Revision of Smoothness Gallery

Um, didn’t your interpretation feel wrong? Even the second value (good) looks wrong to me then? I’m not a skates/skateboard user either, but I do believe that they would be able to pass through a section of smoothness=good asphalt, although with reduced comfort and speed.

Do you disagree (i.e. do you think they’d instead be forced to turn back and return home as it is “completely unpassable” to them)?

Also, would you call car whose engine shuts down every 2 minutes and you have to restart it “usable”? After all, you’d be able to reach you destination eventually. What about if maximum speed you could reach on some terrain in a car was 15 km/h with heavy turbulences? (spoiler: I wouldn’t!)

(…and other examples of “no need to slow down in a car” and similar) As others have mentioned, and has been pointed out several times before in previous discussions, smoothness tag is not car-centric, nor had it ever been, nor should it be. It would be appreciated if the attempts to make the smoothness tag usable only for cars (and nobody else) would not be constantly resurrected (or does it only seem that way).

Yes, for average car user there is no usable difference between e.g. excellent and good (referencing previous discussion), I get it. But for skates there is a difference, which is why excellent exists as a value. The granularity should be there for all users, and not skewed for cars only.

So the high end of smoothness scale is skate-oriented, high-middle range is bicycle-oriented, low-middle range is car-oriented, and low end is more like ATV-oriented. The whole smoothness tag is not about cars only, and especially not only about some specific class of the urban car. (e.g. while not owning either, I can imagine that the two roads what might feel like the same road smoothness to Landrover driver are likely not feeling the same for Tesla Roadster driver)

So while for e.g. your particular car that bad might feel the same as intermediate for that example, for many other users (e.g. bicycles - including two of mine) if definitely does not. Thus it is inappropriate to use the same smoothness value for both surfaces, as I would be using one road, but not the other. If they are tagged the same, I cannot tell such crucial difference. (Just as it would be inappropriate if I - say - suggested that any smoothness=horrible or very_horrible pictures should be in same category as smoothness=impassable, because they behave the same for my trekking bicycle).

Those StreetComplete pictures seems to match StreetComplete descriptions (and vehicle icons) rather well in my experience, at least in English (and Croatian). Are you ignoring those StreetComplete text descriptions too, and going with some other (wiki?) definitions instead? Or are in your StreetComplete language the texts in heavy mismatch with the pictures, so you only ignore the pictures but accept the texts?

Do note that if you use StreetComplete quest to tag smoothness, data consumers can at best infer that you followed StreetComplete definitions of the tag. (it is of course unfortunate if different editors might tag different things with same tag, but that is additional problem to the one primarily being discussed here. I do know that enormous amount of effort has gone in StreetComplete quest getting to the point where it follows the wiki but is simple and usable, and am appalled that it seems to have been for nothing – if even people who participated as much you are ignoring major part [or all?] of the result. It makes me even less hopefull that this thread would fare better).

Not. While I have just been mentioned and have only started on this specific reincarnation of the smoothness thread, I’ll try to give the thread at least some more attention in the coming weeks if I can muster the time away from more productive tasks, but honestly, with its umpteen reincarnation of the mostly the same subject; it is becoming a time sink with ever increasing cost/benefit ratio. (I mean, I barely properly addressed first third of the first post of the thread with this behemoth. I’m not even sure if the effort of going through all of that would be actually end up being beneficial, even if we discount the amount of effort!)


Perhaps we should just slap {{Questioned}} template on that wiki and document what the disagreements actually are; that seems much easier and less controversial task and should inform the data consumers about (seemingly unavoidable) different interpretations of tag values.

I mean, even we all happen to come to some consensus here on the forum what the tag should mean (which seems quite unlikely to me nowadays, unless one “wins” by exhausting the others, not to mention that I am much more of descriptive then prescriptive type when it comes to OSM wiki tag documentation) there are so many mappers using the smoothness tag that changing the wiki definitions at this date won’t have much useful effect, if any. The dozen or so people participating here is sooo much smaller then the number of mappers using the tag. Regardless of what conclusion is reached here, I guess that most people will likely keep tagging what they always did, nobody is going to resurvey all the tags made with old definitions, and further wiki changes would likely just further increase the volatility of the tag definition (and thus make it even less usable for data consumers).

2 Likes

I’ve looked at the current

https://wiki.openstreetmap.org/wiki/Key:smoothness/Gallery

side by side with

https://wiki.openstreetmap.org/wiki/User:Rhhsmits/SmoothnessGallery#Paved_and_compacted_ways

and I have to say that I find the allocation of the pictures to the categories better in the current gallery in every instance where they differ.

The exception is File:Very bad sett.jpg - OpenStreetMap Wiki which looks a bit too similar to bad sett. But we don’t have a picture of sett in even worse condition. Maybe this one, though? File:Sett-5-VeryBad-B.jpg - OpenStreetMap Wiki
(from Surface-smoothness fényképes táblázat - OpenStreetMap Wiki )

2 Likes

for sett I like more in the new version
for unhewn_cobblestone I like more the old version

In the current table in my opinion this is should not be the same category:

I feel like we mostly agree on the ranking of the images and debate about moving individual images left or right, but I suspect that this is an inherent uncertainty of the current smoothness tagging approach.

Would you say File:Sett-5-VeryBad-B.jpg - OpenStreetMap Wiki works better for very_bad sett?

1 Like

Sure, we’re mapping for the end users, isn’t it? However I don’t think the StreetComplete developers forum on Github is the way to reach them, and we don’t have a marketing&sales department that can do market research either. I think the best we can do is to ask ourselves (after all we are also end users of our own data) and the best place to do that is this forum, I think.

1 Like

You mean that SC and the image galleries (only Smoothness Gallery, or also the images in the Values table on the main wiki page?) show images in for instance the smoothness=bad category that according to you should be judges as smoothness=intermediate?

Note that the text in the Description column of the Values table is largely mine and is open for discussion.

former statement was not related to your table.

yes, what I meant was example the paving_stone line in the current gallery:

I do not clearly see “bad” in these images

I don’t see anything that would stop me from using it with a road bike (of course with reduced speed)

1 Like

I tried to change as little as possible in the Smoothness Gallery, and did not consider whether some images in the smoothness=intermediate, good and excellent categories are appropriate.
The picture of sett you are discussion is at the good end of intermediate and I’m not against moving it to good (it looks almost as good as Surface paving stones.jpg). But I’m against moving your unhewn_cobblestone example to bad because it doesn’t show any damage to the surface: it looks like new. I admit that it’s at the bad end of intermediate and it is very uncomfortable to cycle on it unless you have a bike with shock absorbers, but it is possible to use it “without significant risk of damage or injury” by a city bike if necessary.

Agree, esp the left side of the bottom one. Note that concrete surfaces always have joints even when new, so these should not be considered as damage.

Disagree, it fits right into the intermediate category: severe damage, but no potholes, so can be used safely by city bike (though not comfortable).

Disagree: the gaps are much narrower than usual for paving_stones and the stones are all at an equal level, so you would not notice vibration when using this surface (except maybe when using roller skates)

We can put these images up for a vote if you want.

That’s exactly what I mean: What practical relevance does “your” smoothness classification have if it cannot differentiate this serious difference between a path that is comfortable to cycle on and a path that I can just manage to get over “without risk of damage or injury”? In practice, “smoothness” tagging is expected to do nothing else than to be able to reflect such differences (for different traffic modes), so we need to create (or already created) a system that makes this distinction possible.

There is no better transport mode than roller skates to be able to rate a surface as “excellent”. This is why they have always been listed as something like the “reference mode of transport” in the table. If a surface has frequent gaps and therefore creates noticeable vibrations, I look for somewhere else to skate. Like the bike-on-cobblestone-example above, that’s the difference between technically and practically “usability” (if I have to, I can also manage to skate across a cobblestone section on roller skates “without significant risk of damage or injury”, but I don’t understand why this should matter for OSM and why you want to make it the key criterion for OSM).

Oh, wow, I wasn’t aware of this table – just another good source for pre-categorized images.


P.S. But I have to add that, all in all, we have developed a fairly similar understanding of smoothness in the meantime (in the OSM community, but also among those involved here), so I am optimistic that we are very close to a result that seems applicable to everyone. In contrast to four years ago, when everyone had a completely different understanding of the categories and smoothness was really subjective and therefore more or less not usefully evaluable, we have achieved a lot.

So I think we can resolve the disagreements and don’t need to document them.

Regarding SC.
i think the description of the bumpiness (in German “holprig”) fits the asphalt pictures, but the given class is too bad at least for image 4 (bad).

In contrast, the paving_stone images seem to fit the class, but the description “holprig” (bumpy) trivializes the condition. From the small view i could not rank if image 4 or 5 shows the worst surface.

1 Like

I agree that a picture that shows paving stones in worse condition for very_bad paving stones would be better

2 Likes

Not until now… you do have a point: I’ve always felt that smoothness=intermediate is still passable by racing bike, and indeed smoothness=good is passable (though with increased attention) by roller blades. Maybe we should edit the “Usable by” column so that excellent will be “all wheels” and the others adjusted so that the wheels mentioned are “usable with increased attention” (i.e. with lower speed, reduced comfort, etc.)?

I’ve not researched why the phrase “without significant risk of damage … or injury” was included early in the development of smoothness. One big advantage I can see is that this is much more objective and verifiable than “comfortably usable”. Anyone who drives a normal car with ground clearance >14 cm on a road and damages the undercarriage can objectively and verifiably conclude that smoothness=very_bad or worse. It is very personal what people find “comfortable” though: some find it very important, others don’t mind roughing it a bit, and there are masochists… I think it would be better to choose the most objective criterion for determining smoothness and let end users determine for themselves what level of smoothness they are comfortable with. A routing app developer could introduce a “comfort mode” where they allow the end user to set parameters for the amount of discomfort they are willing to accept. A user who doesn’t like to cycle on smoothness=intermediate ways could then set the app to allow max. 100 m on such road (or max. 1% of the total route, or max. 30% longer route to avoid smoothness=intermediate ways, etc.). A user who doesn’t mind cycling on such ways could set the app so no detours to avoid them are allowed.

Agree that it shouldn’t be, and that references to vehicles in the description column should be moved to a new table where the effects of the smoothness values on each vehicle are described (next on my to-do list). In the OP I described why I think an image should be moved, i.e. it is self-centric :wink:
It is unavoidable, because we don’t have an infinite number of categories, that surfaces that are clearly different can still end up in the same category, because they are at the “good end” and the “bad end” of that category.

I don’t want to change it, I want to remove discrepancies in interpretations and the original definitions to be used everywhere, including in the Smoothness Gallery and SC

1 Like

I was invited here, that was 10 days ago. Now I look at a picture, see below, and I have not much of an idea how to tag smoothness there:

Not owning a car, yet I’d say, most cars can go there, if the driver does not care much about the car :wink: I do not think it is suitable for a Riksha, too steep. It is tagged a grade four track. Surface a mixture of dirt and gravel. The administrative GIS says it not useable by lorries, tractors only – they do not care to provide a more fine-grained differentiation and only have three categories: tractor, lorry, lorry+tender. It also says surface unpaved (they also know gravel and asphalt.)

the surface looks like “very_bad” - Car with high clearance, light-duty off road vehicles, adventure motorcycles. The width may be a different problem.

I guess “horrible” is to bad - (off_road_wheels) heavy-duty off road vehicles, dual-sport motorcycles (< 180kg) and all below

1 Like

I also lean towards “very_bad”, just a gut feeling though. This is also what the track is tagged in the data, by me, three years ago though, when I did not have read this topic here and the documentation was different.

I did not make any measurements, but ground clearance >= 18 cm which is documented as the deciding factor certainly not needed. The photo makes that obvious I guess.

Photo of an example type of a light duty off-road vehicle with exactly 18 cm ground clearance. These Fiat Panda 4x4 were used a lot by forestry in Italy and Austria and I have seen them in places where tracks horrible – from the pictures, that is, not from ground clearance necessary.

My experience is that it’s often hard to show in a picture how bad a surface is: it usually looks better on a picture than in reality. Having only the picture, I would guess that it’s just passable by a normal car, because I don’t see obstacles higher than 14 cm, so that would be smoothness=bad (though I would worry about my car, so it’s at the bad end of bad). But you know the situation by personal experience, so I can imagine very well you’d tag it as very_bad I think a Fiat Panda 4x4 is a good example of a “light-duty off road vehicle”.

You might need a 4WD to have enough traction to take the steep slope, but that’s not because of the smoothness of the road surface but because it has loose stones. It might be too steep for a ricksha, but that’s also not because of the smoothness but because of the steep incline. A ricksha may well be able to go down that slope (though it will be very uncomfortable). I think smoothness should not be dependent on the direction of travel.

1 Like

I happened over this track because a part of it recently got improved, to much fanfare in the local press. This is a part of the “city” with isolated dwellings and residents (weekend?) of course drive there with their ordinary cars. I observe those sometimes parking near the hut. My gut feeling told me to map the improved section as “bad” and I changed track_type too, so I did hesitate to map the remaining section the same even though the current documentation would tell me to do that. In the end I just did not carry over any smoothness tag to the remainder.

Ideally, smoothness would strictly refer to surface quality, access tags would only indicate legal permissions, and there would be a separate passability tag to assess general suitability for different vehicles.

However, since no such passability tag exists, mappers have been—and will continue—using smoothness and access tags to convey vehicle suitability.

For example, I use smoothness=impassable on paths that may have smooth surfaces but are inaccessible due to steep inclines, fallen trees, or overgrown vegetation. Some of these paths were also tagged as bicycle=no or highway=footway.

I think smoothness is that tag :slight_smile:

I would tag such paths with incline=* (and let the map user decide if that is too steep for them), barrier=log, obstacle=fallen_tree or obstacle=vegetation It is more informative than just mapping that a path is impassable, and gives the map user more freedom to decide whether to use that path or not, depending on skills, desire, need to use, etc. An MTB rider might decide that he’ll climb over the fallen trees carrying his bicycle and then continue cycling.

2 Likes

I don’t have a particular opinion on the proposed gallery or the specific images, but would it make sense for the example images to be ordered from high quality to low quality for each smoothness value? This would better communicate that there are no obvious smoothness categories in reality and the “border” between adjacent values like excellent and good) is somewhat arbitrary.

For instance, for the asphalt pictures, I think that for excellent category, a natural order would be:

  • 65_Asfaltado_de_calles_(12523687353).jpg (that’s the visibly smooth curvy mountain road with dark asphalt)
  • Výstavba_D8_u_Dobkoviček,_čerstvý_asfalt.jpg (looks a tiny bit less smooth due to more granular asphalt, but surely still excellent)
  • Mala_Hrastice_2020-06-16_Ulice_z_navsi_na_nadrazi_obr05.jpg (still no cracks visible, but the village environment and presence of utility covers in the road makes it clear that there are minor places where smoothness may be at the lower end of excellent class)

While the “right” ordering is up for debate, I think that any ordering would help to better understand the range that a smoothness category entails and better distinguish “borderline” from “typical” examples.

2 Likes