Proposal talk:Cycle Hierarchy

From OpenStreetMap Wiki
Jump to navigation Jump to search

I do think that it would be good to have a 'cycle hierarchy' defined, it is something I have been thinking of as well. However, I see several points where this plan is not as I would envision it. Some points:

  • Split this proposal in two, one for the 'difficult' and 'dangerous', one for the actual hierarchy. They are quite different subjects. I will go further into the hierarchy below, as it is the one I'm more interested in
  • I am getting an impression the definitions are too much based on the road definitions. Because most cycling is done over distances upto 10-15 kilometers, finer grain is more important than extensive networks. More clearly said: There can be some more roads in higher grades than specified here
  • Get rid of the 'unclassified', 'useless' and 'trunk' classes. Primary is high enough, in my opinion, and lower-than-tertiary can better be done as a default than as a special class
  • Giving the London cycle superhighways as your example of 'trunk' roads is laughable to a Dutchman such as me. Large parts of it are less 'adapt for a quick and easy circulation of cycles' than the average 'tertiary' in a Dutch city.
  • I'm not very happy with "highway:cycle" as a key. It does of course have some similarity to highway type, but not enough to show that by reuse of the tag. I'd go with "cycleroute" or "cyclelevel" or somesuch.
  • The sample rendering scheme does not feel very well to me either. It would be good to have the scheme usable when it's only adopted in a few places, not having to wait for the rest of the world. As such, it would be better to have this be something 'extra' on the road, so that for example a cycleway can be distinguished from a road, and bigger roads (for cars) are still seen to function as a navigation aid. In short, I'd like to see something that can be used for example on a variation of the existing cycle map (showing the cycle hierarchy rather than cycle routes, but not otherwise changed), rather than resulting in a wholly different map outlook from what we're used to.

- Andre Engels (talk) 08:57, 30 September 2013 (UTC)

The Dutch manual advises a three-level hierarchy for utility routes - the bottom level is "everything", then there are two proper levels - "main cycle routes" and "cycle routes". These are a different concept from the longer-distance leisure routes (which we already have as ncn/rcn and the node networks in NL/BE). In countries where cycle facilities are less-well-developed than the Netherlands, it can be useful to separately identify utility routes based on whether their dominant feature is directness or quietness. In Oxford I've used mcn(=oxford) to tag the main direct routes, lcn relations to tag the main quiet routes and lcn2=yes to tag subsidiary quiet routes. You can see the result at http://www.transportparadise.co.uk/cyclemap/ --RichardMann (talk) 09:43, 30 September 2013 (UTC)


First thanks for making this far-reaching and thought-provoking proposal that tries to address a number of open problems, of which most of the cyclist mappers are aware.

Now to the issues I see with the proposal:

Your concept of a circulation plan (“plan de circulation”) is an unmappable concept. We map what is there,not what the authors of some plan had in mind. This does not mean that your analysis is not correct, but we do non map plans, we map objects. In fact you state this yourself when you say “Despite the fact that road characteristic are confused with circulation plan, that are 2 different concepts “. Then you continue: “a proposal set of tags to define road hierarchy based on circulation plan for cycles “. This is the basic weakness of your proposal – you presuppose a plan of some kind for bicycles. We cannot map this plan.

Your analysis of the highway hierarchy for cyclists is correct, but that is an assessment that the planners need to do who implement elements of the cycling infrastructure. It is not something that is easily visible on the ground (and often does not exist).

What you are describing is the criteria for establishing a network of cycling routes based on the available road infrastructure or for a routing algorithm to find a suitable route using user-defined criteria. In both cases it is the sum of the properties of the highways that make up the route (or the path from A to B in the case of routing) that actually determine the degree of suitability of the route for a certain use. The use can be anything from commuting to school to a trans-Europe bicycle trip. In all cases the route selection process has to analyse the properties of the available highways to come to a result.

In addition to this there are cycle routes, that are largely also mapped in OSM, that are the result of such analysis by some body, responsible for the route. But that body is not the mapper. The mapper can only map what is there and can map routes if signposted or based on data legally obtained from the body responsible for the route.

A word on your rendering examples: if you were to make a printed map based on your rendering examples, you would not be able to navigate it on your bike, because what you see on the ground and what you see on the map do not coincide. If your map shows a motorway-sized cycling super highway, but omits, because you cannot use it, a real motorway or railway or canal, it becomes obvious that this cannot work. Cyclability should be indicated by an additional mark-up on a “normal”, motorised-traffic oriented map. You may de-emphasize the features that are only relevant to motor vehicles, but you cannot omit them. And the cycle super highway needs to be represented with a completely different stile, obviously very visible. Richard Mann's maps in Oxford are surely pointing the way to go.

A completely different issue, in my view is the signalling of danger levels. This is something which we should seriously discuss. It has two components: the danger-level of a cycle road/path/lane can in many cases be directly derived from road properties: If I have highway with bicycle lanes 1.2 meter wide on a 4-lane highway without speed limit then I know that this is dangerous. On the other extreme if my bike route is in a 30km/h zone, I know that my chances of survival are much better. My tendency would be to map the highway properties much better (width, surface, speed limits, smoothness, …) and use this data for bicycle routing.

In addition we may want to have bicycle-specific danger points mapped. But this is tricky because it is difficult to assess: classical example is the quiet residential road with a school. Twice per day it becomes a bicycle nightmare because of the parents' SUVs stopping/parking their monster cars as close as possible to the zebra crossing leading to the school entrance. This road becomes red-dangerous twice a day for 20 minutes.

In conclusion: I do not agree with the proposal, but agree with much (but not all) of the analysis.

--voschix (talk) 09:07, 1 October 2013 (UTC)

@ Andre Engels

Split the proposal
We can of course do that easily. I don't see any interest to do that but if we do that it is not a problem.
On the other hand, if we split the proposal and only one of the two proposals is adopted, it is a problem. I think "difficult" and "dangerous" are some parts of hierarchy. There is no equivalent for car hierarchy because the road is adapted for cars, but it is a part of hierarchy. There is no sense to tag the circulation plan without difficult and danger.
More roads in higher grades
OK. One solution is to have more road in higher grade. An other solution is to have shorter range in the definition of each grade of hierarchy. You said "the definitions are too much based on the road definitions". I did that to avoid too much subjectivity. Could you propose others definitions ?
lower-than-tertiary can better be done as a default
That is possible.
For "unclassified", I dit this tag to allows making difference between road at the lowest range of cycle hierarchy and road with cycle hierarchy not tagged, but this tag is not absolutely necessary.
For "useless", this tag has exactly the same utility than "service" in car hierarchy. Try to replace all "highway=service" tags by "highway=unclassified" and look at the map you did. You will say "there is a problem".
London cycle superhighways
Cool ! You will easily give a better exemple for "trunk". OK, this tag is not vital for cycle circulation plan.
not very happy with "highway=cycle" as a key
I did this choice for educational reason : this key is explicit. It suggest "that is the same as "highway" but for cycle". Of course this choice may be not the better.
The sample rendering scheme
I worked a lot to perfect the set of tag, but I did't work on rendering. To be quick I just copy rendering sample from highway=* tags, but we can easily improve it.
I don't understand why you speak about wait for the rest of the world. Of course, any rendering scheme would run on a test rendering server first. This is a discussion of potentials for now. We are a long way away from replacing OpenCycleMap. I think we will always have to deal with zones where cycle hierarchy is not tagged, that is why I made a "unclassified" tag and not a default grade.
We can not only add cycles lanes and cycleways on a car map. We can not only change color of roads of a car map to show cycle circulation plan. On a car map there is several informations without interest for cycles and taking a lot of place (the best example is highway=motorway). We have to draw a cycle map where we choose witch information we put and witch size we give at each information. We must have motorways drawn as a barrier and not as a big circulation way. We must have big boulevards drawns tiny and quiet streets drawn large. You can see a first draft of render on this poster : http://www.suna.fdn.fr/OSM/hierarchie-cyclable-V3-EN.odp Of course we will have to improve a lot this traft.

- ahmster 15:02, 1 October 2013 (UTC)

Too complex classification for cycling

IMHO what you propose is too complex and has a structure that goes too deep into differentiation. I do not completely oppose further refinement in cyclebility but we already have a lot of structure for cyclist with the official routes (national, regional, local) and with the actual highway-classification: one of the first things you point out is that the "highway"-classification was a classification made for motorized vehicles, but this is not completely true, instead, it is the classification of the road network (and admittedly as of today roads are built mainly for motorized vehicles although they were already built also before the invention of the motor). Indeed with a bit of local knowledge this classification can also be used to infer the suitability for other means of transport like bicycles or pedestrians. As a second point, why doubling the classification of all highway-classes for cyclists as well doesn't make sense IMHO, the main reason limiting cyclists in speed are not road limits, but they are usually limited by their physical strength, i.e. a cycle-trunk won't allow for higher speeds as (almost) every other cycleway. --Dieterdreist (talk) 13:53, 2 October 2013 (UTC)

Not the answer

As someone who is building a new cycling map and router with OSM data (coming real soon now :) ) I'm afraid this proposal isn't helpful and is likely to be actively detrimental.

OSM tagging should always be verifiable. That is, "a tag/value combination is verifiable if and only if independent users when observing the same feature would make the same observation every time".

The highway= tag, when correctly defined, passes this test. For example, in the UK, motorways are always highway=motorway, A roads with green signs are always highway=trunk, A roads with black/white signs are always highway=primary, and so on. This is not at all subjective.

This might be true for the UK, it isn't for the rest of the world. Many countries, including Germany and Italy, have different hierachies for planning and maintenance. What you can see from the colour of the signs / letters in the refs is not in every case the importance for the road network but merely a reference to who operates (pays for the maintenance) of a road, e.g. the state, the region etc. Of course there is some relationship, but it is not 1:1. Additionally there are further complications for political reasons (the state passing road maintenance to the region or province merely for financial reasons or rich regions building and maintaining better and more important / capacitive roads than parallel national roads). Anyway, there are guidelines and standards that document the characteristics of the road network and define road classes e.g. in accordance to their importance for the road network / connecting function, but they do not corrispond with what you can find on the ground. --Dieterdreist (talk) 11:08, 3 October 2013 (UTC)

This proposed cycle hierarchy would fail this test. As an example, what would you tag as "highway:cycle=primary" between Wolverhampton and Birmingham? There is a continuous path, open to cyclists, along the Birmingham Main Line canal; and for many cyclists, this would be the primary route. But other cyclists, such as those on road bikes, find the canal path slow and unpleasant, and would prefer a route roughly following the A41, or the A457.

I think your example of "cycle superhighway, as it exists in London" unwittingly proves this point. These are believed by many cyclists to be "hopeless", "a disgrace", "a load of rubbish", "the biggest joke ever played on London's cyclists", etc. Yet evidently some people - Transport for London, at least - think they are "highway:cycle=trunk". How do you reconcile that?

Fortunately, there is no need for an explicit hierarchy. We can create excellent cycling cartography and routing with OSM data using factual information:

  • dedicated cycleways (widely mapped in OSM)
  • signposted cycle networks, such as the UK's National Cycle Network (widely mapped in OSM)
  • surface (partially mapped in OSM)
  • maximum speed of other traffic (often mapped in OSM)
  • traffic levels (rarely mapped in OSM)

Where this is present in a city, you can create really, really good cycle routing and cartography that fulfils the needs of many cyclists. I would far rather see more people tagging this information - particularly traffic levels - than introduce a subjective scale where no two cyclists are likely to agree.

--Richard (talk) 15:07, 2 October 2013 (UTC)