehicle=no via a TRO, some may not, and several are sites of ongoing conflict between different types of users.

ftrebien · 2024-08-14T14:19:05.731Z

osmuser63783:

Or are most cases of service=alley e.g. here in Rome wide enough for cars?

Most are wide enough for cars. Expand the Rome table in this study of the existing data I did a while ago.

ftrebien · 2024-08-14T14:38:43.755Z

julcnx:

Unfortunately, there’s no specific designation like highway=motorcycleway for these paths, unlike the available types for cycleways, bridleways, and footways.

At a global level, we probably need highway=motorcycleway (motorised ways primarily intended for motorcycle traffic, but not cars) and highway=alley (narrow non-service public ways not intended for through traffic; different from service=alley). Anyone can start using them at any time, but app support would probably lag for many years (eg. what has been happening with highway=busway), in the meantime some mappers would undo those changes in order to get basic rendering and routing support, with some countries suffering more than others.

ftrebien · 2024-08-14T15:02:11.443Z

julcnx:

Screen Shot 2567-08-14 at 08.38.081540×1080 107 KB

I agree we should mention 4-wheel vehicles (or perhaps “cars” for brevity as the smallest representative of the category). To keep it simple avoiding repetition, I would only mention that in the column headers of the table or maybe a table caption.

I wouldn’t add the new example from Italy yet, it probably needs more discussion. Not only it is a borderline case, there are many similarly narrow old streets in Italy and other European countries mapped as highway=pedestrian.

dieterdreist · 2024-08-14T15:55:11.000Z

Based on my understanding and previous research, roads that are too narrow for standard 4-wheel vehicles should be tagged as pathways (e.g., path, footway, cycleway), which aligns with the current wiki recommendations.

like footway with motor_vehicle=yes? I don’t think this fits particularly well.

On the Italian telegram channel a few people were in favor of highway=residential with a maxwidth restriction, which somehow fits better, but makes it harder for renderers and routers to understand there should probably be specific handling, a distinct highway class is more straightforward (but harder to get started).

In retrospect I think it was maybe too lazy to subtag this kind of alley as subtype of “service”, maybe we should introduce highway=alley for streets (i.e. default to access=yes for all kind of traffic) that are extremely narrow.

What do you think?

dieterdreist · 2024-08-14T16:07:30.000Z

In July 2023, @dieterdreist replaced the original text “Too narrow” with “Illegal for motor vehicles” in the top right cell. The original text tried to contrast this case with “Narrow” in the top left and “Wide enough but illegal” in the bottom right. Would anyone oppose reverting this change? Or perhaps improving the description?

you should have given more context, it is about the reasons why a way would not be an alley but a footway. The case is where it is illegal for motor vehicles. An alley cannot be too narrow to be an alley, but you can be too wide to get through.
https://wiki.openstreetmap.org/w/index.php?title=Tag%3Aservice%3Dalley&type=revision&diff=2573282&oldid=2409669

dieterdreist · 2024-08-14T16:21:55.000Z

@dieterdreist, I am guessing the picture is this way which was previously tagged as highway=footway and you changed it to service=alley, presumably because it’s legal for motorcycles?

yes that’s correct, I also checked with the Italian community on Telegram before I changed it, likely more footways there are actually alleys but I didn’t check it, also the names were mostly missing so I guess these are aerial tracing

Is this case common in Italy? I mean, is this example representative of service=alley?

Or are most cases of service=alley e.g. here in Rome wide enough for cars?

I would not be astonished if there were alleys in Rome which are not wide enough for cars, but I am not sure (the historic center is restricted traffic area anyway), generally these are much more to be expected in rural areas with medieval centers, the less tourists they get, the laxer the regulations and fewer the signs, in remote villages you can drive wherever physically possible with your vehicle, but many alleys may not be wide enough to fit your car

dieterdreist · 2024-08-14T16:30:35.000Z

Most are wide enough for cars. Expand the Rome table in this study of the existing data I did a while ago.

Frankly this study is not useful from my point of view, because this is not a phenomenon to be expected in metropolitan areas such as

Greater London Espandi

These have mostly lost their medieval core, so they are not representative for narrow medieval alleys. Maybe Edinburgh or Rome are exceptions, NYC never has medieval settlements, Berlin destroyed the rest after the last war, … Also these big cities already were big cities with wider streets when in small remote places narrow alleys seemed sufficient or advantageous.

dieterdreist · 2024-08-14T16:32:18.000Z

Not only it is a borderline case, there are many similarly narrow old streets in Italy and other European countries mapped as highway=pedestrian.

if this was a pedestrian area, I would map it as footway, but when no access restrictions apply, footway or pedestrian would not fit.

IanH · 2024-08-14T19:19:31.806Z

In the case of roads that have hard space restrictions, I would use a smaller type of way and then add back vehicle access.

ftrebien · 2024-08-15T03:28:36.659Z

dieterdreist:

The case is where it is illegal for motor vehicles. An alley cannot be too narrow to be an alley, but you can be too wide to get through.

To me, using service=alley for old medieval streets, as in the examples given so far, has always been related to the difficulty of passing due to the narrow width of the way. Where traffic is illegal it clearly cannot be service=alley or any type of highway=service, it must be a non-motorized type (pedestrian, footway or path), reflecting the intended primary use of the way. If it were about legality, then I cannot understand any of the uses of service=alley in Rome or any of the other cities I listed in my study, they would be mapped incorrectly from this perspective, as they are all allowed for wide motor vehicles like cars and many are narrow, but none are that narrow.

SimonPoole · 2024-08-15T06:21:17.105Z

Just as a further data point: the microlino is 1473 mm wide (and BTW it is built in Italy) and that is likely including the mirrors.

There are lots of communal service vehicles that would fit in the ~1.5 m category too.

julcnx · 2024-08-15T07:40:33.510Z

They would fall under the microcars category, as vehicles like ATVs, golf carts, and rickshaws each have their own distinct classifications.

I’ve previously was told that a trail wide enough for an ATV, but not for a 4WD, should be tagged as highway=path rather than highway=track. So, I’d like to clarify: what defines a minor road based on width? What makes sense to classify as a road versus a path on a map?

I’ve argued that using the passability of certain motor vehicles as a guideline isn’t effective, particularly because in many developing regions, pathways are often only accessible by motorcycles.

Would it make more sense to use a threshold of 1.5m instead of 2.0m to make this distinction?

SimonPoole · 2024-08-15T10:23:37.210Z

julcnx:

They would fall under the microcars category, as vehicles like ATVs, golf carts, and rickshaws each have their own distinct classifications.

There’s a new variant of the microlino with 45 km/h max speed that falls in that category but not the original.

FWIW its likely in the L7e category for which I’ve never seen any specific access restrictions.

That said it doesn’t make the alleys in question less navigable by the vehicles (which as the category implies have 4 wheels).

Richard · 2024-08-15T11:55:36.853Z

dieterdreist:

In retrospect I think it was maybe too lazy to subtag this kind of alley as subtype of “service”, maybe we should introduce highway=alley for streets (i.e. default to access=yes for all kind of traffic) that are extremely narrow.

The various different types of service road are basically sui generis from a routing and rendering point of view. A driveway is not a parking aisle is not an alley is not an emergency access. (Also, a service road in the US is not a service road in the UK is not a service road in Germany. Writing routers is fun.)

highway=alley would have been, and would still be, a good idea. It shouldn’t be a massive problem if that’s actually represented in the database as highway=service, service=alley, because any halfway competent router/renderer will already be parsing the service= values anyway.

We do, of course, know that many renderers are not halfway competent - hence the whole highway=scramble debate. So the challenge here is to find something that’s “less wrong”.

highway=footway is very wrong. It doesn’t have the typical physical characteristics of a footway and you’d have to reverse the access implications for motor vehicles, bikes, horses etc. etc. Experience suggests that people are usually a bit slapdash with this and tend to forget one or other class of access. So nope. Same for =cycleway and so on.

highway=path is a clusterf—k beyond all recognition and anyone still proposing using it in 2024 for anything other than informal paths needs their head examined.

highway=service with service=alley is not wrong but potentially a little ambiguous. An “alley” can be a narrow residential street like this. But it can also be a back-access to shops and houses, not really meant as a general-use thoroughfare but as a shared service route.

highway=residential with a maxwidth or width tag is also not wrong but possibly liable to misinterpretation. Routers will potentially ignore the width restriction (OSRM does enforce a minimum width in its default profile: I don’t know about Graphhopper or Valhalla). But then residential roads typically have the lowest weight for car routing so the effect here is perhaps less than it might be.

My gut feeling is residential+width tags is the least wrong and that parsing width tags is table stakes for a car router. But alley would not be a terrible outcome.