That’s how everything started before we had relations and it was not quite useful if you want to add more information than just “that way is part of a local hiking route”, eg. which route. On top it’s hard to maintain and it’s easily forgotten on some small segments.
Maybe there are some renderer which are interested in that, though usually the hiker wants to follow the hiking route he is on. So definitely you need the route relations. Deriving that information from the relations to the ways is an easy task for software and hard for us mappers to keep up to date.
It’s not entirely unheard of in OSM for people to add a tag/value to an OSM object that could also be derived by looking at relation memberships (something that’s doable from, for instance, the flex backend for osm2pgsql but not the pgsql one). Like anything in OSM, there has been a bit of pushback against it because “the same information is stored twice” **.
However, if I came across it without reading this topic first I’m not sure I’d understand what trail=yes meant. The main reason is that I’m English, and so I don’t know what trail means here. I know in American English it can mean something akin to “a hiking route” (Pacific Crest Trail, Appalacian Trail) but that it’s also used for much smaller things that may or may not start at a “trailhead”, and that it has been suggested in the past as a “type of OSM way” (i.e. not route-related at all).
As someone already mentioned, the challenge with “relation tags on the way” is “which relation?”. Here for example there are 3 long distance hiking routes, 1 or 2 regular cycle routes and one horse/mtb/hiking route. That sort of thing seems rarer in the USA, so I guess your suggestion might work there, but it doesn’t sound like a global answer to me.
Unfortunately, as has already been mentioned, that tag is mostly used for something else. With an “OSM data consumer hat” on, having different things in OSM mean the same thing in the real world isn’t a problem; you can easily coalesce into one map feature (and of course any map will do that anyway, even with things that don’t mean the same in OSM).
However, have one OSM tag meaning different things in the real work is a real problem as you have to look at other things for context - other tags, age of feature, geographical location, etc.
** but then OSM has never pretended to be a normalised database.
American “trails” are like British “paths” – they can mean a lot of things, but when you say it in certain contexts, people have a certain mental image. This mental image is pretty strong in the context of a way:
In the context of a route, “trail” can technically apply to a recreational route of any kind – hiking trail, bike trail, bridle trail, or perhaps even auto trail. It connotes a rustic experience as opposed to an efficient means of transportation. Without context, I think most people would associate “trail” with a way rather than a relation. But that’s how we end up with so many route relations that are analogous to making a route relation for a street by a certain name.
Hiking/biking route concurrencies happen fairly commonly in some parts of the U.S. For example, at a junction in Milford, Ohio, this way is part of two hiking trails and six bike trails, and an adjoining way is part of two other hiking trails and six other bike trails – some of which are analogous to a route relation for a street by a certain name. This junction is right where some of the trails depart from a rail trail onto busy highways or city streets.
How would you deal with the Pacific Crest Trail? Last I checked there is a segment into Canada so not only does it cross US state borders is crosses an international border. If we treat trails crossing state borders in the US like national borders in Europe how do we treat trails that cross national borders?
Here we have a couple of routes that go from France to the Netherlands, crossing Belgium that has two hiking federations: one for Wallonia, one for Flanders. Currently that makes one additional level of subrelations
I guess I wasn’t clear. I think we should tag things the same everywhere.
My comment was directed to statement “Like a national route in the US will be equally to an international route in Europe.”
I interpreted that to mean interstate trail routes in the US would or should be tagged the same as international trail routes in Europe. That sounded off to me. And if the US tagged interstate trail routes the same as others were tagging international trail routes, how would we tag an international trail route which partly lies in the US?
To make this a bit more concrete, here are the category definitions for a network:type (or if you prefer, route:scope tag), I would find useful and that might work globally:
base_network : a local route, waymarked or otherwise explicitly signed specifically to create a network of accessibility for a non-car mode of transport (foot, bike, horse, …). Purpose is not necessarily touristic. This is essentially equivalent to the road network we describe with the main highway tags. Routes are usually not named and a single route usually isn’t considered a useful trip when followed from start to end.
node_network: same as base_network except that each route starts and ends at an intersection clearly named or numbered.
local: touristic route accessing one or more local features. Designed so it can be followed start to end. Day trip or less.
regional: touristic route connecting multiple prominent features in a touristic region. Designed so it can be followed start to end. Should be a multi-day trip.
interregional: touristic route crossing or connecting multiple touristic regions or crossing entire countries. Designed so it can be followed start to end. Full trip is 3 or more days.
long-distance: touristic route covering large parts of a continent, usually crossing multiple countries (exceptions are North America, Australia and maybe Russia). Designed so it can be followed start to end. Full trip is at least a week.
Some additional observations:
Whether or not a route is designed to be followed from start to end or if it constitutes a base network is a bit of a subjective decision because many of the base networks for hiking have historically developed out of a collection of local touristic routes. I’d defer to local knowledge there. In Europe a good marker is if the route has an explicit name, not just a ref or a symbol. The US trail system can also be both. I’ve seen some trails that are clearly meant to be followed as a single trip. And I’ve seen trails that made up an entire network to give access to a hiking region so that you would switch from one to another on your trip.
Saying that the base_network is the same as the highway classification somewhat raises the question if it wouldn’t need indeed similar subcategories as primary/secondary/tertiary. So it might be orthogonal with [inrl]wn after all.
In areas where a good base network exists, the local/regional/interregional/global routes usually just follow routes from the base network. The result is that you usually follow the waymarking of the base network and find the symbols and names of the routes only on major guideposts (see also this thread). Ideally, we’d map that by making these routes superroutes containing only references to relations from the local network but practically this is not going to happen because data users can’t process that.
The definitions explicitly use “touristic region” instead of administrative region as these are often the unit of planning. In countries with a more centralized approach to planning (like Switzerland, Czech Republic, Netherlands and partially the US), administrative region might be a better fit.
In Europe, we also have the pilgrims routes, which are a bit of an odd-one-out. It’s not really a route but a network with many starts and one end. Can possibly be covered with a global scope and a special network=Camino de Santiago tag. Or might go into its own scope category.
+1, looks good to me. Just one exception regarding the “from start to end”: there are circular routes (in the case of longer routes, they are often divided in stages, i.e. parts of the route that can be done in one day), which typically do not prescribe a start and end, you can start at any stage start/end and if you do all of it, in the end you will be back to the start.
Ok, another one: for “local” saying “accessing one ore more features” could also find an exception, “der Weg ist das Ziel”, in case there is not specific feature but the general landscape attractiveness is the reason for the route (e.g. leading from a parking in a circle)
Do we actually need a new tag, or could we simply slightly redefine the iwn and nwn definitions?
These definitions are fascinating to me, because I’ve hiked in Switzerland and I’ve hiked along the “base_network” case. There really isn’t anything like it in the US that I can think of1, and I suspect that most Americans that haven’t been to Europe will be confused at it. But, that’s OK.
Looking at the other categories, I’d like to offer American translations.
These last two categories, however, throw me for a loop. They’re the same as the “regional” category above with the only difference being their length. I would categorize North South Trail as regional, from your description, and I suppose the Appalachian Trail, Pacific Crest Trail, or Continental Divide Trail (which are the big three long distance trails in the US) are long-distance.
It seems that once we get to regional and beyond (where it’s overnight-hike or longer distance), the only actual discriminator is length.
I don’t think we should be encoding a value as a proxy for length if data consumers can just measure the length of the route and from that decide what to do with it.
Are there other discriminators beyond how long these are?
1 It’s always a risk to make sweeping generalizations about the US, because somebody will surely come up with an exception.
regional would still be within a hiking area or nature preserve, just a multi-day hike as opposed to a simple day-tour trail that, for example, gives access to a mountain peak. interregional would cross multiple hiking areas/nature reserves.
Forgive the frequent references to Wikidata that follow. That project has also had its ontological struggles with routes and thoroughfares, so a comparison may add something to the discussion.
This sketch has me thinking that network:type=basic_network is basically analogous to Wikidata’s road network (Q358078) (“system of roads in a given area”) or road type (Q7860962) (“classification for any type of road”). This contrasts with the primary, non-recreational usage of network=*, which corresponds to highway system (Q25631158) (“network of highways, usually signed with highway markers of similar design, built to certain mandated specifications, and maintained and/or owned by a single authority”).
It’s two different things, but the terms “network” and “system” are often used interchangeably in practice, especially for something that can be as informal as a trail. No wonder trail system (Q67179750) (“network of trails”) gets used for all of the above, and so do those three-letter acronym network=* values.
Since a given route can be part of both a road trail network and a highway trail system simultaneously, I see the value in distinguishing these two concepts with distinct keys. However, the new values you describe sound like they’d be a third concept. A long-distance trail (Q13405588) is neither a trail network nor a trail system. It is a trail that theoretically can be part of both a trail network and a trail system. So this sketch would make a single key responsible for answering the questions, “What kind is it?” and “What is it a part of?” at the same time.
It sure sounds like the new values should be route:scope=*. This avoids any confusion with the existing values while potentially clarifying what the three-letter acronyms have meant for some mappers. It would also potentially allow for some additional values like alternative and excursion that are currently only available as relation roles, which are unsuitable in systems that fashion such variants as distinct routes.
If we do this, I don’t have as much problem with a term like “long-distance”, because that’s kind of a fuzzy notion in plain language too. It’s a route for long-distance hiking, cycling, or whatnot.
If base_network is just “all the trails in one area”, I suppose that’s distinct. Though I wonder how that differs from simple blazed trails in a hiking area – I have no idea how to differentiate a “yellow blazed trail” - is that the base_network or is that a local route? Remember that trail marking in the US is fundamentally ad hoc.
I still don’t understand the difference you’re drawing from regional to long-distance. It’s just differences in distances.
I can do a multiple day hike and never leave Yosemite. I can also do a several hours day hike that visits several different reserves in my part of the country. You can’t really compare the American west to the east coast - the scale is wildly different.
So, I ask again - is the distinguishing factor merely length or is “crossing different reserves” the thing we’re mapping?
As a European who has only hiked in a few European countries, not including Switzerland, I’m also not really clear about the base network concept. That doesn’t mean it’s a bad idea, and if some countries just don’t really have this concept that’s no problem, there is no need to use it. But it probably needs a bit more explanation.
Looking at the Aldo Poldo Wilderness Trail example, this seems to consist of a large number of routes, each with its own reference, and in many cases its own name. I’m not sure what makes these base network routes rather than local routes.
This is what I was interpreting as the base_network. Or maybe this is the node_network, I’m not really sure. But there’s this network of signposts that show the time to other locations along hiking trails. Here’s a picture I took hiking in the Ticino canton of Switzerland:
It might be not that easy to find a base_network in the US for hiking. Though for cycling it would be the signposted “network” of your local community/city/… without being a dedicated route. For example: https://www.openstreetmap.org/relation/13322711
Maybe all hiking trails in a national park could be a base_network, if the national park is not covered with named routes.
