I think you are just adding to the validity of my argument in another thread ad absurdum that there is more than firmness to tracktype.
By the way now looking again to the poll, I think the questions are misleading in that they are binary. They should have been along a scale.
I tag tracktype solely according to its firmness and nothing else matters.
I tag tracktype mostly according to its firmness with some regard to some general classification like maintenance/quality/developement.
I tag tracktype equaly according to its firmness with some regard to some general classification like maintenance/quality/developement.
I tag tracktype mostly according to some general classification like maintenance/quality/developement with some regard to to its firmness.
I tag tracktype solely according to some general classification like maintenance/quality/developement.
I think you would not get that many answers under 1 or 5 (even I am 4, as firmness is part of quality). I am almost tempted to run such a poll, but if people have better wording for it, bring it on :-).
Hope by “quality” you mean smoothness/usability, because in practice tracktype has become a general classification of how good or bad a track feels to the user.
It’s already pretty clear there are many interpretations and needs, often across multiple orthogonal dimensions, and that finding a single global definition that satisfies everyone will be difficult.
A more interesting poll question might be: “Has your regional community reached and documented a local consensus on how to tag tracktype=*?”
If yes, those examples could be very useful to share, and might help shape a more global template over time.
I hope that most mappers will map tracktype and add smoothness anyhow, together with surface. smoothness alone does not contain all the information available, just a part of it and it is the part which changes more often and faster than surface and tracktype.
Unfortunately that wiki page is full of references to vehicle types I mostly don’t use:
“Will most people be happy if their routing software suggests taking this way with their rollers/racing bike/sports car/normal passenger car/SUV/4WD/tractor, or would they prefer to take an alternative way (assume there is one) with a smoother surface that is 30% longer?”
If I have to imagine whether a sports car driver or a tractor driver would prefer this way over a 30% longer way, I’m likely to stick to tagging tracktype.
Of course you are right, there are paved tracks in the strict sense of getting to the field or the forest.
In the original proposal though, grade1 is introduced as most often leading to the farmyard. So it is for the people to get to their premise or for delivery or customers to get by there. That is why I wrote, they nowadays would be service, due to their function.
Perhaps due to smoothness interpreted by some as: comfortable to ride and by others as possible to ride without severe risk of breaking the vehicle/driver? I have seen tractors drive where lots of people would not dare to walk. The sample picture for impassable is certainly passable by trial bikers.
To the routing: I used smoothness=impassable on some paths in the woods, because that is what they are, in my opinion at least. None of the three routers on the main website did care in their bicycle profile. Might have changed meanwhile?
To the rendering: OSM Carto maintainers explicitly are not fond of the tracktype tags. You might find open ears there (open ears is a saying here for welcoming treatment.)
The last column in your table prompted me to add three more:
Tracktype
Pedestrian Recommendations
Hiking Pleasure
Cycling Pleasure
grade1
Dress shoes
Avoid
Best
grade2
Ordinary shoes
Mostly Avoid
Good
grade3
Ordinary shoes
Good
Rough
grade4
Walking shoes
Sometimes better
Tough
grade5
Walking shoes
Only when dry
Push?
This post mostly thought to make this topic less exhausting. And of course, because I do not care much about vehicles except bicycle. Tracktype is about the type of track, not about what kind of vehicle can go there, isn’t that what smoothness is for?
tracktype=* already seems to bundle several aspects in practice, firmness, degree of engineering/improvement, material presence, and sometimes perceived condition or maintenance — so it functions as a fairly general classification. That’s why I’m trying to understand the boundaries a bit better:
Why should smoothness or practical usability be excluded, even as a rough indicator, if other dimensions are already implicitly included?
From an end-user perspective, what value does a general tracktype classification provide on its own if usability differences are not reflected in some way?
Also, it’s worth remembering that many of these orthogonal dimensions already have (surface=*), or could have their own dedicated tags ( potential firmness=*, engineering=*, maintenance=* tags, etc.). So excluding smoothness solely because it has its own tag doesn’t seem entirely consistent, surface does as well, yet it is still clearly part of how tracktype is interpreted in practice.
I’m reminded of a joke about someone looking for his keys that he is sure ended up under a cupboard when he dropped them, but he’s searching under a table instead. When a colleague asks why he’s searching under the table while the keys are under the cupboard, he answers: “It’s so dark under that cupboard, I can’t see a thing, it’s too difficult to search there. I’m searching under the table because at least there’s some light so I can see something…”
Tagging tracktype is easier than tagging smoothness, but if you tag a track with tracktype=grade3 because you feel that it’s kind of an average track, I still don’t know if I can drive that track with a normal car. To be able to tag smoothness, you need to study the wiki a bit and spend a bit more effort on your consideration, but if you then decide to tag smoothness=bad, then I know I that can drive it although the going will be slow.
You don’t need to be a user of those vehicles yourself to be able to tag smoothness: you can compare the surface with the pictures on the wiki (additional ones are in Key:smoothness/Gallery - OpenStreetMap Wiki ), look at what kind of vehicles actually use that way and how they use it, etc. It’s not as hard as it seems: try it!
A pedestrian might get their shoes dirty on anything unpaved. But I think that a recommendation of shoes depends more strongly on surface and sac_scale.
This may only work for the personal goals of mappers working in isolation. It would drive application developers crazy later on as different mappers requests them different things because each one’s data was generated under different principles and mean different things. And this gets worse the longer those mappers map alone according to personal criteria, instead of shared criteria that works for most.
I would suggest first arriving at a range of possible values given the vehicles you’re familiar with, then making a final choice by trying to imagine the other vehicles. Nobody’s gonna complain if you (or me) got it wrong by 1 smoothness level on an initial survey, a user of such vehicles can check it later and make any adjustments that they see fit (using the same basic idea: only within the range of possibilities they are familiar with). Comparing with wiki images also helps.
I largely agree, but I understand the difficulty of mapping tracktype, smoothness and surface to a complete description of tracks. That’s why I believe mapping surface first and then the rest (tracktype, smoothness) only when needed is an approach that would save time and help avoid many of these cases of uncertainty on the part of the mapper. What really prevents this approach from being satisfactory is the fact that the renderer only considers tracktype for tracks.
I really think you work with a model of humans that rarely exist in practice. Most people will prefer an easier way. In more general way, when designing tags, we need to thing about users/mappers. Hence the preference of a lot of mappers not to apply a lot of qualitative tags.
Why? A table like this: Poll: how do you use tracktype? - #108 by Hungerburg would lead to a mostly usable qualification. When I wrote about it being on the scale, a scale is of course not imaginary, it has some concrete content. It just needs not necessarily be single axis to be usable. Such a scale would give user a general idea what kind of road a track is going to be - as I would argue tracktype is already doing, if not perfectly.
On most tracks I have tagged, I have never seen a vehicle. Anyway I do not use vehicles, so I am just not supposed to tag smoothness, whichs is fine by me :-D. Then my pet peve that I just do not go around with measurement tape to investigate depth of holes apply. Oh, the pictures in the wiki are priceless.
How are these different:
And this and this:
Why the hell is this bad:
(it would be loud, but 50 % speed?)
Anyway, it seems to me there might be two sets of users - drivers, who care about driveability, so for them smoothness maybe makes sense. Than there are pedestrains, who mostly probably do not care. Tracktype is much more useful here.
The thing is - how often does one need to make a decision whether to drive somewhere based on a map when it is a track? Maybe it is different in other parts of the world but the vast majority of tracks I have seen lead to somewhere where you are not supposed to drive usually unless you need to access land (so probably you already have some knowledge or easy access to people who have such knowledge) and they are your final destination anyway. smoothness sounds like a great key if it has less values for actualy public through roads, but tracks are rarely those, based on their wiki definition.
Not often indeed, but occasionally the decision is important:
Just last week, driving on a regional road, we hit an end of a long queue of cars waiting for rescuers to clean up a traffic accident. After waiting for some 15 minutes, I glanced at the map and noticed what looked like a decent detour through the surrounding vineyards over an agricultural track[1]; when we spotted an SUV driving over there we decided to turn back and take that detour…
…it was a narrow but decent asphalted road indeed, so it was the right decision to make. Only if there wasn’t a large snowdrift near the end of that detour that blocked our way, so we had to turn back once again. The main road got opened in the meantime, so we effectively lost some additional time if only we decided to wait it up.
1) Truth be told, it was mapped as unclassified, so somewhat easier to spot on the map. However, vehicle=agricultural reveals it probably should be a track instead.
Well, from the changeset comment and designated=bicycles, I think its primary purpose is for cycling, not land access, so no track, possibly cycleway? And even a track further on: Way: 226859269 | OpenStreetMap, you have got grade1, so it is ok to drive :-).
That’s a touristy, wine area, so I guess it was originally designed for access to vineyards, and later on also repurposed for bike touring. But I don’t think that bicycle=designated is correct here – although I didn’t exactly go for a mapping survey there, I don’t remember seeing a sign to that effect; [motor_]vehicle=agricultural is more appropriate, and plausibly signposted somewhere. Those roads are designated as Güterweg, which can end up as track, service or unclassified.
But we digress; my point was, rather, that routing over tracks and similar low-quality roads can be important in emergency situations (not that mine was serious enough) – and indeed, rescue and emergency services in many areas are known to prefer OSM-based maps because we (try to) map those.
This way should have been split into two parts (as it is now):
This part (blue) is a service road, surface=asphalt, it’s responsible for the change from grade2 to grade1 and lit=yes (yellow) in StreetComplete.
The big part a track through the forest up the hill (in direction of the red arrow). surface=compacted, but was mapped with SC as dirt and lit was changed to lit=no, but the tracktype wasn’t changed. It’s now back to tracktype=grade2.
It’s one tracktype=grade1, surface=dirt less now. It was not easy, a 1,100-metre route with 120 metres of elevation gain.
Roughly 20% of the land area of the United States is public land that is either part of a national forest or managed by the Bureau of Land Management. Once you get away from the main tourist areas, vehicle access is generally done on logging or ranching roads – exactly the sort of thing that gets tagged as highway=track.
4 Likes
Mammi71
(One feature, Six mappers and still More ways to map it)
121
Not OTG, but I also took a look at some of these strange, unexpected combinations and came across the following cases in my samples:
Absolute mappers newbies, only 1 changeset: residential with asphalt and track_type=grade5 at the edge of a newly built and recorded commercial area. Before, there was only field and probably farm tracks. Beginner’s mistake.
Seen very often: A farm track with track_type=grade5 branches off from a larger road. However, the first 10-15 metres are still asphalted. These 15 metres were later separated and tagged with surface=asphalt. Without exception, this is a StreetComplete challenge – obviously, with SC you can change the surface – even of tracks – without checking whether the track_type is still correct. And the editor doesn’t even notice. Fatal error!
Not so common, because it may not happen that often: a former farm track is upgraded, improved, perhaps even to highway=service or higher – and someone forgets or overlooks changing the track_type. In most cases, it did not have a surface before, and in some cases the surface was updated, but without adjusting the corresponding track_type.
(open ears is a saying here for welcoming treatment.)
.jpg)
