Aligning OSM hiking sections with Wikivoyage

General talk
, ,
62

The point is, mappers have already acted as editors in the vast majority of cases, and it’s working.

I’d like to standardise around this: let’s use relations for logical stages or sections that are meaningful for hikers. Also meaning: let’s not use them for other purposes. In the examples I put in the original post:

  • The Appalachian Trail was split into relations representing the operators. These are also added as tags on the individual ways.

  • The GR 10 has large sections, for which I don’t understand the rationale, and couldn’t find clear “from” or “to” tags to add.

In these cases, it seems like the relations are not used to represent logical sections or stages of hiking trails.

If you do hold the position you’ve outlined above, and want to standardise this to an extent, you should propose removing the existing stages or sections on trails where there are no on the ground markings.

So which is it?

We could standardise: split trails into well-defined sections using relations, either only when signs exist, or more broadly, as is common practice now.

63

If you’re interested in a good example of a hiking guide created using OSM on Wikivoyage, I recently edited this: Skye Trail – Travel guide at Wikivoyage

It’s still a work in progress (and you’re welcome to contribute!) but I think it’s a good example of the general structure, and what’s currently possible.

Another example would be Lapplandsleden, for which I have also contributed to OSM (see also Kungsleden, the High Coast Trail, Nordland trekking trail, etc.): Lapplandsleden – Travel guide at Wikivoyage

Not trying to advertise, these are free guides, just wanted to showcase what’s possible through collaboration :slight_smile:

64

Not necessarily, I can also rather take the whole superroute and split it with my own choices, and let the local mappers map those giant beast the way they find more practical. As someone said earlier, it’s the mapper time that is precious.
That said I tend to think that standardisation for the sake of it is not necessary. Should a consensus arises, fine !

3 Likes
65

Thanks for the comment.

and want to standardise this to an extent

It sounds like you don’t see standardisation as necessary. I’d be interested to hear more about why. From my perspective, the argument that “mapper time is precious” doesn’t really hold up here. As mentioned several times before, mappers have, in practice, already done the work. Logical stages are widely present in the data across many trails.

So it feels like the resistance isn’t coming from the mapping activity itself, but more from a reluctance among some forum users to acknowledge what’s already common practice. If the data is there, and it’s working well for downstream uses like Wikivoyage or hiking apps, why not build some consistency around it?

66

operator describes the operator of the object the tag is put on. operator added to the route-relation equals who maintains the route, where operator added to the way describes the maintainer of the path/road.

It might be just keep the members below a certain number to make maintenance work of the relation do-able. Which should be our highest priority. The best split route is useless, if no one is maintaining it anymore.

Why do you think “your” split is more useful for OSM than any of the others? And even if undefined stages are more useful for OSM, why “your” stages are better than “mine”. I understand your aim, though the OSM-DB might be the wrong place to realize it without any post-processing.

1 Like
67

If I am reading the history right, the mapper who originally mapped the route with these sections is still very active in OSM, so maybe worth asking if you haven’t already done so.

But don’t rely on what I say, I’ve just walked a 45km route, so maybe I’m not in the best shape to parse OSM object histories. It has no stages, you see, so I had no choice but to do it in one day!

3 Likes
68

operator describes the operator of the object the tag is put on. operator added to the route-relation equals who maintains the route, where operator added to the way describes the maintainer of the path/road.

I’m not sure I follow the distinction here. If a relation groups together ways maintained by a given operator, then the relation is defined by the operator. That seems consistent with tagging.

The best split route is useless, if no one is maintaining it anymore.

Fair enough, I think it’s good to have a conversation around this. It’s not my goal to make routes difficult to maintain.

Why do you think “your” split is more useful for OSM than any of the others? And even if undefined stages are more useful for OSM, why “your” stages are better than “mine”. I understand your aim, though the OSM-DB might be the wrong place to realize it without any post-processing.

I’ve given my response to this here: Aligning OSM hiking sections with Wikivoyage - #62 by jpolvto

As well as here: Aligning OSM hiking sections with Wikivoyage - #16 by jpolvto

But to complete your argument:

Why do you think “your” split is more useful for OSM than any of the others?

You’ve partly answered that yourself:

It might be just keep the members below a certain number to make maintenance work of the relation do-able. Which should be our highest priority.

In reality, I think you’ll find this aligns with how users actually perceive trails. The purpose of sections is to make a trail easier to understand and manage mentally. Too many, and it defeats that purpose. You could divide the PCT into 200 stages, but hiking guides don’t, because that would make the trail harder to grasp, not easier.

I think this makes a strong case for aligning sections in a way that reflects how people actually use and understand trails.

It might be helpful to consider a few questions at this point:

  1. Would you agree that most hiking routes in OSM are already divided into meaningful segments for hikers (such as stages or sections)?
  2. If so, what do you think is the reason behind this common practice?
  3. Given that, could we aim for a consistent approach in how we structure and present these segments?

This could help us avoid going in circles or debating points that are already reflected in existing OSM data.

69

Thanks for the tip!

70

I can’t speak for the PCT but there are definitely Camino de Santiago guides which are divided up into village to (the next) village sections, which is a whole lot of stages if maybe not quite 200!

71

But there is only ONE way to split the route in OSM and there are thousands people out there. So which of them is the ONE? If you split the route now based on how you hiked the trail, why should any other mapper not change the split any time later based on his hike? Which of the two splits is more meaningful?

If splitting the PCT in 200 stages because of easier maintenance, then that might be the OSM split and if stage 178 get’s too long, OSM might split it in to stages at any time and PCT will contain 201 stages.

The path/road itself might be maintained by an organization and a hiking club might maintain the signage of the route. So operator=organization will be added to the way and the route-relation will have operator=hiking club.

72

Even those guides typically include an overview that isn’t broken down into all 200 sections. And for the record, I would not propose splitting a relation in that way.

To give a sense of what I think we should avoid:

  • We could split the Camino every time it crosses a river. Some participants in this thread seem to consider that a valid option, but I disagree. It doesn’t reflect how hikers actually experience the trail, based on any of the examples I’ve seen.
  • We could also split the Camino by village for the first five villages, then have a single giant section for the rest. That kind of inconsistency also doesn’t seem reasonable (see Kungsleden a month ago).
73

I assume you mean most hiking routes of substantial length. I think this depends on whether you feel that defining sections by administrative units is meaningful for hikers.

This is the section of the Spanish GR 7 (part of the European E4) in the Murcia region. It starts and ends at the regional border, at points of otherwise no significance, where it is almost impossible to start or leave the route. I would see this as not meaningful for hikers.

This is not unusual, the Camino Norte de Santiago is also like this (in contrast to the Camino Francés). So is the GR 1 Sendero Histórico.

I would guess these splits are motivated mainly by the technical difficulty of maintaining monster relations. Splitting by region is objective and may fit well with mapping communities and sometimes with trail operators if they are also regionally organised. But as I said, I find it hard to see stage endings at random border points being motivated by usefulness to users.

But I don’t know what proportion of long routes worldwide follow this pattern.

I’d also mention that hiking routes in Ireland are generally not split at all. It’s true that only a few routes in Ireland are over 200km. But the Kerry Way, for example, takes 9 or 10 days to walk. Nobody seems to have tried to split this kind of trail into daily stages or any other kind of section.

3 Likes
74

I would think there is no such common practice, as it depends on the trail. On a rather remote trail like the PCT I would expect a split at the few supply locations, like where I can start/stop the hike. But using the same practice on trails going through populated areas like Lutherweg you would end up splitting the route in each village.

So splitting a route by that measure can’t be consistent and “meaningful”/useful.

In OSM splitting at admin boundaries is quite common. You could go by distance, like each stage should be 50 miles. Both would be consistent, but that might be not that useful for hikers, as usually you can’t start/exit the trail exactly at the boundary or after 50 miles.

EDIT: Another, documented method is the quantity of members, with same results as mentioned above.

1 Like
75

Thanks for the response! Hope you enjoyed the hike today!

But I don’t know what proportion of long routes worldwide follow this pattern.

I’m not sure I would consider this an anti-pattern. At least it’s consistent, and state boundaries can be quite useful points of orientation. I own hiking guides that split trails along state boundaries.

One case I’ve come across is when smaller trails that are part of E-routes are not actually included in the E-route relation. Instead, the same path is traced twice, once for the E-route and once for the local trail. Some might argue that this makes maintenance easier by having fewer components. But wouldn’t it be better if the smaller trail were included in the relation from the beginning?

I’d also mention that hiking routes in Ireland are generally not split at all. It’s true that only a few routes in Ireland are over 200km. But the Kerry Way, for example, takes 9 or 10 days to walk. Nobody seems to have tried to split this kind of trail into daily stages or any other kind of section.

Fair enough. The West Highland Way is also an example of a (famous) hiking trail that is not split into sections. There are a few more exceptions to find.

76

Some examples from the Nordics:

  • Arctic Trail
  • Kungsleden
  • Massiv Trail
  • Bergslagsleden
  • Roslagsleden

Some more examples:

  • Pacific Crest Trail
  • Pennine Way
  • Cape Wrath Trail
  • Continental Divide Trail
  • Now also the Appalachian Trail!
  • GR 11
  • Haute Randonnée Pyrénéenne
  • All of the Via Alpina routes (these need some work)
  • Te Araroa
  • French Way

Maybe I can make a graph using overpass.

In OSM splitting at admin boundaries is quite common. You could go by distance, like each stage should be 50 miles. Both would be consistent, but that might be not that useful for hikers, as usually you can’t start/exit the trail exactly at the boundary or after 50 miles.

Can you give some examples for hiking trails?

77

I don’t get what you want to tell me by listing the samples.

78

It supports the claim that dividing trails into meaningful sections, for both hikers and guidebook editors, is a common practice. I’d like you to specifically support the claim that splitting trails into 50-mile sections is a widespread convention.

79

It might be just keep the members below a certain number to make maintenance work of the relation do-able. Which should be our highest priority. The best split route is useless, if no one is maintaining it anymore.

It seems contradictory that when the idea of keeping member counts low is brought up, as a way to make maintenance more manageable, it’s framed as a top priority.

But then, when I point out that this same principle supports the idea of splitting trails into meaningful sections, suddenly the opinion flips:

If splitting the PCT in 200 stages because of easier maintenance, then that might be the OSM split and if stage 178 get’s too long, OSM might split it in to stages at any time and PCT will contain 201 stages.

This kind of reversal suggests that the response is less about the actual reasoning and more about disagreement for disagreement’s sake. That’s a bad-faith way to engage, and it derails productive discussion.

80

I’m stepping away from this topic for now. Thank you to everyone who engaged constructively in the discussion. I really appreciate those contributions. I believe a consensus has been reached among active mappers, even if that’s not necessarily reflected here on the OpenStreetMap Forums.

81

In generaI, I think so, yes. Provided of course that the entire small relation is part of the longer route. I have never really liked the fact that the E8 in Ireland duplicates the Wicklow Way and several other routes, instead of including them. But both the E8 and many of the local routes were mapped 12-13 years ago, maybe there were good reasons at the time, or maybe the idea of route relations containing other route relations was not well known.

I’m not sure if this is really relevant to the main topic, I guess your point is that the contained routes could be sections of the longer route? That could lead to a few slightly odd “sections” like the bit north of Bunclody here, where the Wicklow Way and the South Leinster way don’t quite join up. But that might be OK as a through hiker really would see this section as different due to the absence of waymarks.

1000051808
10000518081080×1340 206 KB