Proposal talk:Cycle expressways

A USA perspective

If we see a need for "this" (tag for semantic) in a number of countries/jurisdictions/cultures/languages/departments of transportation... and it seems we do, then let OSM make it a "thing." There is a similar expressway=yes which allows me/us in a part of California to untangle a decade (or two, almost) of tagging a certain highway I've watched evolve over many decades. I also map (in OSM) bicycle routes (and infrastructure). I see this tag making sense, while listening (widely). I'm in the USA, where these (cycle expressways, super-cycleways) don't seem as prevelant, though I see the trend towards them and states or local jurisdictions beginning to brand them with some evolving-standard that is "around" this concept of "super-cycleway." Some edges to feather, I might assume, but "largely there" with this. Stevea (talk) 23:22, 3 August 2022 (UTC)

Belgium and the Netherlands map them as routes only. This might be enough for your usecase perhaps Joost schouppe (talk) 09:55, 5 August 2022 (UTC)

Unfortunately not, since it is becoming clear that the NL/BE cycle network tagging is not strongly correlated with physical road quality, and that the correlation is weakening with time. This proposal is concerned with physical road quality. The fact that NL/BE are now having issues with a weak correlation between cycle route network and road quality IMO underlines even more the need to have both. Please see the discussion below for clarification. --Balchen (talk) 10:21, 5 August 2022 (UTC)

I left a reply in the thread on the he Dutch forum. This one was not addressed to the propasal as such, but to stevea. Joost schouppe (talk) 10:34, 5 August 2022 (UTC)

Hello Joost; unfortunately I do not read Dutch or its forum, so I kindly ask you to repeat your reply (or link it) here. A link below to a Dutch forum doesn't seem to have it and I am eager to read it! Stevea (talk) 17:58, 7 August 2022 (UTC)

Interference with existing tags

First of all, I'd like to say I think this is a good idea and should be captured in OSM. However, one problem I see is that the key could interfere with existing uses for cycleway=*, most notably cycleway=crossing for mapping the crossing way when it crosses a road (even if this should happen more rarely than with average cycleways). Therefore, it may make sense to use a value of the established designation=*, invent a new tag cycle_expressway=yes, or add some new key to a bicycle route relation to designate the route of the cycle expressway as such. --Popball (talk) 00:18, 4 August 2022 (UTC)

Yes, similar to how an "expressway" (for automobiles, not bicycles) has no (or severely limits / minimizes) "crossings" (such as business entrances, cross-traffic as other streets / driveways entering and exiting the way), a "bicycle expressway" should also have similar standards. We might need to enter specifics (for each country?) where the meaning of the proposed tag allows or does not allow (or allows only under strict circumstances which are well-defined) where such "crossings" are / are not allowed on a cycle expressway. I disagree with needing to change (like adding a new key) to a route=bicycle relation: this seems completely do-able by tagging the way (or node, as in some crossings) rather than a route relation. Stevea (talk) 04:13, 4 August 2022 (UTC)

There is an alternative already in use that prevents clashing with existing refining tags in the form of cycle_highway=yes, which is used on route relations (not on the ways themselves). This in itself is already a way to mark cycle highways. On the Dutch forum the consensus as of 2021 seems to be that cycle highways often lack physical characteristics that would enable a tag on the physical ways to make sense. Mostly a cycle highway is a route that consists of cycleways that may be slightly wider or smoother, but mostly enable fast commuting due to a lack of slower same-grade crossings and other traffic features that would impede speedy progress. I would recommend looking at this (it is used massively in Belgium and as far as I can see for all cycle highways in the Netherlands too) and see if a physical counterpart is needed at all. I would recommend reaching out to the Belgian and Dutch communities to ask why they didn't go for a hierarchical tag on the ways themselves. --JeroenHoek (talk) 07:35, 4 August 2022 (UTC)

• Overpass Turbo link, look at how Flanders is well-covered by a network of fietssnelwegen.
• Discussion on Dutch forum in 2021. --JeroenHoek (talk) 07:55, 4 August 2022 (UTC)

That is very interesting. I've read the forum discussion and had a closer look at the definition of Fietssnelweg. My understanding of Fietssnelweg from Norwegian sources discussing our sykkelekspressveg is that they are the Dutch and Belgian equivalents to sykkelekspressveg. This now seems to perhaps have been slightly inaccurate? A sykkelekspressveg is clearly physically different from a regular cycleway. The highest standard non-express cycleways in Norway are 3 m wide. A sykkelekspressveg must be at least 4 m wide. A Fietssnelweg itself seems to not have any minimum width requirement? If there are no observable differences between a Fietssnelweg and a regular cycleway if you look at a small segment by itself, the concept is perhaps not a cycle expressway. I also immediately think of the difference between a motorway and a high-standard dual carriageway. In many cases, if you observed a random segment of it, you wouldn't be able to the tell the difference. The difference would only be visible at the junctions, where a motorway will have an interchange and a dual carriageway might have a roundabout or an intersection, and in curves, where a motorway with its higher design speed would have a larger curve radius and longer line of sight. All these are physical traits that would be hard to observe from a short, random segment. I guess in that context two more questions arise: 1) are there no observable differences between Fietssnelwegs and regular cycleways because regular Dutch and Belgian cycleways are already so well built that the only improvements left are at the junctions -- which you can't see if you observe a random segment of it -- and so they actually are cycle expressways, but the only way to tell is observing the junctions -- and this would perhaps be a uniquely Dutch and Belgian "issue" since in other countries, the differences would be clearly observable at any random segment of the road and 2) are Fietssnelwegs actually the equivalent to London's cycle superhighways -- i.e. no physical traits, but rather an (inaptly named, perhaps) attempt at creating a road network, and not a higher quality road class? I would love to have that discussion with the Dutch and Belgian communities. How and where? But just to clarify, the intention behind cycleway=expressway is to define a higher quality road class, with clear physical traits that in most cases would be observable on any random segment, but in comparison to the next lower road class would perhaps not be visible unless you look at a larger segment. --Balchen (talk) 08:08, 4 August 2022 (UTC)

Yeah, I think that sums it up. Newer cycleways do get built to higher standards than older ones, and you can be sure that those that are part of cycle highways lack sharp curves and are generally speaking much more convenient. I think that the tag you propose (although perhaps not using cycleway=* given the clashing with other tags noted above) could also be used in the low countries, but probably only where they are physically up to a certain standard, like motorways. I'll let the cycling experts from the Dutch and Belgian communities weigh in on that. Conversely, for Norway the route relations used with cycle_highway=yes can be useful too in addition to tags on the highway=cycleway itself for numbered/named cycle highway routes. These two approaches don't seem to be mutually exclusive, so it's probably best if they do align in documentation. I've invited the Dutch language community (Dutch and Flemish) via the forum thread linked above. You are welcome to join in there in English (or Dutch, obviously, if you happen to speak it, but don't bother with translating software, just use English). --JeroenHoek (talk) 08:47, 4 August 2022 (UTC)

I've posted two messages to the forum, and we'll see what the feedback is. Regarding clashing with other uses of cycleway=*, see my response below. --Balchen (talk) 08:55, 4 August 2022 (UTC)

Thank you for bringing that up, Popball. I agree that cycleway=expressway would interfere with tagging a segment of that cycleway as cycleway=crossing. It's a valid issue to raise. When I think about it, though, we don't have this issue for carriageways, neither when they cross eachother nor when they cross a cycleway or a railway. The reason we don't have this issue -- IMO -- is that carriageways are viewed as being continuous through a junction. A cycleway often is not. That's why we don't call them junctions -- we call them crossings. If a cycleway were continuous at a junction, it could simply be drawn straight through, and there would be no need to tag a segment of it with cycleway=crossing. One of the important traits of a cycleway=expressway is no or very few at-grade junctions with other road classes, and right of way where such junctions exist. Right of way certainly means it is continuous. A junction with right of way for the cycleway=expressway therefore means it can and should be drawn straight through, with no cycleway=crossing. Any relevant information about the junction can be recorded on the junction node, like we do for highway and railway. It could easily be argued that if there are at-grade junctions where the cycleway=expressway is not continuous for some reason, it has also ceased to be a cycleway=expressway at that junction, very much like a motorway with a roundabout instead of a grade-separated interchange ceases to be a motorway at that point. In this scenario, cycleway=crossing could be applied without conflict. In fact, the conflict itself prevents obvious cases of mistagging; the combination cycleway_express=yes + cycleway=crossing makes very little sense. --Balchen (talk) 07:39, 4 August 2022 (UTC)

+1 to Popball. While cycle expressways are tagged with highway=cycleway + cycleway=expressway, how can we distinguish between ordinary cycleways and unsurveyed sections whether they are expressways? If cycle expressways and ordinary cycleways are exclusive (a way cannot be both at the same time), we also need some cycleway=* for ordinary cycleways, but it is very verbose and interferes with other cycleway tags, and it doesn't correspond to real features. Any boolean tag such as cycle_expressway=yes will make sence (no as default for untagged cycleways). --Gyotoku810 (talk) 11:03, 4 August 2022 (UTC)

Good point. That's certainly an argument in favour of creating a separate key for it. cycleway:expressway=yes? --Balchen (talk) 11:17, 4 August 2022 (UTC)

On a related note, wouldn't this issue be same for highway=footway + footway=sidewalk? Perhaps we can use that example to discuss the severity of the issue. How do mappers know if a footway is an unsurveyed sidewalk or a regular (non-sidewalk) footway? --Balchen (talk) 11:23, 4 August 2022 (UTC)

Yeah, I know also the footway issue well, and also highway=service + service=* has the same issue. I think all these have bad semantics in my opinion. A subtag should be a refinement of the parent tag, so features with subtags should be included semantically in features with no subtags, but in these cases, features with and without subtags have different meanings. That's beyond the scope of the cycleway discussion anymore... --Gyotoku810 (talk) 12:01, 4 August 2022 (UTC)

I do agree from a semantic viewpoint. This is also the case with cycleway=crossing, which was the main reason a concern was raised in the first place. A cycle crossing is a different real-world object (or primarily a lack of a real-world object). Strict semantics aside, if we accept that cycleway=crossing is a valid refinement of highway=cycleway, then cycleway=expressway should be an equally valid refinement -- a cycle expressway is a proper subclass of a cycleway. The real semantic problem is perhaps not cycleway=expressway, but the lack of cycleway=regular? Off the top of my head, I also cannot think of a precise term to explicitly describe a regular (non-motorway) highway either -- the term highway just implies that it is not a motorway unless explicitly specified. The underlying semantic issue seems to be universal. So much so that in zoological classification, the "regular" class of an animal is denoted by simply repeating the name again (e.g. Rattus rattus). In that spirit, perhaps cycleway=cycleway could be a way to explicitly denote a regular cycleway :) I don't know if I seriously meant that or not, so please don't let that derail the discussion at hand. --Balchen (talk) 12:45, 4 August 2022 (UTC)

OK. Then, I found most of cycleway=* are not subtags of highway=cycleway, but added to road ways. Exceptionally cycleway=crossing can be added to highway=cycleway, but not only to, and can be added to other path/footway ways. On the other hand, the proposed cycleway=expressway can only be added to highway=cycleway. I think cycleway=* shouldn't be used like as a subtag. This is the another reason to recommend a boolean tag. Next, about the key of a new boolean tag. For a suffixed key, the suffix part should indicate common information. Thinking about cycleway:expressway=*, this tag should indicate common information with expressway=* that is "it is a expressway (for car)". Cycle expressways are obviously different from expressways for car (similar to parking and bicycle_parking). Therefore, it is thought cycleway:expressway=* is not a good key name. After all I like cycle_expressway because it's just simple. --Gyotoku810 (talk) 14:57, 4 August 2022 (UTC)

If there is a an agreement on highway=cycleway + cycle_expressway=yes/no, I will be happy with that. --Balchen (talk) 15:56, 4 August 2022 (UTC)

cycleway:expressway=* already means it is different from expressway=*. --- Kovposch (talk) 03:23, 5 August 2022 (UTC)

cycleway=* should not be used because it may conflict with cycleway=sidepath, and the few cycleway=crossing on such routes.

Variation of *expressway=* sounds too similar to cycle_highway=*. Aside from mention of "supercykelväg" and "cykelsupersti" (guessing their meaning by looking at them),  Bike freeway starts with "also known as a cycling superhighway, fast cycle route or bicycle highway", while referring to the longer distance or commuting use, and including London's case that's excluded here. Until that is resolved, it may be preferable to use something else.

Adding eg designation=* should be done, with international differences in mind. That being said, the national status is not mutually exclusive with a general tag. I personally don't like expressway=* when it is unsigned, because its meaning can be vastly distinct.

"Expressway" seems to be a status, or specific definition.

Some listed criteria are more subjective (a range of acceptance), or less easily determined (require technical judgement). Not sure about requiring even some of them together either. Eg when the route has only a few few at-grade crossings, priority may not be essential overall-speaking. When a country has different cases, how do you distinguish between them?

From other features, I'm reminded of highspeed=* on railway=rail. Using it prefixed cycleway:highspeed=* could be more direct than needing to interpret the meaning of "expressway", and allow more detailed description than using cycleway:expressway=* to lump every characteristics together. As a result, cycleway:expressway=* will only be used when it is designated as such.

--- Kovposch (talk) 04:35, 5 August 2022 (UTC)

A cycle expressway should never be cycleway=sidepath nor cycleway=crossing. If you want to map a cycleway in this manner, you are probably mapping a designated "highway" route, not an expressway. Consider as an example if someone were arguing that we can't have a tag highway=motorway because what if they need to use highway=residential on part of the same way? In this case, the problem is not with the tagging schema, but with your road planners or road builders.

It's true that the terminology is often mixed (see discussion below on the difference between physical road class and road network function), even on such reputable sources as Wikipedia, and may vary from country to country. This is universal; highway=* is used in OSM to designate any road or path, but in the US, "highway" often means "motorway". Yet we manage to tag highway=motorway in OSM with no apparent confusion.

The fact that expressway carries a designation or definition in several countries (with different names) is the basis of my proposal. Actual classification and designation will vary from country to country. Having a range of accepted values is very common for all roads. Norway has motorways without emergency lanes/wide shoulders. In many other countries, this would be strictly outside of the definition of a motorway and would be classified as something else. Still, no one from Germany or Serbia is objecting to us tagging highway=motorway on those roads.

I don't see the principal difference between highspeed=* and cycleway:expressway=*. highspeed=* also lumps many defining characteristics of high-speed rail lines into a single tag. It also lumps wildly varying national classifications into a single tag. A high-speed railway may be steel, dual-track, wide-gauge, surface-built line with a design speed of 200 kph, or it may be a single track, elevated, maglev line with a design speed of 500 kph. We are not restricted from detailing the characteristics of a high-speed railway or a cycle expressway with further tags, such as width, lighting, lanes, intersections, surface, etc. The intention was to use cycleway=expressway when it is designated as such. How and where it is designated will vary from country to country. For the three Nordic countries, this seems to not be an issue at the moment. For NL/BE, it is still unclear if they actually designate Fietssnelweg as a physical road class, or if they designate Fietssnelweg as a route network classification.

--Balchen (talk) 07:28, 5 August 2022 (UTC)

Why would you say it should never be cycleway=sidepath when the example photo File:Fietssnelweg_De_Liemers_Westervoortse_brug_(3).jpg 60852612 60852612 looks like one? The proposal itself includes "very few at-grade crossings ". That's not cycleway=crossing? Kovposch (talk) 12:13, 5 August 2022 (UTC)

if you want to tag the adjacent road with cycleway=sidepath to indicate that this road has as adjacent cycleway, please do. With regards to crossing, I quote my previous comment on this: " A junction with right of way for the cycleway=expressway therefore means it can and should be drawn straight through, with no cycleway=crossing. Any relevant information about the junction can be recorded on the junction node, like we do for highway and railway. It could easily be argued that if there are at-grade junctions where the cycleway=expressway is not continuous for some reason, it has also ceased to be a cycleway=expressway at that junction, very much like a motorway with a roundabout instead of a grade-separated interchange ceases to be a motorway at that point. In this scenario, cycleway=crossing could be applied without conflict." --Balchen (talk) 12:25, 5 August 2022 (UTC)

Umm no, cycleway=sidepath is used on the highway=cycleway. That's why I'm asking. You are thinking about cycleway=separate on the roadway,

Why would priority means it is not a cycleway=crossing? Certainly highway=crossing and footway=crossing is still used for any crossing where pedestrian has priority?

--- Kovposch (talk) 12:29, 5 August 2022 (UTC)

See Tag:cycleway=sidepath and Proposed features/Key:is sidepath for the usage I described and for alternatives to what you are describing.

highway=crossing is used on the node, not the way, you can still do that without conflict with cycleway=expressway. Any additional information you want to capture about the junction can be captured on the node. footway=crossing means the footway has ended and the pedestrian is now on the road, even if the pedestrian has priority. The argument is that if a cycle expressway has priority, it will be continuous through the junction and will not have ended, therefore cycleway=crossing isn't necessarily the correct tag. You are free to use it, of course, but that means the cycle expressway will have a little gap at this segment, which accurately reflects the real-world situation if the cycle expressway does not have priority. That is fine.--Balchen (talk) 12:43, 5 August 2022 (UTC)

1. You are only describing one usage, not all. Then that needs to be solved before this. You are not proposing your alternative here. There are 9k cycleway=separate, compared to ~3k cycleway=sidepath combined, plus ~1k cycleway:*=sidepath, This is not counting path=sidepath. is_sidepath=* is used on highway=footway and highway=path altogether.
2. footway=crossing may still be used on "continuous" sidewalk flat crosswalks. If not cycleway=crossing, what tag should be used to indicate section is crossing a road? Why would you have a highway=crossing without a cycleway=crossing there? Asking for everything to be tagged on highway=crossing is not a solution. You already lose the length and alignment (angle/shape) information here.

--- Kovposch (talk) 09:38, 6 August 2022 (UTC)

I am happy if the consensus is to move forward with a tagging scheme like cycleway:expressway=yes. This also means there are no schema conflicts to discuss. What we are discussing now is how to combine a cycle expressway with cycleway=sidepath or cycleway=crossing, and even how to map crossings, and that is evolving into a different discussion than the original discussion about tag conflicts. Also, the benefit of being able to tag cycleway:expressway=no to indicate a successful, but negative, survey, for me makes this a valid and valuable change to the proposal. --Balchen (talk) 20:51, 6 August 2022 (UTC)

highspeed=* is immediately clear it is about "high speed". Expressway doesn't necessarily mean high speed, then other criteria are added on the list.

Anyway, that's only an idea I had. My main concern is cycle_expressway=*'s superficial similarity with cycle_highway=*. I'm half apathetic with cycleway:expressway=*, as I pointed out its usability above.

--- Kovposch (talk) 12:37, 5 August 2022 (UTC)

I think what I see (better) here (and now) is that this is a rich subject that probably shouldn't be captured with a single tag. I continue to think it important that OSM distinguish between "infrastructure" tagging and "fits into a bicycle route network at what our country calls a trunk network" (for example), which has nothing to do with its infrastructure. (Although, as has been mentioned, these do influence one another). There are so many "this but not that" versions of this, and likely with (subtle) distinctions between how these exist and are created / named in different countries that some research might need to be conducted. Indeed, we seem to be doing (part of) that now. This is a rather good start, but I'm not sure the single proposed tag (feature) is rich and deep enough to describe the ("faster bicycle" infrastructure) realms which are in the real world. So, some "syntax improvement" to the proposal seems to be necessary ahead. Stevea (talk) 06:41, 5 August 2022 (UTC)

I get what you're saying. Still, we don't let those issues stop us from using highway=motorway or highway=trunk. For your reference, this is a Norwegian highway=trunk road: https://www.google.no/maps/@60.3428329,6.656998,3a,90y,22.15h,77.3t/data=!3m7!1e1!3m5!1sAOa9PUV3S74Ph6iR5_ZX5A!2e0!6shttps:%2F%2Fstreetviewpixels-pa.googleapis.com%2Fv1%2Fthumbnail%3Fpanoid%3DAOa9PUV3S74Ph6iR5_ZX5A%26cb_client%3Dmaps_sv.tactile.gps%26w%3D203%26h%3D100%26yaw%3D286.13422%26pitch%3D0%26thumbfov%3D100!7i16384!8i8192?hl=no&coh=205409&entry=ttu https://www.openstreetmap.org/way/68901039#map=15/60.3445/6.6597. It seems at the moment that cycle expressway has a stricter and more internationally uniform definition than trunk road. If anything, we should stop using highway=trunk and start using cycleway=expressway. :) We can capture variety by enriching the road with other tags, as we should! But the overall concept of a cycle expressway still deserves recognition, like motorway or trunk. --Balchen (talk) 07:59, 5 August 2022 (UTC)

There are actually more specific tags such as dual_carriageway=*, and access_control=*. I prefer these over a vague expressway=*. For comparison here, since you elaborated on this. --- Kovposch (talk) 12:16, 5 August 2022 (UTC)

@Balchen and Kovposch: There's nothing wrong with dual_carriageway=* and access_control=*, and from this proposal it seems clear that these keys could apply to more than just roadways. These keys and expressway=* all have potential practical uses for renderers and routers. The main point in favor of expressway=* is that American maps and signs visually distinguish expressways from other roads (even if map legends confusingly describe expressways as a form of access control). Because of this distinction, ordinary users expect things like "avoid highways" to mean both freeways and expressways, not necessarily limited-access and/or divided highways, even if their dialect has no word for expressways. If this is a mess in OSM, it's just a reflection of the mess in reality.

highway=* classification has been largely focused on macro functional roles and cycleway=* has been focused on micro physical attributes, leaving a gap for classifications at a middle scale that has been filled so far by a dizzying array of top-level Boolean keys: expressway=*, motorroad=* (effectively), cycle_highway=*, cyclestreet=*, just to name a few. I'm not sure if it's better to strive for a unified classification scheme or continue proliferating these top-level keys, which are dependent on regional definitions even if there's some commonality.

 – Minh Nguyễn ^💬 22:18, 6 August 2022 (UTC)

Object tag or not?

Is this supposed to be an attribute on existing objects or a tag that defines a real world object on its own (creating a new top level key)? The proposal text is ambiguous.

The intended use is highway=cycleway + cycleway=expressway. I'll amend the proposal text to clarify. --Balchen (talk) 09:15, 4 August 2022 (UTC)

The difference between physical road class and road network function

Based on feedback on this proposal and discussions in other forums, I see a need to address the distinction between physical road class and road network function.

The physical road class and the road network function are two independent, but of course correlated, features of a road. In a sparsely populated country like Norway, you'll frequently see highways designated as Riksveg (the highest national road network) or Europaveg (European highway network) that are so narrow that two large vehicles cannot meet. In terms of the road's network function, it's clearly of high importance, but the physical road class is comparable to a residential street.

A road's physical road class concerns how it is built. Road classes are observable, physical, mechanical traits. The effects for road users of a road having a certain physical class are mainly capacity, speed, and safety. Cross section, physical barriers, line of sight, curvature, width, number of lanes, surface, lighting, gradients/slopes, at-grade or grade-separated junctions are physical traits that are defined in a road class standard.

A road's network function concerns how it allows travel between localities, districts, or regions, or to/from destinations (houses, businesses, parking, etc). The effects for road users of a road having a certain network function are mainly signage, continuity in road numbering, length of road continuity, right of way, AND in the public desire to change the road's physical class.

At least in Norway, this principled distinction between physical road class and road network function is easily observable, because the correlation is so low. In countries with a much higher correlation between the two, the distinction might be blurred to the point where they appear to be the same dimension. I would still argue that they are not, and that they would best be captured independently.

I wanted to address the issue since I see that the two concepts are often mixed and people are arguing that network function does not match exactly with road quality -- and I agree! That's why I want a tag to explicitly specify road quality. --Balchen (talk) 15:05, 4 August 2022 (UTC)

@Balchen: I think your parallel proposal Proposed features/cycleway hierarchy makes this distinction pretty clear. Maybe mention it as part of the proposal to alleviate concerns from those who care more about the hierarchy than the design characteristics? – Minh Nguyễn ^💬 21:57, 6 August 2022 (UTC)

More similar name as cycle_highway=yes

Currently cycle highways or cycle expressways are already marked with a route relation with the tag cycle_highway=yes. If I understand correctly this proposal will additionally add the tag cycleway=expressway to the ways of the cycle highway. Wouldn't it then make more sense to use a more similar tag like cycleway=cycle_highway? Otherwise this proposal will just make things more confusing. A67-A67 (talk) 18:51, 4 August 2022 (UTC)

The feedback from NL/BE is that their "cycle highway" network does not always consist of expressways, so only some ways could correctly be tagged as expressways. Cycle network and road quality are two different dimensions. There are of course correlations between the two dimensions, but these correlations vary wildly between countries and regions. I think there is a probability that confusion arises from mixing physical road quality with cycle networks and using one dimension to represent the other -- or both. So to summarise this part: It seems cycle_highway=* is not used to denote a cycle expressway today -- it is used to denote a type of cycle road network that is sometimes implemented with expressways, and sometimes not.

Regarding using the term "expressway" or the term "highway" as the name for a specific physical road quality, "expressway" is understood to be the highest level road quality across all roads. "highway" is commonly understood as a generic road, and this is the meaning in OSM. The tag combination highway=cycleway + cycleway=cycle_highway appears completely redundant and doesn't immediately convey any new information unless you are specifically aware that "highway" means something special only for cycleways, but something plain and generic for all other highways. I would also think that the use of cycle_highway=* has much higher risk of being confused with or seen as an alternative to highway=cycleway. For these reasons, I'm not convinced that "cycle_highway" reduces confusion.

It would be beneficial to clarify how the route relation tagging and the physical quality tagging could and should work together, like Jeroen has also mentioned, so that is something we would look at when/if documenting the tag.--Balchen (talk) 19:56, 4 August 2022 (UTC)

A fietssnelweg (cycle highway) is supposed to be the same as what here is described as cycle expressway. Although in reality there will always be money and space constraints when building infrastructure. So there can be substandard parts in a fietssnelweg. These are for example narrower parts, parts with more crossings than desired or parts that are a cyclestreet for motorized access to houses along the route. A67-A67 (talk) 21:27, 4 August 2022 (UTC)

This can always happen, and especially in those cases it is valuable to capture (in OSM) what was actually built, not what was supposed to be built.

Compare this to the E-road network. I quote Wikipedia: "Originally it was envisaged that the E-road network would be a motorway system". I don't think anyone considers it correct to tag a road as highway=motorway when the built road has a single carriageway, just because it carries an E-number or because a motorway was supposed to be built. Different road segments will have different road classes depending on how they are really built.

If a Fietssnelweg is an expressway, and someone built something that isn't an expressway, then it isn't a Fietssnelweg and it shouldn't be tagged as Fietssnelweg, even if it was built on a route that is supposed to have Fietssnelwegs. It doesn't become a Fietssnelweg simply because that was supposed to happen. Now, I emphasize that I say "if", because at the moment it is a bit unclear if Fietssnelweg is actually a physical road class or if it is the name of a cycle network system (like the E-road network).

In Norway, sykkelekspressveg is a physical road class. It is also signed slightly differently, although I don't know if this is formalised. --Balchen (talk) 21:59, 4 August 2022 (UTC)

Just to remind everyone: we are in a fairly murky space of "how countries/cultures/departments actually classify both bicycle infrastructure AND routes" mapping nicely and neatly to OSM tags. This is difficult, let's not kid ourselves. Keeping separate infrastructure tagging and route tagging (and/or how a 'route fits into a network / hierarchy') is a good start. Keeping all of this straight means keeping track of a lot: laws, signs, public perception, the need to "tie it all together so all of Earth's People agree." I often find building a rough structure of framework, the simpler the better, as a place "many agree to vote on" a great next step. There is a lot to unfold and unpack here. I think keeping it simple, as "THIS" means "what we agree it to mean" is how we get there. That's squishy, and it should fit into a small tube. I think we can do this. Stevea (talk) 00:10, 5 August 2022 (UTC)

I think different countries have similar bicycle infrastructure, that can be called cycle highways, bicycle highways or cycle expressways. These are all cycleways for functional (as in non-recreational) cycling that make higher speeds and less stops possible. In Openstreetmap the name cycle highway is used to describe this, so it would make sense to me that other tags than cycle_highway=yes, which is used on route relations. So cycle_highway=yes on routes and cycleway=cycle_highway on ways makes more sense to me than the inconsistent cycle_highway=yes on routes and cycleway=expressway on ways.

In the Netherlands and Flanders (Dutch-speaking Belgium) this kind of bicycle infrastructure is called fietssnelweg, snelfietsroute or doorfietsroute. There is special signage that is used on the entire finished route, also on substandard parts. In the Netherlands the signage used to be very inconsistent with every province inventing their own. However the signage was formalized in 2021 and is now slowly being applied to all cycle expressways.

Because the only objective way to determine whether something is a cycle expressway is the signage, the substandard parts would also get cycleway=expressway on the ways. Only using the tag on parts that are cycle expressway according to some guidelines in the OSM wiki would result in subjective tagging. Someone would consider a part too narrow, while someone else thinks it's wide enough.

You make a comparison with the E-road network, but this is not a fair comparison in my opinion. When an E-road isn't build as a motorway, it isn't a motorway according to the countries guidelines and signage. It should therefore not be tagged as motorway, even if it was supposed to be. For a cycle expressway that is signed as such, including the narrower and twistier parts and parts where local car traffic is allowed, everything is build as cycle expressway. You could better compare this to a narrow or twisty part in a motorway, that is not up to the guidelines for constructing motorways. That part would still be considered a motorway.

I think you should take differences between countries into account. The Netherlands and Flanders are densely populated areas, so space is limited. This will result in parts that are narrower, twistier or which are a cyclestreet with local car traffic. Besides Dutch cyclists are cycling quite "unregulated" and more negotiating. Things like turn lanes are very uncommon on cycleways, because people decide for themselves how they cycle there. They almost only exist at traffic lights, so people waiting for the traffic light don't block people that don't have to wait. A67-A67 (talk) 08:07, 5 August 2022 (UTC)

This is a Norwegian trunk road: https://www.google.no/maps/@60.3428329,6.656998,3a,90y,22.15h,77.3t/data=!3m7!1e1!3m5!1sAOa9PUV3S74Ph6iR5_ZX5A!2e0!6shttps:%2F%2Fstreetviewpixels-pa.googleapis.com%2Fv1%2Fthumbnail%3Fpanoid%3DAOa9PUV3S74Ph6iR5_ZX5A%26cb_client%3Dmaps_sv.tactile.gps%26w%3D203%26h%3D100%26yaw%3D286.13422%26pitch%3D0%26thumbfov%3D100!7i16384!8i8192?hl=no&coh=205409&entry=ttu. This is a Norwegian motorway: https://www.google.no/maps/@60.3428329,6.656998,3a,90y,22.15h,77.3t/data=!3m7!1e1!3m5!1sAOa9PUV3S74Ph6iR5_ZX5A!2e0!6shttps:%2F%2Fstreetviewpixels-pa.googleapis.com%2Fv1%2Fthumbnail%3Fpanoid%3DAOa9PUV3S74Ph6iR5_ZX5A%26cb_client%3Dmaps_sv.tactile.gps%26w%3D203%26h%3D100%26yaw%3D286.13422%26pitch%3D0%26thumbfov%3D100!7i16384!8i8192?hl=no&coh=205409&entry=ttu. We have our own definitions. Believe me, I strongly do take variations between countries into account.

If the Dutch or Flemish definition of a fietssnelweg is a physical road class with certain requirement for quality, but still allows stretches of narrower, twistier, lower quality road, then I would say this is how the world actually is and you should tag that as cycleway=expressway because that is what cycleway=expressway means in NL/BE. Exactly like how the Norwegian definition of a trunk road allows a road where two large vehicles cannot meet, and the Norwegian definition of a motorway allows no emergency lane/shoulder.

If the Dutch or Flemish definition of a fietssnelweg allows for parts of the fietssnelweg to be a totally different kind of road (like a pedestrian street or a living street), then fietssnelweg simply isn't a physical road class, but a route network class, and you could not and should not tag those pedestrian streets or living streets with cycleway=expressway just because they are part of a fietsnelweg network. Those parts of the fietssnelweg network that in other countries would normally be officially classified as cycleway=expressway would then not have any official classification in NL/BE, despite being part of a fietsnelweg network, and so it would be up to the NL/BE OSM communities to decide to not tag them as cycleway=expressway.

The signage for fietssnelwegs that you refer to, is it this signage? If so, that is network/route signage, not road class signage. Again, I strongly encourage participants in this debate to be mindful of the difference.--Balchen (talk) 08:35, 5 August 2022 (UTC)

The signage you refer to is indeed the Flemish signage of fietssnelwegen. The new Dutch signage is this. There is no signage like start motorway, that marks the road class. A67-A67 (talk) 08:44, 5 August 2022 (UTC)

Then this is in fact cycle route network signage, and any correlation with physical road class is coincidental/"supposed to be". The Norwegian signage for sykkelekspressveg is and the route network signage on the same road is .--Balchen (talk) 08:50, 5 August 2022 (UTC)

This is exactly like how an E-route sign does not mean something is a motorway, btw. The comparison seems very fair.--Balchen (talk) 09:00, 5 August 2022 (UTC)

Except that this signage is only to be used on cycle expressways, not on other cycleways. But my point is that these road and route classes are strongly related, so consistent naming would be desirable. A67-A67 (talk) 09:20, 5 August 2022 (UTC)

E-route signage was only to be used on motorways, and then people started using them on other road classes as well. Exactly like you are describing with fietssnelweg route signs, and like others (or perhaps you) have been saying on the Dutch OSM forum. The quality of the route network designation is devalued because it is being used on roads it wasn't meant to be used on.

Consistent naming of two things that are completely different will probably introduce even more confusion. Again, it's like saying we should be using "consistent" naming for E-routes and motorways just because the intention was that a road would be both at the same time. --Balchen (talk) 09:27, 5 August 2022 (UTC)

Countries will indeed probably chose the cheaper option over an agreement they made in an UN commission. E-roads have existed since the 1950s and they have never fully been build out to be all motorways. So therefore I don't think the comparison is fair. I'd compare it to the A-road network in the Netherlands or France, or the M-road network in the UK. Those are supposed to be only used on motorways. Those motorways can have substandard parts, with narrow lanes, sharp turns and movable bridges, but they are still motorways, even the substandard parts. A67-A67 (talk) 20:54, 5 August 2022 (UTC)

If I were to un-murky here, I continue to believe that this sort of precision language, especially as it discerns between infrastructure tagging and route tagging as "quite distinct" is a winning strategy here. There really are those two different things, they wildly vary all over Earth, and we best be careful. Recognizing the many differences and "smears of imprecision" (like "dilution of what the network was supposed to be with what it is because it has lower-standard infrastructure..." these and more are all true) is what people need to keep in mind as we both tag these and discuss tagging these. This proposal seems to pretty strictly be talking about infrastructure, as it is meant to convey a "higher or highest standard" (let's use that for now) bicycle infrastructure elements. Yes, this is going to mean different things in different countries (in 2022, maybe in the decade to come these harmonize...don't hold your breath). So, we must keep that in mind as we tag. Let's go slow and careful here, discussion and good dialog like this is great, we do well to say what we mean, mean what we say, and for people to really read and understand the wide perspectives here. The specific tagging seems to better emerge.

Again, I am in the literal middle of an experience (Mission Street, Santa Cruz California) of this 2.5 km of highway being "promoted" from highway=primary to highway=trunk, but with the specific tag of expressway=no, deliberately and quite distinct from the expressway=yes segments of highway=trunk it is connected to on both sides. This is happening because of how USA's definition of "trunk" has "grown up" (to mean something specific, AND distinct FROM "expressway" because we use that key, too). There are short-, medium-, and longer-term perspectives on tagging in OSM and I have the longer one on this topic. I also think "when it makes sense to dwell in design, do not skip to implementation." (In this case, the Proposal moving too quickly to a Vote, as it continues to receive important Comments and dialog). So, I'm 100% OK with this continuing as an excellent conversation, including a nice, "wider" (could go even wider) sub-community; good for us. I will add that emergences from here of specific flavors like cycleway=expressway (maybe with an additional well-documented boolean), cycle_highway=yes and cycleway=cycle_highway still have "green lights ahead" (in my opinion as possible candidates), as long as we are crystal clear what we mean by them. And then we should be able to hammer a firm lid down on that and Vote. No rush on Voting, it's good swimming in these (design, not implement) waters now. Stevea (talk) 02:18, 16 August 2022 (UTC)

not a fan of this tag

As can be read from several arguments here/discussions on the forum this tag in the end just describes the physical condition of a cycleway, that could be described by excisting tags. It does not describe a relation ( the actual route ) and does not solve the fact that other parts of a cyclehighway as route still need to be tagged. From a routing point of view it adds no specific value. If you would like to describe the quality of a road with a single tag than you should design a generic taggingscheme that would apply to all road types. --Noordfiets (talk) 12:27, 5 August 2022 (UTC)

After reading the numerous comments here, "design a generic tagging scheme" (for bicycle 'superhighways / expressways') seems to be a way to move these intentions forward. However, such a scheme must not be simply this single tag, but it must accommodate all countries / signage schemes that are (or will be, through extensions to the proposed syntax) in the real world. Based on the comments, we are not there yet, not with this proposal, not with a larger scheme that might include it. Again, I think we can do this, but it will take much more discussion to make CRYSTAL-clear the distinctions that must be made (to both bicycle infrastructure and bicycle routing / networking) so that everybody can understand it. I'm fairly sophisticated in these realms (I am a long-time, serious OSM mapper of bicycle infrastructure and routes in the USA, having "done" thousands of miles of both, I give talks on these topics at national conferences...) and I do not understand the proposal as it currently exists, especially with all of these comments. We'll need much more clarity if we are going to vote upon something (in the future). These are good discussions, but they must continue to better develop, so they refine something such that it is very clear to everybody. We are not there yet. Stevea (talk) 01:04, 6 August 2022 (UTC)

As it stands the tag is specific for one country only, Norway, by it's current definition. But in several forum discussions it is unclear if the tag describes the physical state of a road only or includes the legal state.

If it is just the physical state than the tag is not really needed. If it descibes a road-class ( as a combination of physical and legal state ) than the tag should be reconcidered to be in line with the excisting tagging for highways. E.g. highway=cycle_express would fit the general tagging scheme better and make more sense.--Noordfiets (talk) 07:05, 6 August 2022 (UTC)

I think using the concept to describe both physical and legal state is a good idea and already implicit in the proposal, but it can be made explicit. Certainly whenever a cycle expressway is signed, that would automatically be the case anyway, even if it shares the general cycleway sign. This would cover the case when a cycle expressway has different rules from a regular cycleway, and the case when they have equal rules. The difference in rules would vary from country to country, I assume. This is certainly the case with highway=motorway.

I am not opposed to introducing a new tag highway=cycle_express for the purpose. The general issue with that is it isn't backwards compatible, so all data consumers will be unable to interpret this new highway value until they are updated. Tagging it as a highway=cycleway and then refining it as an expressway allows everyone to automatically interpret it correctly as a cycleway, and then further refinement can be done as needed. But if a new highway value is the correct way to move forward in principle, and the community agrees, I have absolutely no issue with that. --Balchen (talk) 08:12, 6 August 2022 (UTC)

There are many different 'cyclehighways', depending on country ( and even province in NL ). The Norwegian type does not excist in NL. Perhaps there are some in DK. If it is a true cyclehighway the most logical tag is highway=cycle_express. And as I interpret your proposal, a cyclehighway has a special legal state in Norway: it is truely a road-class.

It does not so in NL/BE/etc ( with even different access rules per province in NL ). Sometimes it looks like the real thing, but many times it is just a chain of excisting, partially upgraded, partually new, cycleways and rural roads. That makes it more logical to regard them as routes and combine them in a relation. They are not a special road-class, legally they do not excist but are just a cycleway. That is why they are called 'cycle-on-routes' or something simular. But the name cyclehigway is still often used by the general public. The tag cycleway=expressway would result in a patchwork of tags, not connected, and lose it's meaning other than stating that it is a wide cyleway.--Noordfiets (talk) 09:25, 7 August 2022 (UTC)

For the remainder of the discussion, I think it would be useful if we agree on using the the term way to denote a physical road, and the term route to denote a combination of ways that are signed towards a destination. There can then be cycle express ways that are meant for safe, uninterrupted, high-speed cycling, and cycle express routes that provide a fast/the fastest combination of ways to get to a destination. Conversely, there are ordinary cycleways, that do not provide safe, uninterrupted, high-speed cycling, and there are ordinary or tourist cycle routes, that do not provide a fast/the fastest combination of ways to get to a destination. Note how the inclusion of cycle express ways in a cycle express route greatly helps achieve the goal of providing a fast/the fastest combination of ways to get to a destination, so therefore people will often -- inaccurately -- talk about them as being the same thing that we are building. And one could easily argue that a cycle express route is possible without a cycle express way. There is correlation, but there is no causation.

I guess the trouble I have with your argument is that you are arguing that the cycle express routes in NL/BE do not consist of cycle express ways, and therefore the concept of cycle express ways does not exist in NL/BE. This is not a logical conclusion. I have watched videos of some of your cycle express routes on Youtube (F261 and F35), and yes, I agree that those do not consist of cycle express ways. But just because your road planners are manifesting cycle express routes using fietsstraaten, does not mean that cycle express ways do not or could not exist. There is no opposition between the two. We can have cycle express routes and we can have cycle express ways without issues. One is a quality of the route and the other is a quality of the way. There can be an intention or a promise that a cycle express route is built with cycle express ways, and then this promise can be broken and you end up with fietsstraaten; that does not mean that cycle express ways do not exist.

To say that "My country does not have this road class, therefore OSM should not have this road class, even if other countries do have it" sounds very wrong to me. The road class exists, even if it so happens that Norway this one time happens to be the global leader in road class definitions (which I highly doubt, but I am willing to accept it).

Is it valuable to capture in OSM that something is a cycle express route? Yes.

Is it valuable to capture in OSM that something is a cycle express way? Yes.

Can we create a scheme to capture road classes generally? Maybe. It will be a much larger proposal and one that would require magnitudes more work. It is often better to take small steps rather than not doing anything because a gigantic step is possible. --Balchen (talk) 17:31, 7 August 2022 (UTC)

I do not understand your long comment, as it supports my previous suggestion: your tag describes a roadclass, I am not opposed to that, I suggest that a roadclass should be tagged within the excisting scheme of roadclasses. The scheme highway=class excists and my suggestion to use highway=cycle_express does not exclude anyone nor blocks any futere developments. And if it's route it should be tagged as such, using a relation. That is after all the way OSM currently defines roadclasses and routes.

On top of that you added several concerns below this discussion that show the possible conflicts your tag creates, that are not introduced when you make it a real roadclass tag, and using a relation when it is only a route.--Noordfiets (talk) 06:57, 8 August 2022 (UTC)

The scheme highway=* is not really a scheme for road classes. It's a mashup of several different concepts. Road classes in that scheme by your definition are highway=motorway, highway=cycleway, highway=footway, and a few others. Other values, such as highway=trunk and highway=primary, are not road classes. They describe a network function or network hierarchy, and are more similar to your usage of cycle_highway on route relations. See also Proposed_features/cycleway_hierarchy for a more thorough discussion. The most visible consequence of this mashup of road classes and network functions is that motorroad=*, which IS a road class, is not a value for highway=*. Yet other values, such as highway=construction, is a chronological state.

I do not object to introducing a new value highway=cycle_express. It seems this was done quite recently for highway=busway. It means data consumers will not be able to recognize that a cycle expressway is a cyclable way, and so routers and mappers will not be able to use these ways for a time. This is a downside that did not apply to highway=busway, since they are not for general public use. But if consensus were reached on highway=cycle_express, I don't doubt that it would be implemented, so I don't necessarily see that as a road block. --Balchen (talk) 07:32, 8 August 2022 (UTC)

I don't think that is the way to look at highway=nn. A trunk or primary road is a combination of importance in the road network with road construction and as such does not differ from motorway. It is a clearly recognizable type of road. The classes you recognise as beeing real classes are user-classes, not roadclasses. Road construcion and use are related, as in your tag fo expressways.

As for the proposal for a classification of cycleways: imo that is a completely different thing and describes use, not roadclass. It over-complicates things by losing the relation between construction and use. It is a bad proposal. As a cyclist I am not interested in the importance of the cycleway. It does not add any information a cyclist needs.

Enough said. Let's just await the vote.--Noordfiets (talk) 08:53, 8 August 2022 (UTC)

Like I've pointed out before: this is a Norwegian trunk road: https://www.google.no/maps/@60.3428329,6.656998,3a,90y,22.15h,77.3t/data=!3m7!1e1!3m5!1sAOa9PUV3S74Ph6iR5_ZX5A!2e0!6shttps:%2F%2Fstreetviewpixels-pa.googleapis.com%2Fv1%2Fthumbnail%3Fpanoid%3DAOa9PUV3S74Ph6iR5_ZX5A%26cb_client%3Dmaps_sv.tactile.gps%26w%3D203%26h%3D100%26yaw%3D286.13422%26pitch%3D0%26thumbfov%3D100!7i16384!8i8192?hl=no&coh=205409&entry=ttu. Do you clearly recognize this as a trunk road based on road construction? I certainly cannot. Maybe in NL the correlation is always 100%, but that is not the case here. --Balchen (talk) 09:46, 8 August 2022 (UTC)

Combination with cycleway=sidepath

The originally proposed tagging scheme cycleway=expressway conflicts with also tagging the same way as cycleway=sidepath and cycleway=crossing. A non-conflicting tagging scheme cycleway:expressway=yes or cycle_expressway=yes is proposed as an altenative. Even so, the combination of tags can be discussed.

How could cycleway=expressway (or cycleway:expressway=yes) be combined with cycleway=sidepath to indicate that the expressway is attendant to a road? --Balchen (talk) 20:09, 7 August 2022 (UTC)

Meaning of "attendant to"

Firstly, what does it mean to be attendant to? The wiki does not define this.

Linguistically, "attendant" means "accompanying or following as a consequence or result". Synonyms are "consequent, consequential, due (to), resultant".

A cycleway that exists attendant to a road could be interpreted as existing as a consequence of the road existing. This is a natural property when discussing cycle tracks, since they are part of the same road construction (at least in the OSM usage). Cycle tracks can be mapped as part of a road or as a sidepath. To know that a separately mapped cycleway is actually the cycle track part of a road, can be very useful. If you visited, you would not expect to find two roads; you would expect to find one. This is also the purpose of cycleway=separate. The Wikipedia description of cycle track also supports this: "A cycle track, separated bike lane or protected bike lane (sometimes historically referred to as a sidepath)".

A cycle expressway would often not be a constructed part of a parallell road. The opposite is much more likely to be true: that the cycle expressway must be well separated, and regularly take a different trajectory from that of other roads. While a cycle expressway may share constructions such as bridges with other roads or railways, that does not make them attendant to (existing as a consequence of) the railway or the other road.

In densely built cities or in city centres, cycleways and roadways will often share the available space between buildings. Such cycleways could, by national definition or community consensus, be classed as expressways, although they would probably have only some of the common characteristics of an expressway. In these cases, a cycle expressway could also be a sidepath. I see no issue with this being the real-world situation and mapping it accordingly. This proposal does not aim to provide one, strict, absolute, global standard for a cycle expressway.

"Attendant to" could also be interpreted as "next to", in the meaning of accompanying the road, or being a companion of the road. I imagine it is also a potential intuitive interpretation -- that is, without having looked at a dictionary -- based on observing the usage and noticing that sidepaths are always next to a roadway. In this case, we are capturing that two distinctly separate ways are simply running in parallell. That is to say, that they offer travel in the same direction.

If the second interpretation is actually why people use the tag, could someone explain the practical use of this information? --Balchen (talk) 20:09, 7 August 2022 (UTC)

@Balchen: I'm pretty sure "attendant to" is just a mistranslation from German to English that sounded smart enough to stick all these years. The original author probably meant that the sidepath is spatially "associated with" or "accompanied by" the roadway. But if two paths overlap entirely, sharing the same space, I don't think that meets any normal definition of a sidepath. (Incidentally, this is a common term in American English that has a pretty intuitive meaning, but cycleway=sidepath isn't being used in the U.S. [1]) – Minh Nguyễn ^💬 00:53, 8 August 2022 (UTC)

So those instances would be the second interpretation, i.e. simply "next to". In that case, I wonder what information we are actually capturing in OSM? --Balchen (talk) 06:07, 8 August 2022 (UTC)

@Balchen: I would imagine it's similar to the information captured by sidewalk=* on a roadway or footway=sidewalk on a footway. In places where sidepaths and sidewalks are mapped as separate ways, some renderers automatically hide or deemphasize sidewalks and would in principle do the same to sidepaths at some zoom levels. Some pedestrian routers could also omit detailed instructions about sidepaths that would be seen as pedantry. But there's no usage at all in the U.S., where sidewalks and sidepaths are mapped as separate ways as a general rule. So that leaves us with the basic idea that if there's a word for it, there's a tag for it, regardless of immediate practical need. – Minh Nguyễn ^💬 07:47, 8 August 2022 (UTC)

Tagging

cycleway=sidepath is mainly used in Germany, with ~3K ways being tagged, and there are two main schemes:

1. cycleway=sidepath is tagged on the highway=cycleway -- around 1/3 of the usage is this way
2. cycleway=sidepath is tagged on the parallell highway=* -- around 2/3 of the usage is this way

There is also is_sidepath=* with ~13K ways tagged, with the associated is_sidepath:of=* with ~11K ways tagged, a lot of these also in Germany.

If all cycle tracks in the world were upgraded to cycle expressways, in total less than 5% of the tagging would statistically be done in a manner conflicting with cycleway=expressway, and the trend is clearly toward is_sidepath=*, which offers much richer functionality, so this percentage will continue to skew in favour of non-conflicting tagging. Realistically it would be magnitudes less than even 5%, since cycle tracks are rarely cycle expressways. So there seems to be no real conflict in tagging either? --Balchen (talk) 20:09, 7 August 2022 (UTC)

If there is a real conflict, the root problem would be that cycleway=* is a homonymous key. By contrast, footway=* only classifies the current feature, because roadways have a separate sidewalk=* key. But anyhow, cycleway=sidepath;expressway sounds reasonable to me. Hopefully no editor or data consumer is already too tightly wedded to cycleway=sidepath on its own. – Minh Nguyễn ^💬 00:38, 8 August 2022 (UTC)

This is an incomplete interpretation of statistics. Have you have queried how many is_sidepath=* is actually on highway=cycleway and highway=path + bicycle=designated? Not to mention the dominance of footway=sideawlk. Both should use comparable syntax. --- Kovposch (talk) 04:08, 8 August 2022 (UTC)

No, because the original objection was that there is a tagging conflict, and is_sidepath=* is never in a tagging conflict with cycleway=expressway, no matter on which way it is used. --Balchen (talk) 06:01, 8 August 2022 (UTC)

But you claimed "in total less than 5% of the tagging would statistically be done in a manner conflicting with cycleway=expressway, and the trend is clearly toward is_sidepath=*, which offers much richer functionality, so this percentage will continue to skew in favour of non-conflicting tagging." on cycleway=sidepath? You find that there is no conflict because something else can be used, but you are not solving that here. --- Kovposch (talk) 09:57, 8 August 2022 (UTC)

Symbolically, it looks contradictory to have an example photo indicating this problem. --- Kovposch (talk) 10:05, 8 August 2022 (UTC)

For the record, I don't agree is_sidepath:of=* is "richer functionality" when it may not handle multiple different parallel features on both sides. Would that be highly meaningful functionality? How do you decide what is associated, or more important? is_sidepath:of:name:left=* doesn't seem to handle a road with name:left=* either, while in this case name=* can show the side's name directly.. handeSomehow is_sidepath:of=river is mentioned together... Despite this not being a common practice yet, ~125k footway=sidewalk is already used with name=* at a top level, compared to the ~10k is_sidepath:of:name=* at lower level that would be found on highway=path and highway=cycleway as well. In extension, I can imagine a sidepath having a name different from the associated road, but that distinction is not explicitly represented by these tags either. If you want more functionality, sidewalk:*=* and sidepath:*=* can also be explored. --- Kovposch (talk) 04:20, 8 August 2022 (UTC)

Combination with cycleway=crossing

How could cycleway=expressway (or cycleway:expressway=yes) be combined with cycleway=crossing to indicate that the expressway is crossing a road?

What is a road crossing?

I have to emphasize that what I write below is based on the road jurisdictions that I know. There are a lot that I don't know, and principles may be different there. Please supplement with details from other jurisdictions if that is the case

A cycleway is defined by a traffic sign, and the effect of traffic signs end at junctions. A cycleway therefore by default ends at a junction with another roadway. Upon leaving the cycleway and entering the other roadway, cyclists who want to continue straight ahead are not seen as travelling on the intersecting roadway, and do not enjoy the legal status conferred to travellers (and cyclists) on the roadway. They are instead crossing the intersecting roadway, and are under a completely different legal regime. This legal regime lasts until the crossing is completed and the cyclists enter a new cycleway, as defined by the appropriate sign. The other roadway, however, does not end at this junction. The traffic signs on that road maintain their legal status through the entire area, uninterrupted.

The cycleway ends; it does not cross. It is the cyclists who are crossing the intersecting road. For this reason, we have a concept of a cycle crossing (this is the legal term). The same principle applies to sidewalks and footways. We have the concept of a pedestrian crossing, not a footway crossing or a sidewalk crossing. In this sense, the OSM nomenclature is a bit misleading. Travellers on the crossing may or may not have priority, depending on the markings and rules of the crossing.

It is also possible that a cycleway continues uninterrupted through a junction. This happens when the cycleway is crossed by an access road (to a property or a business) or a private road (not open to the public). In this case, the cycleway does not end, and the cycleway (and any travellers on it) has priority, even if there are no markings.

How do we map this?

When two ways intersect in OSM, there is a junction node. For roadways, this is normally all that there is. The junction node may be tagged with traffic lights, or there may be separate traffic light nodes scattered about, but the way segments of the intersecting roadways are normally not split up and tagged as being highway=crossing.

A cycleway, on the other hand, ends at the junction, and then a separate cycleway appears on the other side. The part that would physically intersect with the other roadway doesn't exist. When the roadways are drawn as lines with no width, such as in OSM, this really can't be captured. The cycleway must be drawn continuously to the junction node at the centre of the other roadway. When micromapping, on the other hand, it is possible to split the cycleway into several segments, where the segments on either side of the junction node -- that don't really exist legally or physically -- can be tagged with cycleway=crossing, so that we know to treat these segments differently -- if we want to. They offer continuity for routers and mappers, but really they are describing way that don't exist.

As for the other types of junctions, with access roads and private roads, these cycleway junctions could be mapped with the cycleway continuing uninterrupted, like in the real world, and there is no need to micromap segments of it with cycleway=crossing, since at no point is there a crossing -- it's a junction.

How does this translate to cycle expressway

1. If the cycle expressway is seen as being continuous through the junction, it can be mapped as being continuous. This means no cycleway=crossing. In this scenario, it makes most sense that the cycle expressway also has priority.
2. If the cycle expressway is seen as legally ending, and the cyclists are crossing, it can be mapped as cycleway=crossing. In these cases, it would make sense that there is no cycleway=expressway, since it really has legally ended, and what cycleway=crossing does is really map a cycle crossing.

Let me know if that makes sense. --Balchen (talk) 05:58, 8 August 2022 (UTC)

This is a very thorough explanation. I can only speak for the U.S., a jurisdiction that not only eschews the Vienna Convention but also mostly has a legal system based in common law. If the situation in the U.S. doesn't fit into the model you've expressed, it's probably OK because the cycle expressway concept is but a glimmer in the eyes of cycling advocates here – Seattle has a few high-performance, commuter-centric bike trails paralleling freeways that might qualify, but there's not much else.

In general, sidepaths and off-road bike paths are not very standardized, especially in terms of signs. highway=cycleway is mostly distinguished from highway=path bicycle=designated by people calling it a bikeway or bike path as opposed to something else in everyday conversation. Traffic laws vary by state, but I don't think they would consider a bikeway to dump the cyclist into ordinary roadway before continuing. It's a crossing precisely because the bikeway passes through the roadway's legal right of way (landuse=highway) and neither is totally interrupted. Generally, a pedestrian or cyclist must yield the right of way (priority) to motorists, but once in the crosswalk, they have the right of way over motorists.

It seems like what you're trying to do is find a rationale to avoid the need for cycleway=crossing;expressway. But I think it's easier to cite precedent: option (1) is consistent with aeroway=aircraft_crossing and railway=level_crossing, while option (2) is consistent with footway=crossing and cycleway=crossing at highway=crossing or railway=crossing. When an ordinary highway=footway crosses a road mid-block, it briefly becomes a footway=crossing. Similarly, when a highway=footway footway=sidewalk crosses a road at an intersection, it briefly becomes a footway=crossing too. I don't think this is really because of fine legal distinctions; it's probably because people primarily want an easy way to associate the crossing with road furniture along the bikeway (kerbs, islands, etc.), as opposed to road furniture along the roadway.

 – Minh Nguyễn ^💬 08:55, 8 August 2022 (UTC)

1. But highway=crossing is exactly not railway=level_crossing or something else. There is nothing stopping the possibility of a proposal to do that either, cf junction=intersection if it is not considered a "crossing"; or when someone decides to draw area:highway=crossing.

For driveways, it is not that there is no "crossing", but that it is "informal". It is not uncommon to see highway=crossing being added to driveways, let alone street corners without formal crosswalks.

In the case here, there is indeed a formal crossing. Then why would there be a highway=crossing as suggested, while a cycleway=crossing is not "allowed" here? --- Kovposch (talk) 09:53, 8 August 2022 (UTC)

@Minh_Nguyen @Kovposch It's not merely me trying to find a rationale to avoid the need for a crossing, although that is of course also a way to look at it. The principle of the issue been on my mind this summer when I spent 4 weeks doing QA on mapped cycleways in my county. Here is a typical modern cycleway being "ended": [2]. Compare to this: [3], which is the exact opposite. Compare to junctions with access roads: [4] [5] and finally you have these... [6] and these.... [7] (turn 90 degrees right to see the continuation). The second example, where the cycleway is continuous, is not mapped as a crossing.

For the sake of this proposal, though, it makes no real difference. I would be very happy to vote for any of the proposed non-conflicting tagging schemes, so you could easily do e.g. cycleway:express=yes + cycleway=crossing. I thought it would be prudent to also discuss how you would potentially address this. I think we will see that cycleways in the future increasingly will be treated as regular roadways, and so a lot of the implicit second-hand treatment that we take for granted today will need to be revisited at some point. --Balchen (talk) 10:39, 8 August 2022 (UTC)

@Balchen: On the other hand, maybe one of the many proposals to explicitly tag the part of a roadway within an intersection will finally gain traction. Or we'll finally reach the point where mapping roadways and pathways as areas becomes normal. Personally, I don't see the practice of mapping footway=crossing ways as demeaning to pedestrians at all. If anything, mapping pedestrian-centric details like Proposed features/crossing:markings is empowering to pedestrians and pedestrian advocates; likewise for cyclists and cycling-centric details.

In the U.S., unmarked crossings are very common. It wouldn't be particularly surprising to come across a segregated shared-use path where the pedestrian crossing lane is marked but the bike crossing lane is not. (That said, segregated shared-use paths are relatively rare over here.)

 – Minh Nguyễn ^💬 01:56, 9 August 2022 (UTC)

@Minh_Nguyen I'm sorry, I think I phrased that unclearly. I didn't mean that OSM mapping of cycleways is second-hand treatment; I meant the real-world treatment that cycleways get from road planners and builders is second-hand -- at least in Norway. I do all of the public consultation on proposed plans for the regional chapter of the cycling union. I see a lot of ... things. --Balchen (talk) 06:04, 9 August 2022 (UTC)

A Finnish perspective

Personally I wholeheartedly approve this new tag, though I can also understand much of the criticism already expertly expresssed above. For a Finnish perspective, we do have such cycleways in Finland (mainly in Helsinki, but also elsewhere). These cycling superhighways have been branded "Baana"s (of "Banor" in Swedish) here. These ways are currently not tagged in any particular manner, and their status as "expressways" is only inferable from the name-tag that contains the word "Baana". So a specific tag for these expressways is in my opinion a good idea. Regarding the possible namespace clash with cycleway=crossing, could it be thought that if an expressway contains a crossing, it simply ceases to be an expressway at that node or section? Currently the proposal suggests that one criterion for an expressway would be that it contains only a few at-gare crossings. I think that this is a somewhat suspicious or at the very least a highly ambiguous criterion. What number of crossings constitutes 'few', or at what interval could such crossings exist for them to be considered 'only a few' in number? If the expressway would simply cease to be an expressway at a crossing, no ambiguity would exist and the frequency/infrequency on crossings on an expressway would be directly observable (for any interval of the expressway) from the data/on the map. Does this make sense? There's a lively discussion on the Baanas happening in the Finnish section of the OSM Forum, so any Finns reading this are naturally invited to comment this tag here also! --Tolstoi21 (talk) 07:35, 16 August 2022 (UTC)

Yes, back in the analogous world of "expressways for cars" (as distinct from motorways that are almost always exclusively for cars), "crossings" are allowed (as they are one of the distinguishing characteristics separating expressway and motorway), but they should be "few and far between." If they are not a "cloverleaf interchange" that has a bridge / under-crossing (no at-grade crossing) they likely have traffic_signals instead. As these should "rarely exist," these should be spaced far apart. For bicycles, I could see either a crossing meaning "not an expressway through here" (maybe with "begin expressway" and "end expressway" signage / legal meanings) or a crossing meaning "this is allowed on a cycle expressway, but just barely, so you won't see one of these for many more kilometers, if at all." Something like that. Thank you for wider (Finnish!) perspective! Stevea (talk) 04:57, 18 August 2022 (UTC)