Dashed/Soft Cycle Lanes

I am trying to achieve an agreement on an extension of the cycleway tagging scheme. Specifically on how to tag cycle lanes with dashed road markings, also simply called “dashed cycle lanes” (source).

Since it worked so well with the recent discourse on sett vs cobblestone, let’s try to repeat this success and find a solution for this tagging.

Let’s start with a picture:

For the sake of that there are no misunderstandings during discussion here, let’s use the following words in this topic because many different names are used for these things:
A - bikelane
B - dashed cycle lane
C - shared lane
(cars share road explicitly with bikes)

Why distinguish bikelanes from dashed cycle lanes?

In general, the difference between A and B is, that dashed cycle lanes are less strictly demarcated from the rest of the road than bikelanes. The legislations in different countries usually distinguish between these two as well. How they differ exactly varies from country to country, but in general, cyclists are not obligated to stay on these lanes and cars may cross into the lane (and stop on it). It is a soft segregation.
(For more details on the legislations, there is an overview on the German Wikipedia (translated), scroll down to “International Overview”.)

In practice, dashed cycle lanes are often used where there is not enough space on the road for both normal car lanes and additionally “proper” bikelanes: i.e. two cars cannot pass each other without one of them having to cross into the cycle lane. This, the fact that they are often quite slim and/or right next to parking cars with no safety zone and that there is ignorance over that they are usually as soft of a demarcation for cars as for cyclists, has made dashed cycle lanes extremely unpopular amongst cyclists because they are seen as both dangerous and stressful.
Some users suggest that dashed cycle lanes should be treated by router applications worse than shared lanes or even worse than no cycleway at all. (example)

In a nutshell, it is necessary to distinguish bikelanes from dashed cycle lanes because

  1. in many countries, a different legislation applies to each of them

  2. practically, it results in a different situation on the road

What is the current situation?

The wiki currently does not make a distinction between bikelanes and dashed cycle lanes and mentions in the paragraph for cycleway=lane:

Furthermore, the wiki page on Tag:cycleway=lane actually shows a dashed cycle lane. (Note: The German version of the same page shows a picture of a bikelane, but at the same time mentions that “It is demarcated from the road by a solid or dashed lane.”).
Basically, the wiki is pretty clear on that dashed cycle lanes should (also) be mapped as Tag:cycleway=lane.

Still, some people (in the German forum) previously argued for tagging dashed cycle lanes as shared lanes, so there might be instances where dashed cycle lanes have been tagged as shared_lane, especially in Germany. Also, some users from Germany report that most dashed cycle lanes they have seen have been mapped as shared lanes.
This only happened because in Germany, the concept of shared lanes does not officially exist and only recently, towns started to print those shared lane markings on the road.


Now, here it gets interesting - how to tag it? There are two decisions that have to be made - what key to use, and what value to use. From an earlier discussion on the German forum, which did not reach any consensus, I will consequently reiterate the arguments for/against each decision item.

At this point, I should mention that there was a proposal for soft lanes in 2014 which was rejected, mainly because it would change the meaning of existing tags (cycleway=lane) and because there was a (slight?) preference for the use of subtags instead.

A. The Key
Basically, as with any extension of an existing tagging scheme, to not change the meaning of the existing tags, there are two options (examples given are just examples, more to that further below!):

  1. Phase out cycleway=lane and replace it with i.e. soft_lane and strict_lane. This is the same process as with landuse=farm to farmland and farmyard or surface=cobblestone to sett and unhewn_cobblestone

  2. Define two subtags for cycleway=lane, i.e. cycleway:lane=soft and cycleway:lane=strict to distinguish between bikelanes and dashed cycle lanes

  1. Pro own tag:
  1. more straightforward to map. Important detail is not hidden in subtag.

  2. as explained initially, in practice (and in some countries in legislation), dashed cycle lanes are in effect somewhat different from bikelanes, this difference may be big enough to justify using two separate tags.

  1. Pro subtags:
  1. it is more backward compatible for data consumers that only know how to handle cycleway=lane

  2. introduction of own tag would be a large re-tagging effort (246k ways)

  3. introduction of own tag was rejected in the past, partly because of a preference for subtags

  4. introduction of own tag would further complicate the old mapping style for contraflow cycle lanes in oneway streets: i.e. cycleway=opposite_soft_lane and cycleway=opposite_hard_lane would be necessary. (Not a strong argument, as there is a better alternative to cycleway=opposite_* now anyway.)

  5. the appearance of dashed cycle lanes is very similar to that of bikelanes

B. The Value
There have been a few proposals on how to name the values of such a tag. The main two groups are:

  1. by appearance of markings: dashed, solid/continuous

  2. by meaning of markings: soft/shared and hard/strict/exclusive

  1. Pro appearance:
  1. mapping how it appears on the ground supports verifiability.

  2. it may not be immediately clear for surveyors that when they see cycle lanes with dashed lanes, they should tag it as “soft” (as they do not see anything “soft” there)

  3. legislations differ per country and may change, i.e. in some countries, a cycleway with solid lane markings may not be exclusive to cyclists after all or the other way round. Tagging by meaning implies meanings that may not be there

  1. Pro meaning:
  1. bikelanes (=with solid lane markings) will be dashed nevertheless on an intersection or at the entrance of driveways. Users might be tempted to micro-map these sections as dashed when they tag it by appearance but less so if it is tagged by meaning. (Micro-mapping these section is not desired)

  2. the presence of dashed cycle lanes (over bikelanes) carries meaning beyond how the road lane markings look like but also tells something about the situation of cyclists on the road (quality, usability, safety). I.e. higway tags also often carry the meaning of a road, not the appearance, such as residential or unclassified

Alright, that’s it! Did I miss anything pro- or contra?

I envision the procedure now as follows: First, this is open for discussion. Then, in case not everyone has unanimously the same preference, we do a vote on the key and on the value.

Since noone voiced his opinion yet, I am going to be a bit cheeky and say that this is the best solution. Any objections?:

cycleway:lane=dashed / continuous

I value backward compatibility and avoidance of retagging efforts over the mostly opinion-based argumentation (~“difference is important enough”) for a new top-level-key.
Similarly, I find on-the-ground verifiability by tagging what is visible on the ground more important than possibly seeming to encourage micro-mapping by how the value is named.

I think Tagging list is a place where you should talk about tagging schemes.

Why, so it can be “bike shedded” to death there instead of here?

So more people are aware of it.

I like the idea of a subtag how westnordost proposed it. It helps a lot adding this fine detail on all currently already tagged bike lanes without generating a real mess. :slight_smile:

I support cycleway:lane=dashed / continuous

  1. no need to review/retag every single cycleway=lane

  2. people not caring about distinction will not be forced to care about it (at least in Poland there is no real difference between these two types and single cycleway lane may have segment of both types what would make painful to tag this).

“cycleway:lane=dashed” is meaningless. It tells the user exactly one thing: that there’s a dashed line between the bicycle lane and the nearest vehicle lane. It says absolutely nothing about the meaning of that line. Locally, it means the bicycle lane is ending, but judging from the other posts, in at least parts of Europe, it apparently means the bicycle lane is semi-shared with motorized traffic.

As a motorized-vehicle equivalent, it would be like tagging a road as “centerline=white_dashes”. Here in the US, it means you’re on a one-way multi-lane road and lane changes are permitted. In other places it means you’re on a two-way multi-lane road (or you’re at an airport and you should probably drive off the runway before an airplane arrives).

If you want to tag road markings for any purpose other than providing rendering hints, figure out the meaning of those markings and pick a tag that reflects that meaning.

And that is what a dashed line says its is “shared”, if you want to call it like that, with a different vehicle, mostly cars. The motorized vehicle equivalent has no value here, because its called cycleway.

In the UK, a dashed line is advisory, whereas a solid line is mandatory. I believe a solid line, in this context, is only mandatory for motor vehicles.

Advisory means that there is no offence committed when motor vehicles encroach on the cycle lane.

Unfortunately, it also means cars can park on the cycle lane, so many advisory cycle lanes are useless for cyclists.

In Belgium, shared cycleways are marked with a continuous red paint, see e.g. https://www.mapillary.com/map/im/U4eN0FazGTZTtycovXdfKQ

What do you mean with shared? cycleway=shared_lane?

First of all, i support the tagging scheme proposed by @westnordost (cycleway:lane=dashed / continuous). The reasons given look “solid” to me :wink:

A dashed lane has the meaning you want to give it. If its an ending cycleway, then it still means that cars can cross the dashed line. However, these are probably only few meters and would definitly be “micro mapping”.

Though, if the example posted by escada shows a cycle lane with a soft demarcation, then if we tag by appearance, as I proposed, we would need give this red stripe another tagging, as it is clearly not dashed but its *meaning *should still be a soft demarcation. cycleway:lane=red or something. This somewhat seems wrong to me, hmm :confused:

SimonPoole brought the by-appearance tagging up in a previous topic, I wonder what would be his 5 cent on the information on the Belgian situation.

The suggestion was more along the lines of “here’s a Gordian Knot approach to the problem if you can’t agree on something sensible”.

The problem is, as so often with bicycle related tagging, that we have parties wanting to make an advocacy statement, in this case that non-exclusive use cycle lanes are bad, in OSM. It isn’t that I don’t agree, just that OSM is the wrong place to voice that opinion. A further problem is that one of the obvious two adjectives that could be used to describe the properties of Euro-style bicycle lanes: exclusive and shared, shared is already in use for a slightly different concept (shared lanes as in the US).


I oppose cycleway:lane=dashed/continuous because a dashed cycle lane does not mean the same thing everywhere.

In Norway a dashed line is the normal marking for cycle lanes. Only bikes are allowed to drive in the cycle lane. Other vehicles, such as cars or mopeds, are not allowed there. A solid cycle lane line simply means that cyclists are not allowed to change lanes.

Tractor, could it be that you misunderstood something? The reasoning behind cycleway:lane=dashed/continuous in the thread starter is exactly the reason you state: that in practice and in legislation, the meaning of a dashed line can differ from country to country. The dashed value does not make any statement about the meaning of it, it only states that visually, there is a dashed line.

Okaay, I think I gradually get why the community hasn’t been able to agree on a tag yet …

…and that is because it is viewed from different angles here.

My angle is the angle of the surveyor, the angle of verifiability: The surveyor sees a dashed line - ok - and wants to input this information into the OSM database. It would be quite a lot to ask from each and every surveyor that he knows about the exact legislation and input this accordingly. In other words, if you require (this) expert knowledge from the surveyor, you can count on that this will only ever very sporadically (or wrongly) be mapped and thus, the information is pretty worthless for any data user.
The other angle (yours, I guess?) is that of a data user. In the end, he is just interested in the definite information: is the cycleway exclusive, is it compulsory for use, is it of a certain standard etc… Sure, it would be nice to have this information “pre-packaged” like this, but in the end, it is derived information. Derived from where? From the visual appearance, as surveyed by surveyors, plus country specific legislation.
It comes down to one fundamental question: Whose job is it to derive the on the ground information to usable information - The surveyor’s, or the data user’s?

As the author of a surveyor assisting application, I may be biased, but I value easy contribution, verifiability and thus ultimately (better) maintainability of the map over easy direct applicability of the data. It is a similar thing with implicit speed limits like i.e. maxspeed:type=NO:urban, really. Data users can not directly make use of that data, they need country-specific metadata that translates that to a concrete speed limit. This “problem” could be solved with **one **meta data library, that any data user (= an application/service) can use.
Surveyors do not have that luxury, and even if they have a surveyor assisting application that could in theory have the meta data to translate the observation to the applicable data, it would mean that surveying a feature can not start before when this meta data is defined in detail for any given country.

Are we on the same page now?

Also, are you sure about the rules as you mentioned them? Did not read anything relevant about it here.

(Slighty off topic now but here is an instance where requiring surveyors to know the legislation really backfired. Read it if you are interested: maxspeed=: in US)

Uh okay, I had the impression that we were on the verge of coming to a consensus before you mentioned the dashed-tagging though. (cycleway:lane=soft/hard)

Sure, but with a subtag, it’d still be distinguishable. cycleway=shared_lane for sharrows and cycleway=lane + cycleway:lane=shared for dashed cycle lanes.

Yes, it is a cycleway=shared_lane. It does not create a real, separated lane, and cars typically drive with the right wheels in this coloured strip.

That’s why I do not like any tagging that is just mentions what it painted. The mapper should know the local traffic law and map the result of that.

That would seem to be a rather high hurdle for contributing to OSM.