Consuming highway=path, Take 2

There are several hundred surface=compacted in Spain and Portugal alone, and some more surface=unpaved. I didn’t check other unpaved surfaces. From randomly looking at a few of these, a significant proportion are former railways. (My.spot checks also brought to my attention that horse_scale is a thing…)

3 Likes

Hi

Guilty as charged! Although, I sound even more hyperbolic if you cut my quote mid-sentence. The rest of that sentence was:

So, yeah. From a highway=cycleway tag alone I might infer that the surface is probably not ground or dirt, that its smoothness is probably not bad or horrible and that its at least as wide as most handlebars. That’s not enough for me and my beloved roadie. So I threat these Duck Tagged cycleways as impassable, and drop them from my routes.

More to the point & topic at hand: couldn’t one similarly say that a highway=path is OK, because you clearly can walk on all of the examples posted? It’s true that you might not want to take your leather-soled Oxfords to some of them (compare: “you might not want to take your road bike to all the Duck Tagged cycleways”). I mean, didn’t this fellow report that he “survived” the SAC path?

That we need to emphasize the importance of the extra descriptive tags on highways is a very important point! In my opinion they are also just as valuable on all (or most) highways other than paths.

In the part that I bolded for you, you just proved my point.

If you don’t cycle highway=cycleways, you couldn’t cycle any cycle route in Nederland! Also, there are many unpaved cycleways. I’m pretty sure most of those have a surface and/or smoothness tag. Include highway=cycleway and exclude (or give a very high penalty to) surface=unpaved|ground|dirt|grass|… (not asphalt and not paved) and you’ll be fine on your Cervélo Calidonia-5. Nevertheless, if you really want to race, you need more info and you won’t find that in the OSM database. Maybe a raceability scale could be introduced…

Exactly! Same here in Finland. My rant concerned Duck Tagged cycleways tagged with only highway=cycleway and with no other tags.

In order for anyone to exclude the ones with a grass surface, they would have to include a surface=grass tag, for example.

The discussion at hand is about adding these additional tags to highway=path-tagged ways. My point is that most other highways would benefit greatly from these extra tags as well, and that paths are not that exceptional in this respect.

2 Likes

The difference between path and cycleway is in how likely you are to need mountain rescue if you take the wrong turn because the way was missing important tags.

Anyway, in case anyone else is wondering, here’s an Overpass query that show cycleways that would probably benefit from a visit.

highway=cycleway means “this is a cycleway”, i.e. it’s safe to assume it was constructed to cycleway standards.

cycle.travel routes over highway=cycleway and users are happy. The only two challenges I have are (a) differing expectations on surface (in Europe it’s mostly paved unless there’s a surface/tracktype tag stating otherwise; in rural areas of the US it will often be crushed limestone or similar), and (b) very occasionally people tag an MTB trail centre with highway=cycleway and no surface tags, but that’s clearly a tagging error.

Believe me, compared to the world of hurt that highway=path brings me, highway=cycleway is blissfully simple.

7 Likes

In many biotopes/situations such images would have to be updated frequently and the year & date or season clearly stated if they are to be useful.

1 Like

A big difference is that for cycleways/cyclists:

  1. Road bike users with special surface requirements are a minority. The overwhelming majority of ‘average’ cyclists can use anything tagged as a highway=cycleway just fine.
  2. ‘Bad’ ways tagged as highway=cycleways that are completely unusable for road bikes are probably not that common.
  3. Being routed over a ‘bad’ cycleway is mostly an inconvenience where in the worst case, you need to push your bike.

while for path/pedestrians:

  1. The average pedestrian (= person, almost everyone walks at some point) would have some troubles if a route sent them over a difficult path.
  2. ‘Bad’ highway=paths are comparatively more common than ‘bad’ highway=cycleways
  3. Being led down a difficult (hiking) path that people are not prepared for can easily lead to injuries or require rescue operations.

People (on foot) having to be rescued because they got stuck after following an unknown path on a map is a common occurrence. I’ve never heard of that happening to a cyclist.


People’s perspectives on this likely depend heavily on where they live. Outright dangerous paths in flat regions are bound to be much less common than in mountainous regions.

5 Likes

I think this is missing the point. Legal access is not the issue here, practicality/safety is.

1 Like

The same principle holds for other tags: establish implicit tags and worldwide default values, then tag what is different or missing in other countries and other cases. Defaults should be chosen to include the majority of cases. If no sensible default value can be agreed on, for a particular key that is essential for data use, then all mappers in all countries have to add the tag all the time, else routers will send the end user to Mordor.

1 Like

There are essentially no/very few implicit tags or defaults that can be inferred from highway=path, that’s the problem.

3 Likes

In Austria the bicycle default for path is dismount. As documented on the openstreetmap wiki defaults page.

So, if the world wide default is bicycle=yes, many paths in Austria should get the tag bicycle=dismount. Otherwise only country specific routers and renderers get it right.

I think the country specific defaults tables document what should be explicitly tagged in those countries, in order to “translate” the situation to OSM-wide understood tags.

Do you have any numbers to put behind that statement? This wikipedia page suggests a dozen jurisdictions where that is the case. There are around 200 different countries in the world.

Thats about freedom to roam, not about access to a specific path. Not applicable.

What’s the difference - they’re both about explicit legal access?

Are you saying that even in countries without this sort of legislation there is some other sort of legislation that allows people to cycle anywhere they want, even across private land, or are you saying that “most paths across private land aren’t add to OSM, so we don’t care about those, and only worry about public roads and paths”?

Are you saying the ancient concept of freedom to roam is about cycling or not cycling anywhere you want?
If on a particular physical path this concept somehow disallows cycling even when there is no visible indication, I think that should be tagged explicitly by the people who know such things about these countries.
How else would the data user know?

Can only think of MTB downhill single-trails. In my area these mostly tagged path, some even bicycle=designated, so the “modern” counterpart of cycleway. OSRM, graphhopper, valhalla, they all route there in their bicycle profile regardless the mtb: niche tags set. cycle.travel does not :wink:

If I had any say in that, I’d duck-tag them highway=singletrail.

I’d consider MTB trails tagged as highway=cycleway to be mistagged.

(But regardless of the highway tag, bicycle=designated is correct and unfortunately leads to the situation you describe where many routers consider it a perfectly fine way to route you over with a normal bicycle)

2 Likes