Automatically split long ways because they are hard to survey

[Zaczero]

Use case

I am sometimes hesitant to answer quests on VERY long ways. I suppose I am not the only person. It would be very convenient for such ways to be automatically split. I know there is a split feature but deciding where to split from the pedestrian perspective is difficult in such cases. Doing it automatically would be a welcome feature.

Proposed Solution

When loading a quest, the way is automatically split above some threshold. Only the nearest part is then shown to the user as part of a quest. The split way is submitted to OSM once user answers the quest.

There should be a different limit for footways and roads. The split should prefer an intersection location, if such is nearby the threshold distance.

[westnordost]

I don't see how this could be done in a meaningful way automatically. And certainly, we don't want the app to be acting on its own, splitting ways "just in case" the user is going to answer it differently for the sections presented. The changes are made in the user's name after alll.

[Zaczero]

There is no risk to splitting a way assuming it's done correctly. Initially this could be only applied to ways without relations for safety.

I don't see how this could be done in a meaningful way automatically

I already explained the idea in the first post.

The current issue is that some ways are just too long and they discourage survey. It would be better from user experience perspective if a way were cut automatically above some reasonable threshold. It's not possible to survey such long ways accurately. And there is quite much friction to using the manual split feature.

[matkoniecz]

but deciding where to split from the pedestrian perspective is difficult in such cases

Split into surveyed and not surveyed part

[Zaczero]

If you both agree in this being a non-issue, maybe it's just me.

[amandasaurus]

I have this problem often when hiking, and I would like an easier way to split long ways. I don't think it should be done automatically, but I think it should be easier & quicker than it is now.

(a) When I'm hiking in the forest, some of the OSM ways are large, and I'm also uncertain about adding a tag to the whole thing in one second. If I could split a way into ~ 100m (?) sized ways, and then add the surface many times. A downside of this is that it makes many small OSM ways, which may, or may not, be needed in the end.

(b) Sometimes I walk past a junction, and can see the first ~50m(?) of the track. and can see the surface, or smoothness. I'm not walking down that path, so I can't say for certain. I'd like to be able to split the way into the first X m (?) quicker & easier than currently.

I think (b) (“Quickily split off first X m”) is few downsides, and would be benefitial. (a) (“Quickly split into X m sized chunks”) has downsides, and I'm unsure if the outways the benefits

[mnalis]

(a) (“Quickly split into X m sized chunks”) has downsides, and I'm unsure if the outways the benefits

yes, that one is definitely too problematic IMHO, as it would both create a ton of unneeded chunks and would increase mapper load significantly.

(b) (“Quickily split off first X m”) is few downsides, and would be benefitial.

I'm all for making manual split quicker & easier, but I don't have an idea how design could look like which would accomplish that¹? Can you detail how you envision this easier way would look like in steps, @amandasaurus ?

But I personally do not do that ("mark surface of first 50m of some way I do not intend to walk") because:

• only having 50m tagged does not really help all that much
• if the rest of the way becomes worse (as it too often does in my experience, esp. for forest trails: e.g. starts as compacted for 100m and then becomes ground after the house, or starts as an asphalt and then becomes compacted track later etc), it creates even worse situation then if I didn't map anything at all (as it will actively mislead people)
• it hides it from my (and other mappers) eyes for the next time when I pass that crossing, making it less likely to get fully tagged eventually - as if I can see it is untagged yet, it is quite likely that I might modify my route to go that way instead, even if it is longer or in completely wrong direction. If it however looks tagged (because first 50m or whatever are mapped), I would likely just pass it...

So because of that I personally actually prefer to either tag the whole way correctly (splitting manually as needed to tag changes in properties), or at least significant part of it (e.g. 30-60% or whatever) for ways which are very long (e.g. dozen kilometers with several intersections).

Footnotes

1. Only idea I personally can see how this idea with "first X m" might work, would be to have user have to select the way the want to split by clicking on it, and then entering number of meters (and direction, if it is not start of the way) , which seems to me to be even more complicated than current method of just clicking on the position of the way where you want to split it - so something better than that specific method would likely be needed to actually improve the situation. ↩

[amandasaurus]

On Di, 12 Mär 2024 17:29 +01:00, Matija Nalis ***@***.***> wrote: yes, that one is definitely too problematic IMHO, as it would […] increase mapper load significantly.

My suggested “Split every Xm” would only be done manually by the mapper, never automatically. Since the mapper has chosen this, it's no longer a bad thing for “increasing their work load”

…

-- Amanda

[matkoniecz]

I admit that I have no idea how interface for doing could look like - and be noticeable improvement to current one.

Occasionally I split overly long ways when I surveyed only part and this never seemed to be a real problem in mapping to me.

Can you propose which specific steps with new interface mapper would take to split way?

[mnalis]

My suggested “Split every Xm” would only be done manually by the mapper, never automatically. Since the mapper has chosen this, it's no longer a bad thing for “increasing their work load”

My concern was if that mapper didn't end up walking 100% of that path, it would negatively affect other mappers.

E.g. if the mapper asked SC to split that path every 50m, and the way was e.g. 1.5km long, that would create 30 segments. But if they only ended up walking first e.g. 250m and then took a turn (or even didn't go on that way at all, but only "saw" first 50-100m) they'd only map 1-5 segments, leaving remaining 25-29 segments for someone else to solve, thus significantly increasing the load of those subsequent mappers.

And if they are 100% going to walk the whole way, then I do not see the need for such auto-splitting-in-equal-chunks at all? Simply walk until you see the difference, and only then split the way and tag it differently. That would create optimal (minimal) amount of splits, thus producing better result for everyone. And it would reduce load for both current and subsequent mappers. Looks like win-win to me?

---

But perhaps I am misunderstanding what exactly are you suggesting 🤷‍♂️ ; could you please give us step-by-step example showing use case where your suggested new way of splitting whould be helping the mapper(s)?