Highway=path wiki page additions and updates

The problem is that, even in Germany, it’s not an implicit rule everywhere (as noted already). It’s an assumption that you can make if you don’t have better data. It still makes sense to add an actual tag that says “yes, cycling (or walking, or whatever) is allowed here” if someone has checked all of the myriad of things (is the path in BaWu? is it <2m wide?) that might affect it.


I know, but still many routers treat those access tags the way I described above, even commercial ones:

those examples show that Komoot does not take the 2-m rule into account which in turn means they also make a difference between implicit yes and explicit yes.

Again, I’m not saying we should care, I’m just saying that forcing the use of explicit access rules will certainly break things - just as getting rid of explicit rules in favor of implicit rules broke things as you mentioned above.

Then I guess you must believe that an implicit vs explicit value for any access tagging would differ in meaning. Seems a bit confusing. Perhaps that’s why someone recently came through my area and added bicycle=designated to all the previously mapped highway=cycleway. It seemed entirely unnecessary to me. :grinning:


We have had the discussion implicit vs. explicit several times in the german forum in regard of different tags (in most cases acess=*). The problem with an implicit access=yes (or no) is that it does not give any proof if the mapper is happy with the implicit value or if the mapper just has forgotten to check if the implicit value is correct. You do not have this problem with an explicit value. Anyhow these discussions never came up with a clear consensus as far as I remember. I believe it depends on the given situation which approach is the better one.

1 Like

It seems to me that this discussion is based on different views of what “implies foot=yes / implies bicycle=yes” mean:

  • (a) some interpret this as "when bicycle= is missing on a way the suggested fallback-value for data-users is “bicycle=yes”*
  • (b) others go a step further and interpret this as “if bicycle=yes on a path in the real world, you should NOT tag this on a highway=path in OSM, because this is already implied / default”

That is a very good argument against interpretation (b) -that prohibits tagging implied values-, but not against interpretation (a) -that is aimed at data consumers and not mappers.

If you look at the history of the proposed default values in the country-specific tables in OSM tags for routing/Access restrictions - OpenStreetMap Wiki , you’ll see changes in cases where the interpretation of the default value have changed (even if the traffic/access rules themselves have remained the same) .

When such a change in default value (from A → B, for say bicycle=* on a bridleway) occurs and you believe that this is also a rule for what to tag and what not to tag, you are left with a problem:

  • removing all the tagged (bicycle=B) 's from ways in OSM -because it is the newly declared default- is deleting valid data (where a mapper has concluded that the presumed default indeed applies on that way)
  • adding value (A) to all the bridleways in OSM that are not tagged for bicycle=* is also not a viable option, because you can not tell which of these ways are (1) evaluated for bicycle-access and found to be the same as the default value or (2) which ways have not been evaluated for bicycle-access by a mapper

These problems can be avoided by being clear and explicit to your fellow mappers and data users about access tags and use as little hidden assumptions as possible.

Only consider leaving out tags when they can be inferred by definition (such as bicycle=designated on a highway=cycleway), not by some assumed default that may be interpreted differently by area, by different people, over time and per usecase.

This is the first time that I have heard of this, and I have not read this in the access-documentation either. Surely some people will act this way (I’ve seen stranger things come along), but I would be very hesitant to assume that this is a widespread interpretation internationally, since it seems to be in such clear violation of the described primary scope of the general wiki for key:access :

Access values describe legal permissions/restrictions and should follow ground truth; e.g., signage or legal ruling and not introduce guesswork. It does not describe common or typical use, even if signage is generally ignored.

1 Like

I don’t think so. Komoot answered a few months back in a PM that they -based upon the OSM wiki- just assume you may ride there that when a highway=path is not tagged for bicycle=* . They don’t seem to use a penalty for a missing bicycle=yes like Brouter does.

This despite the data in OSM showing that the majority paths that are tagged for access in that state have a negative value (bicycle=no / private ).

The routing behavior was consistent with this and this was the reason for contacting Komoot after complaints by forest owners about routing suggestions over paths that are prohibited for cycling in real life, but untagged for bicycle in OSM

As noted earlier, on most paths in OSM that are still untagged for bicycle=* in my country cycling is NOT allowed, they just arent’t tagged yet.

So you can not be sure that an untagged way complies to a presumed default, the only way to reduce uncertainty is to be explicit in tags, regardless if they do or don’t comply to a default.


“meaning” is probably the wrong word, but yes, I would argue that there is almost always a difference between implicit vs. explicit. We also add source:maxspeed to speed limits (which are only implied by a road sign) because of this.

No, this is actually an argument for removing unnecessary access tags whenever possible. In BaWu, if you tag a way narrower than 2m with bicycle=no because it is narrower than 2m, then you would also have to add width=x where x < 2m, or source:bicycle=2m rule Baden-Wuerttemberg, because otherwise a change in legislation would cause a massive amount of work.

Yet another argument for keeping it implicit, because otherwise such a change would become impossible.

I fully agree that there are advantages to making it explicit, but I am not quite convinced yet that they outweigh the disadvantages.

I would love to gradually shift to a phase where highway=path becomes analogous to highway=road. This would of course require that we have a consensus about which alternative tags we use to cover most of the use cases that highway=path now fulfills.


In Austria there are 53744 km highway=path, of which

1082 km - 2% are bicycle=yes,
1928 km - 3.5% are bicycle=designated,
5465 km - 10% are bicycle=no and
45132 km - 84% are “not bicycle”=*.

That sums up pretty well, not much lost here. On how many of those paths not tagged for bicycle the implicit bicycle=yes holds?

Along the lines of the proposal process, it must hold at least in three out of four cases (75%) to find approval as a generally applicable statement, that is, at least 33849 km of path on the ground must be bicycle=yes|designated at a minimum, no matter any individual markup.

Spoiler: Out of those 53744 km highway=path:

23492 km - 44% are sac_scale=*

Of those:
192 km - are bicycle=yes and sac_scale=hiking
750 km - are bicycle=no and sac_scale=hiking
3407 km - are bicycle=no and sac_scale=*
3851 km - are bicyle=* and sac_scale=*

That makes it look like cycling is generally not allowed on hiking trails, that make up for close to half of all paths. Will the assumption that cycling is allowed on paths hold in Austria?

(Data scraped from Ohsome - Dashboard)


(is the path in BaWu? is it <2m wide?)

if it is width>2m and no explicit access (designated), it is probably a track.

1 Like


  • if you explicitely state an implicit rule, what do you do if the implicit rule changes later on?

That is a very good argument against interpretation (b) -that prohibits tagging implied values-, but not against interpretation (a) -that is aimed at data consumers and not mappers.

it is not an argument against tagging, you could tag that it is implicit.

1 Like

Or are they being marked bicycle=no by people that think “This is a rough walking track, so you couldn’t possibly / comfortably ride a bike along here”?


I would be very surprised if this was true. Here two fastbike examples that - in my opinion - are impossible to explain if Komoot does not penalize missing bicycle=yes tags:

Below you can see what they interpret into a 220 yd long path segment, depending on whether a bicycle=yes tag is present or not:

I have given this some more thought and what I wanted to say is: there seems to be a strong correlation between the presence of foot or bicycle access tags and the walkability or bikeability of the path - and at least some routers seem to exploit that.

I actually do both, but I also had to realize that even in Germany, which presumably has one of the highest densities of active mappers, in the world - high quality surface and smoothness data is at best wishful thinking and that probably the majority of all bicycle=no or bicycle=yes tags (which were not deduced from an actual traffic sign) are actually wrong.

It would be interesting to know how many mappers actually know what bicycle=yes is really supposed to mean.


Lots of them are in the woods. Cycling there generally is not allowed. So even if tagged for the wrong reason, result will be right :slight_smile:

Your analysis is supported by these statements on the general rules on cycling on paths in the forests as summarised here

In a nutshell: Mountain biking is forbidden in Austrian forests unless it is explicitly permitted.

Which is based on a reference to §33 of the Austrian Forest Act which reads

C. Use of the forest for recreational purposes

types of use
§ 33.

(1) Notwithstanding the provisions of paragraphs 2 and 3 and Article 34, anyone may enter and remain in the forest for recreational purposes.
(3) Use that goes beyond paragraph 1, such as camping in the dark, camping, driving or horseback riding, is only permitted with the consent of the forest owner, with regard to forest roads with the consent of the person who is responsible for maintaining the forest road. […]

These are two things - the value itself and the trust level of the value. But also an explicit value may be wrong.

1 Like

I think that is a fair assesment.
That correlation could also (partially) be explained / supported by the fact that in some areas there are regulations where the “bikeability” is a factor that determines whether a path is or is not legally accessible (and if access is legally enshrined or not).

Two examples:
Nordrhein-Westfalen official state website ( Radfahren im Wald → Rechtslage)

What is meant by “fixed trails” was clarified by the Administrative Court of Cologne in a ruling (14 K 5008/07) on 02.12.2008: “Fixed trails” according to § 2 para. 2 of the NRW State Forestry Act are not only artificially paved trails, but also trails with a naturally solid subsoil, which are also suitable for bicycle traffic in the forest due to their width. Typical “single trails” outside woodland trails , which are popular with mountain bikers, are not fixed paths in the sense of the jurisdiction and the law. Their use is therefore not permitted.
The suitability of trails for cycling is judged not only by the acquisition of the surface, but also by whether cycling can lead to the disturbance of other recreationists. Therefore, narrow paths and trails that do not provide sufficient space for cyclists to meet other forest visitors may not be used by bicycle.

Dutch roads act
(not to be confused with the road traffic law)

The existence of a restriction in the use, other than by virtue of a legal provision regulating traffic, may also be assumed on the basis of the condition of the road and of the use which is customarily made of the road.


Perhaps part of the problem is that the wiki currently only shows well-behaved examples. If both casual mappers and data consumers were aware of the fact that both of the following paths:

could be tagged like this (without necessarily violating the wiki):


mappers might become more inclined to also add smoothness tags and data consumers might become more careful when evaluating the access tags.

Yes and for example in Bavaria this ruleset changed less than 2 years ago and the wall of text describing it contains things like:

If there is a risk that travel on the path will loosen the soil surface, increasing the risk of soil loss and soil erotion on the path, the path is regularly unsuitable for travel by bicycles or other non-powered vehicles.

Width, gradient, curves and clarity are to be considered, also in connection with the frequentation of the path by other nature users.

The suitability of a path may also be limited to the time of day or certain periods of a day for reasons of community compatibility.

This applies, for example, to unpaved paths that lead across alpine pastures where there are animals. Particularly during the nighttime (between sunset and sunrise), entering these paths can trigger panic reactions in grazing animals, leading to injuries and damage.

In my opinion we should not even try to boil stuff like this down into foot=yes/no or bicycle=yes/no, because the correct answer in such cases will always be foot=maybe and bicycle=maybe, no matter how much we do not like it.


I think it is time update those “default access tables” that were linked above. If a tag on a stretch of path is that much difficult to tag, no way there can be country wide defaults. In doubt though, I rather trust local knowledge will get tagged with enough consideration to be much more true to the ground than defaults from the heavens.

1 Like

Some data consumers will do nothing if they can get away with it. Sometimes (as here) attention needs to be drawn to grossly inappropriate routing. No-one says it’s easy (in GB you need different rules in Scotland and England+Wales) but working out whether it is legal to send a cyclist somewhere is literally the least you have to do.

Things like the above (“OSM-based router sends cyclists in error through the Royal Parks”) reflect badly on everyone else in OSM.