Trails: Use ‘Name’ and/or ‘Hiking Route?’

Communities United States
, ,
1

This is a continuation of a topic on Slack. We are moving here so that the conversation will be more permanent.

Slack discussion link: Slack
Also, this discussion is essentially a continuation of a discussion from 2021: Slack

Based on my sampling, trails in Connecticut had names without controversy until 2020-22 (mostly 2020) when a large-scale project was undertaken to remove nearly all trail names and instead add hiking route relations if they weren’t present before. And the project continues today with extensive maintenance, continually reverting people’s contributions anytime anyone feels the trail name might be an important thing to have and tries to add it (with the hiking route being retained). It seems like this project has been done based on ‘unwritten rules,’ contrary to certain norms in English-speaking regions of the map and guidance on the OSM Wiki. If I am missing something and these aren’t actually ‘unwritten,’ it would be great if this guidance could be shared. But I looked around Canada, Ireland, the UK, the US, Australia, and New Zealand and it was not hard to find named trails anywhere, except a slight idiosyncrasy in New Zealand where ‘track’ seems to be the standard term for a trail. When I did encounter no-name paths with hiking routes, it was typically reserved for more informal routes. Examples: “Tararua - northern crossing” (a ridge route in New Zealand) and “Mourne Wall Challenge” in Ireland. As an aside, the later route seems to have only been added after extensive discussion: https://www.openstreetmap.org/way/60716708/history

To me, the name field is the most important text tag. It’s the one text tag that is almost universally displayed, including on OSM. Not displayed on every feature, but almost always on some features. I am genuinely interested in finding out if there are any other features besides Connecticut trails where there is a rule against adding the common name as the name. Unlike the name, hiking route relations seem niche specific, and mostly seem helpful where there are gaps in the trail. And if there are gaps in the trail, then the hiking route relation will inherently not duplicate the name the entire way.

Many have stated that there shouldn’t be a problem having the same name and hiking route text. Concern has been voiced that having the same trail name and hiking route relation may cause some renderers to display the same text twice. To me, this seems to be a problem with the renderers and it’s not really that big of a problem. There are technical solutions to this such as: displaying only one or the other; having a check against displaying the same text twice; having a label that clearly indicates which is the name and which is the hiking route. I feel sad for the Connecticut mappers in the past few years who may have become frustrated that their contributions were being removed when they were only following the OSM guidelines. Even if the text is the same, are they really duplicates if the tags / relations are different?

2 Likes
2

We have had this debate in my country (France) but nowhere near the same level of controversy. Here, the consensus appears to be:

  • that instead of trails (“sentiers”) which is a polysemic word, we focus on routes vs ways.

  • that ways are named only when they have their own name, independent from that of routes (e.g. the route “Chemin de Compostelle” goes through the way “Chemin Neuf”)

  • that giving names to ways to recall that they belong to a route borders on creating descriptive names.

We find this more practical to manage routes, because we often have more than one route on the same way, and because routes change over time.

From my work on international routes I have the feeling that this works more or less the same in other European countries. Maybe it is different on other continents where hiking routes do not reuse pre-existent ways?

2 Likes
3

“Way” is even polysemic. There is a dictionary definition: “a road, track, path, or street for traveling along”. And then there is the OSM version e.g. highway=path. I say this because even a 0.25 km long path/way might be broken up into multiple highway=path segments to capture different physical attributes like surface roughness, a bridge segment, etc. A better tagging schema model would separate the attributes that describe the physical attributes from those that describe the political or “human-centered” attributes like name, access, operator, etc. I presume this is the spirit of using route relations. We’re grouping/demarcating all the smaller “physical” segments into a relation and applying a single instance of name or operator, etc. This allows us to get closer to the ideal of “one feature, one element”.

For various reasons of mapper adoption and data user parsing of route relations (including OSM-carto), trail relations are used relatively infrequently in the US, especially for shorter paths/trails. Name is applied to all the individual highway=path regardless of “trail” length and regardless of if the highway=path is part of a route relation. I summarize the regional common practice as:

For highway=path/footway where it is signed/known as a single name, the name gets applied to all the highway=path/footway segments where applicable (and the route relation if there is one) as in here

When a highway=path/footway is known by multiple names, the lower level way’s name is what’s applied as in this path that is part of the Appalachian Trail.

There is almost no such thing as a “trail” having a name but the underlying “path” (way) not having a name (for ways highway=path/footway). It’s basically how name is used for “roads”. We’re not making route relations for every street.

The crux of the issue stated by the OP is a mapper is not comporting with these mapping norms, with explicit requests made in various online forums and changeset comments, and with mapping guidelines laid out in the wiki and the OSM US Trails Access Project.

I hope as a community we can full-throatedly say that it’s OK for name to be applied to individual highway=path segments regardless if it’s part of a route relation. And it’s NOT ok for valid names to be proactively removed.

3 Likes
4

Maybe I’m confused what you mean by “hiking route”, but there are 100s of them in the UK and Ireland. I do “gap checks” on the major regional, national and international ones here and here, and I had to split the checks up into two because there would have been more than 1000 lines in the file.

Here are a couple - you can see that the Pennine Way and the Kirklees Way are both shown, that they actually share some segments, and that some segments actually have their own “road name” too.

What you might be suggesting (and apologies if I have got this wrong) is “copying the name of the route relation to every piece of road that makes up that relation”. That doesn’t work very well, as in the example above some pieces of road name have their own (road) name, and some belong to multiple hiking routes. There may be exceptions (this is arguably one) but most signage of routes makes it clear that what is being signed is a route, not just a local street name. As well as being true in the UK and the rest of Europe, it’s also true in Australia and my limited experience of the USA, such as the PCT.

5

Sorry for barging in this local debate, I had not noticed the Communities/US tag. Just keep in mind that the practices in Europe tend to be different from what you describe. Your local consensus seems closer to the current practice for ski pistes and ski routes here.

6

I think what @ConradWard is saying is that there shouldn’t be ways without names, not that every member way of a route relation needs to carry the the relation name. In the case of multiple routes covering the same way, I think @TomPar is correct that the convention in the US has been to have the name of the “more local” trail on the way.

2 Likes
7

We’re not saying that. We’re strictly speaking about highway=footway/path and maybe highway=track.

When a highway=path has two names, and there is a hierarchy, the “lower level” way name is applied. If both names are of equal prominence, then yes there is a messy issue about what name to apply. In this instance, I leave the name off (of the highway=path).

1 Like
8

To level set, yes we’re talking about big trails like the AT and PCT but even more so we’re talking about the hundreds of small conservation properties like this. name is proactively being removed from these types of paths.

9

Note that at least in some places path may have named hiking route, without being named in any way. Including hiking route name not applying to path.

This is common in Poland, as I understand it is less typical in USA. But I would avoid assuming it as always applying across entire USA.

10

I agree with this, and would add that I think it is OK if someone makes a trail/path where the name is only on the relation. Maybe we can work towards a consensus at some point to where @TomPar’s “human centric” stuff is all on relations in the future.

11

Thank you for raising this issue @ConradWard. I am sorry to hear about your work being reverted and I support the names you’ve added being restored. I hope we can all agree that systematically removing other people’s contributions without at least reaching out them is not in the collaborative spirit of OSM.

The problem here (as I see it) is that we have some trail map renderers that only display trails mapped as route relations and others that only display trails mapped as ways. This inconsistent mapper feedback results in some mappers adding route names to ways in places where it isn’t appropriate, and other mappers adding route relations for every single named path including many places where that is not appropriate. So we have duplicate data. Personally my practice is to simply live with it and allow that trails can be mapped either just as ways, just as route relations, or as both. Maybe someday all trail map renderers will handle way mapped and route mapped trails equally well, but that is not the current situation.

3 Likes
12

I think StC had a great point early on that it does make a difference whether or not the trails use pre-existing ways. I would also say that a given name can evolve over time. If the Appalachian Trail was brand new when OSM started, the section heading up the Green Mountains of Vermont might have just been named the Long Trail. But over time, the Appalachian Trail gained prestige and prudently the OSM community decided to use both names.

I’m going to take a stab at articulating the differences we have been talking about between routes and trails. This is going to be United States-centric and may not apply to other places. I see a route relation as being more suited for when a combination of these circumstances occur:

-The route has been ‘designated’ by a niche community where the name is only meaningful to one or a few related communities.

  • The ‘designation’ was accompanied by little to no physical change or construction of the route.
    -The route was not purposely built specifically for the community, is not predominantly used by the community, and does not have signs indicating the name.

Here are some examples where I think a route name would be most prudent without a standard name tag:

  • A road designated as a bicycle route by a cycling club. (i.e. American Discovery Trail).

  • A hiking route designated by a hiking community, often goal-oriented or destination-oriented, usually over other named trails/ways (i.e. Presidential Traverse in New Hampshire, 100-Mile Wilderness in Maine).

-A mountain bike route designated by the mountain biking community, but not in use outside the community (i.e. visit the website Trailforks). These could evolve over time from routes to names, if it enters into common usage, especially likely if the trails would otherwise not be named.

  • A group of roads that have been designated by a state as a scenic byway or scenic route.

  • A ‘history route’ created by a historical society or community (i.e. Black Heritage Trail in Boston, MA).

  • A route created by a local chamber of commerce designed to bring attention to local business and/or tourist attractions.

On the other hand, a trail is defined as a physical place and the vast majority of places called a trail are a trail in a literal sense. If a landowner decides to name a trail, puts up a sign with the trail name, makes a trail map with the trail name, and creates a guide with the trail name, that seems to exceed the threshold required for adding the name to the name field. It is not much different from naming a road. With this status, the name becomes universal regardless of what activity you are doing. Even if you are just driving over the trail when it crosses a road, that trail has a name. And if you aren’t even there, but are just browsing OSM, that trail has a name. I think things may be a little different in the U.S. compared to Europe in that I would venture a guess that the median named-trail length is less than three miles / 5 km and any road sections often represent an undesirable gap in the trail, not a contributing part of the route.

The above thoughts are in reaction to very different mapping realities in Connecticut, but I feel like in most places there is much more harmony on what constitutes a name, with variations for different locales / languages. I will say I’m probably more liberal than average when it comes to adding names, but I’ve been toning it down for this conversation. Of course there will always be gray areas and exceptions to the rules. And that’s the beauty of it. Maybe the chaos is a feature, not a bug. I might be making it all too complicated. When we are mapping, I wonder if we should picture ourselves at the location and imagine someone asking us “what is the name of this place?” If you reply “this is called XYZ,” then this place by default should get that name in OSM unless there is an exception to the rule.

1 Like
13

based on the MassAudubon example given above, I would add a fourth criterion: that there is indeed a route and not just a simple path.

I fully perceive that this criterion is unclear but it tries to capture the fact that using type=route on a relation just to remove names from a few ways that constitute a single path can be perceived as artificial and misleading.

14

A big part of the problem is that many people’s concept of OSM doesn’t extend beyond what appears on osm-carto, and osm-carto is both (a) car-centric and (b) unwilling to parse route relations. There are numerous hiking-focused sites that use osm-carto despite it being unsuitable for the task.

Route relation names are the standard way of tagging any non-negligible hiking (or cycling) route, as @StC says. But I can understand people’s keenness to keep the name tags given that the route would otherwise disappear from the map they use.

Two suggestions I’d make.

One is to firm up the boundary around “non-negligible”. It seems inappropriate for ways to have the Appalachian Trail name. Someone walking their dog along a section of path that happens to be part of the Appalachian Trail won’t say “I’m just about to take Fido for a short hike on the Appalachian Trail”. Similarly you wouldn’t expect to rename a county road with name=Appalachian Trail just because the trail follows it for a short while.

So you could potentially say something like “for trails extending beyond a single municipality, or n miles, the name should be on the route relation alone”.

Second, in circumstances where the name is duplicated between relation and way, please do make an effort to keep the name identical across the two. This allows smart renderers/routers to deduplicate them (and to forestall @Minh_Nguyen asking, yes I do do this a little bit for cycle.travel :smiley: ).

Hopefully the advent of vector tiles on osm.org, even if they don’t fix this particular problem, will mean that osm-carto has less influence on perceptions of OSM going forward.

3 Likes
15

People would absolutely say this. I don’t understand this comment.

4 Likes
16

This situation certainly does exist and I agree that in this case the county road is definitely not named Appalachian Trail. However, much of the Appalachian trail is a path that was created to carry the route. This path didn’t exist before and has no other name besides Appalachian Trail. It doesn’t make sense to say that this is a nameless path that was created for the Appalachian Trail to use. The path name is just Appalachian Trail. There also are some sections that do follow older paths that already had other names at the time the AT was created. In these cases the path name remains the older name.

5 Likes
17

Probably not the kind of path and the kind of people (or maybe the kind of dog) Richard has in mind. My in-laws have a “Chemin de Compostelle” that passes in their town just behind their house, and I’m sure that the neighbours don’t walk their chihuahua on the Camino but rather on the “chemin creux” or the “Ancienne route de Paris”.

1 Like
18

This goes even deeper. Sorry if I err a little off-topic here, but the very idea that OSM might be a relevant source for hiking routes is still not widespread.

In that context, using Carto might actually be exactly fit for the purpose of its users (“get the paths, not those OSM routes we are not interested in”) and, coming back to our topic, pushing names to ways can be seen as tagging for the renderer :slight_smile:

1 Like
19

Got it. In that case then I absolutely do see your point.

1 Like
20

Thanks for raising this issue. I view named route relations as a bonus/ancillary to the way tags. It is nice to add route relations but we don’t force data consumers to use them. There have been many practical examples in the thread for when you might add a trail name and a route name. I can also think of ones where a larger national trail like the East Coast Greenway or 9/11 Trail run along a trail with a different local name.

Regarding Connecticut. Wearing the DWG hat, I’ve invited @Mashin to join the conversation on a recent changeset where a number of other mappers have left comments critical to the practice of removing trail names after adding relations. In that comment I’ve cited a few of the most popular hiking trails in the USA (based on AllTrails).

  1. Queen’s Loop @ Bryce Canyon NP. The most popular hiking trail in the United States.
  2. Angels Landing Trail @ Zion NP. Second most popular in the US. Includes a similar named route.
  3. Appalachian Trail @ Vermont The AT hopefully needs no introduction. Looking at the VT relations most are named.

AllTrails is a partner on the OSM:US Trail Access Project and is arguable one of the most seen and sought after OSM trail data viewers in the US. They show OSM trail names on their Mapbox-driven map. The National Park Service, also a partner and user of OSM data. NPS shows trail names as they are named in the ways, not the relations. Relation support in mainstream uses seems sporadic and niche to me. Relations are often split up arbitrarily by state or region, given non-names like “Apalachin Trail MD/DC”, and are challenging to work with. I believe we should support all the use cases with the data by not pigeonholing tagging into the most normalized/tidy design.

5 Likes