Id: Preset: Cycle & Foot Path should produce highway=path rather than highway=cycleway

[naomap]

The preset highway/cycleway/bicycle_foot.json produces highway=cycleway + foot=designated + bicycle=designated.

According to the case S5 in https://wiki.openstreetmap.org/wiki/Bicycle#Miscellaneous, a "Cycle and Foot path" should instead produce highway=path + foot=designated + bicycle=designated.

This bug causes a lot of burden in the Greater Paris area, where a few mappers have damaged a lot of data.

[sun-geo]

I agree, same tagging (highway=path + foot=designated + bicycle=designated) is a tool recommending for German signs, see:

http://osmtools.de/traffic_signs/?signs=240
highway=path
bicycle=designated
foot=designated
segregated=no
traffic_sign=DE:240

or

http://osmtools.de/traffic_signs/?signs=241
highway=path
bicycle=designated
foot=designated
segregated=yes
traffic_sign=DE:241

[BjornRasmussen]

I personally prefer using highway=cycleway on any path that is designated for both cyclists and pedestrians, since those paths look more like cycleways than what I would consider a "path". Although the two methods are equivalent, I think that using cycleway implies that the way is more than just a path in the woods, but rather one that is more formal, and suitable for cyclists.
So really, it doesn't actually matter which tags are used on this preset, because data consumers will interpret them all the same, so I don't think we should change the preset without a good reason.

[charlesmillet]

I personally prefer using highway=cycleway on any path that is designated for both cyclists and pedestrians, since those paths look more like cycleways than what I would consider a "path". Although the two methods are equivalent, I think that using cycleway implies that the way is more than just a path in the woods, but rather one that is more formal, and suitable for cyclists.
So really, it doesn't actually matter which tags are used on this preset, because data consumers will interpret them all the same, so I don't think we should change the preset without a good reason.

This is a long standing subject concerning the path. I used to prefer to chose highway=footway or highway=cycleway rather than use highway=path but I admitted that highway=path was something very generic, justly adapted to those cases where you can't tell and that some additional information like surface, width, segregation, etc. were widely sufficient to distinguish with simple paths.

[naomap]

There is a good reason for changing this preset : it lets beginners damage data that have been carefully created by others. The request http://overpass-turbo.eu/s/Mao shows some of the path shared by cyclists and pedestrians, that have been changed to cycleways using iD and this new preset, where this is wrong : these are not cycleways, there are merely wide sidewalks authorized to cyclists.

There has been long discussions in the past on the Tagging list, as well as on the french Talk-fr mailing list, on the meaning of highway=path. Even though I agree with you that that highway=path should refer to "a path in the wood" (words have a meaning !), the agreement of the OSM community is that highway=path describes a "generic path", i.e. a path that cannot be defined as reserved to a specific usage.

Also data consumers may want to make a distinction. For instance in France so called "voies vertes", which are shared by cyclists and pedestrians, are precisely mapped with highway=path + foot=designated + bicycle=designated. The result of this preset is that they may get changed from "voies vertes" to "piste cyclables" (cycleways). However we need to distinguish them on a map, and in statistics.

So please consider changing the preset so that it conforms to the documentation.

Regards,
Antoine.

[quincylvania]

So I originally added the Cycle & Foot Path (greenway) preset for use in countries like the US where cycleways generally allow pedestrian access, in contrast to Europe where they generally don't. In the meantime we added a property to presets that lets us limit them to specific countries, so to address this issue I just restricted this preset to North America. Europe can continue to tag greenways as they see fit.

Personally I find it unsustainable for a global map to use different tagging conventions in different places, but if that's the nature of OSM then iD will try to reflect that.

[systemed]

highway=cycleway is standard tagging in Britain for this. Could you reinstate this for the UK bounding box?

[quincylvania]

@systemed Yes! We can support any number of countries where this preset is appropriate.

[systemed]

\o/ Thank you!

[simonpoole]

Arghhh.

For starters @naomap any data consumer that doesn't treat the small number of possible cycleway/footway variants either with path or cycleway/footway as the base object as equivalent is IMHO broken. The wiki page in question has been subject to lots of fiddling over the years (because it is a matter of taste and people tend to get upset about such things). [Edit: the page you are referring to is actually not the one I thought of, but leaving out the alternative taggings for cycleways is still wrong, as there is no consensus that only one of the variants is correct, as this issue shows].

Then because it is a matter of taste, and that tends to vary a lot between countries, having a complaint from one country (France) lead to a global change in which it seems to be every country now has to complain to get the what fits them best reinstated (after they have noticed potentially months from now) is just silly.

@quincylvania is it possible to invert the preset per country selection, aka get one specific for FR and whoever else wants it, and leave the other countries as is?

[naomap]

@simonpoole of course a data consumer needs to consider every path, cycleway and footway, the point is not to leave any but to make a distinction between them : they are not the same thing. Or maybe I misunderstood your point.

I do agree that having different ways to map similar things in different countries is undesirable. However when this occurs we don' t want to loose the information, which is what happened here.

Mapping cycleways is getting IMHO too complex, with too many variations between countries. The SOTM talk "Is the OSM data model creaking?" mentions cyclists needs for detailed mapping, I'm looking forward to attending it.

[simonpoole]

@naomap it is long established convention (before I even had mapped a thing) that

• hghway=cycleway + foot=designated
• highway=footway + bicycle=designated
• highway=path + foot=designated + bicycle=designated + surface=*

are functionally equivalent and at best only differ in nuances (in Germany you might find "official" instead of "designated").

Not that the nuances are not important, in urban areas you may find lots of facilities that have the characteristics of a footway that are shared with bicycles, while in rural areas it may be just the other way around.

PS: there is not one solution to the mapping of detailed facilities, particularly as we want to enable the progression from less detailed to more. But as has been shown we can create unified routing graphs with pre-processing and other measures without burdening our contributors with ever more complex modelling.

[quincylvania]

@quincylvania is it possible to invert the preset per country selection, aka get one specific for FR and whoever else wants it, and leave the other countries as is?

@simonpoole Good idea. This is better than whitelisting hundreds of country codes.

I changed the greenway preset to only be blacklisted in France. Mappers should let us know what other places to exclude it, if any.

[mvl22]

I used to prefer to chose highway=footway or highway=cycleway rather than use highway=path but I admitted that highway=path was something very generic, justly adapted to those cases where you can't tell and that some additional information like surface, width, segregation, etc. were widely sufficient to distinguish with simple paths.

I personally prefer using highway=cycleway on any path that is designated for both cyclists and pedestrians, since those paths look more like cycleways than what I would consider a "path". Although the two methods are equivalent, I think that using cycleway implies that the way is more than just a path in the woods, but rather one that is more formal, and suitable for cyclists.

Co-author of CycleStreets here. Yes, we agree on these two.

We treat highway=cycleway as primarily designed for cyclists, which users on foot may sometimes be permitted to use (when foot=yes is added), highway=footway as a footway that cyclists are sometimes permitted to use (when bicycle=yes is added), with the latter considered slower because cyclists using a path designed for pedestrians should be more cautious in cycling. In the UK, typical rubbish shared-use we would expect to see the latter. Even though both end up implying bicycle=yes,foot=yes - we treat "cycleway" vs "footway" as actually having semantically-different meanings.

highway=path we regard as an ill-defined tag and therefore inherently poorer, compared to cycleway/footway which have a much clearer intention. Adding things like surface and width being good really indicates that it it is not a generic path and so should be tagged fundamentally as cycleway/footway in the first place, especially in cities.

Our notes on highway=path state: "Horridly complex tag, everyone uses it differently. [...] This includes snowmobile trails, ski trails, hiking trails, horse trails, bike trails and paths, mountain bike trails as well as combinations of the above and other modes of transportation. These routes may have any type of surface."

So really, it doesn't actually matter which tags are used on this preset, because data consumers will interpret them all the same, so I don't think we should change the preset without a good reason.

No, we think there is a different semantic meaning to these three highway types.

[mvl22]

Mapping cycleways is getting IMHO too complex

+1 ...

There is a lot of subtlety needed in interpreting the data. But this is all fundamentally because cycle provision is incredibly varied (and often poor) around the world, so what one place thinks is good may not be by, say, Dutch standards. Anyone attempting cycle routing without really understanding cycle infrastructure and country-specific real-world norms is probably going to get it wrong.

The SOTM talk "Is the OSM data model creaking?" mentions cyclists needs for detailed mapping, I'm looking forward to attending it.

Hope you managed to come to our talk!

The video of it is now at:
https://media.ccc.de/v/sotm2019-1038-is-the-osm-data-model-creaking-

and the slides at:
https://www.cyclestreets.net/blog/2019/09/22/sotm2019/

[henry4442]

Please add "de" to the exclusion list: "notCountryCodes" in data/presets/presets/highway/cycleway/bicycle_foot.json
It causes retagging of correct data by new users.
highway=path with *=designated is the existing data in germany:
https://wiki.openstreetmap.org/wiki/DE:Bicycle/Radverkehrsanlagen_kartieren