The most prominent consumer (a renderer) devises paths signature by “paved|unpaved|unknown” surface. Curiously, the unknown signature lies between the others. While I do not know how many signatures a normal person can adequately discern, when looking at a map I honestly do consider the choice not sufficient to represent the plenitude of what is on the ground.
It would be much easier for mappers that care about what this topic was started, if footway got used for the “decent” ways in the park and “path” for the others, but this ship sailed long ago, too lazy now to find the quote.
Which one is it that you’re referring to? I’m guessing that one of the styles used by Facebook might be the most popular, or maybe one of the ones with some OSM input (like Apple or Bing in some places). Or do you mean a map on osm.org itself?
In Greater-Toronto-Area-English-as-spoken-by-me, a trail is mostly in a wilderness or a park area; a rail trail is most often out of an urban area, usually gravel; a path is a catch-all; an MUP is a paved cyclable way in an urban area on which pedestrians are also allowed; a walkway is most likely a short connection in an urban area; a promenade is a wide way designed or used for strolling, often along a linear destination like a lake or river shore.
In fact, I’d wager more people both inside and outside of OSM could distinguish a “rail trail” from a “promenade” than could distinguish a Highways England “primary road” from a “trunk road” or a Landstraße from a Bundesstraße if they don’t know the ref.
I understand the history of why OSM ended up with five main highway values for car roads that could, in different regions, actually all end up looking the same, while for pedestrians we have only footway and path which we cannot decide if they are different or not, and pedestrian which is maybe different. But I think it’d be better to just leave it as “well it’s not great but realistically it’s not going to change so here’s what we could do instead” rather than trying to pretend that there’s a systematic, well thought out reason why that came to be.
To me, a footway is a path with foot=designated and bicycle=no. Pedestrian is a street or area with foot=designated and vehicle=no. Other access tags can be added for conditional or additional access.
Hierarchies of paths exist, but only on the level of e.g. single nature parks. In other areas there is no hierarchy at all. I couldn’t build a general hierarchical OSM-wide path tagging system for that.
That is also true for car infrastructure, but that one is
a. always there (or is that western bias?)
b. by country and/or state mostly stable and official
So each state or country’s road hierarchy can be systematically ‘translated’ to the OSM-hierarchy, but there is no such thing for paths.
I think so, yes. One remark was that a path could be narrower or less smooth somewhere on the way, and OSM is excellently equipped to map exactly that: a section of highway=path with the appropriate physical tags. Or steps, or a ferry, a ford, or…
Yes, there can be different types of pedestrian infrastructure (in all kind of dimensions), but there isn’t any widespread “importance” or if you so want “traffic channelling” classification of pedestrian infrastructure (as @SomeoneElse notes there is a bit of that for cycling but that’s it).
Particularly when we are discussing paths in a park.
For a different take: sometimes there is a hierarchy expressed for walking or cycling networks. Page 30 of Scotland’s Cycling by Design defines primary, secondary, local access, and long distance routes. An example of how these four tiers might look:
The main thing to notice here might be the yellow local access routes, which tend to be short and kind of random.
These four tiers are quite subjective and specific to network planning. My only idea to assign some kind of intuitive “importance” to paths in a park would be to estimate which paths are used by the most people. If you had reasonable origin/destination data showing where people in the park start and end, then you could calculate this with a shortest paths algorithm, if you make assumptions about people following such routes. Betweenness centrality is related. Of course in a park, people are not going from A to B in a rush, and their choice of route might be influenced by surface, width, and other physical factors mentioned. But the idea of figuring out which paths lead between major landmarks and entrances/exits, vs little random spur paths or dead-ends, might help.
(I’m not suggesting any of these ideas are anywhere near objective or on-the-ground verifiable enough to be used for tagging; just giving a different perspective.)
You examples would seem to be more of a route kind of thing (for which we do actually have classifications), and not a classification of the physical infrastructure (yes this opens up yet another can of worms :-)).
According to The Wiki, track is defined as designated for predominantly 4-wheeled motor vehicle use, for access to e.g. farmland for harvesting, forest for forestry vehicles, nature reserve for maintenance vehicles. Access could include non-motorized transport methods, such as cycling and walking.
In Nederland, tracks are not typically unpaved. Especially tracks for access to farmland are often paved, and increasingly so. Most of these are not two-tracked, but wide enough for one motor-vehicle. We usually map track:type and surface for tracks.
That said, lots of mappers map all two-tracked roads as tracks, even if they are true connecting roads, part of The Grid. Again, many of these have been paved (two paved wheeltracks with grass or ground in between), and serve cyclists and pedestrians as well as 4-wheeled motor vehicles.
It is always a joy to learn how much regional differences there are in how something that is conceptually the same (at least in OSM terms) looks on the ground (that is what we are supposed to map?)
I was debating this in my head last night (and even asking Copilot what it thought) after running and walking some wooded and heathland “open access” areas in Surrey, UK and then trying to reconcile my “experience”, heat maps and what had been mapped.
There was almost a random mix of highway=footway, path, track which I view as a sort of offroad hierarchy but then also got highway=bridleway as some of the “paths” are designated public bridleway ROWs although the whole area is effectively right to roam so the access nuances get complicated to say the least. In other words by and large all paths are available on foot or bike (and often horse too) so unless explicitly signed having highway=footway in such places does not make much sense to use a footway as a kind of “lesser” path.
Indeed the main path from the car park was tagged footway despite extensive bike usage, wide enough could probably drive on it and a “decent” surface at least in the dry and in the UK everywhere is remarkably dry right now. So footway made no logical sense to me there especially as it is not an official public footpath ROW.
Conversely I have also regularly seen highway=track used then mappers slap on motor_vehicle=no just because seemingly a path looks like a “track” even if no vehicles are likely to be able to get on it even if you owned the land so might be private at very best. I also saw highway=path mapped off a “footway” which when I passed I could not really see on the ground, no obvious heatmap usage and no real Lidar evidence either. To me they should not be there at all even if tag informal. That said unless you extensively “survey” an area “on the ground” and in different seasons then it would be hard to establish a hierarchy anyway. Lots of these areas don’t reveal a great deal, especially compartively, from aerial photos especially if taken in summer. You can’t of course “cheat” and look on StreetView either. Assume these “ghost” paths were added by armchair mapper who thought they could detect a “line” in some aerial picture.
