Add smoothness quest

[westnordost]

Alright guys, here is the new PR that fixes #1630. I created a new one because the old one (#3257) already had so many comments that it became quite messy. So here is a summary.

Since by now, so many people are involved and engaged in this PR by researching, by collecting photos and otherwise, I took great care to check and refine the texts again for conformity with the following sources:

• https://wiki.openstreetmap.org/wiki/Key:smoothness (and linked pages)
• https://wiki.openstreetmap.org/wiki/User:Rhhsmits/Smoothness_details
• https://wiki.openstreetmap.org/wiki/Berlin/Verkehrswende/smoothness

I also looked through all the photo collections contributed so far, either by linking to some wikimedia commons or by actually posting collections of self-made photos. Thanks again everyone for contributing, here is a summary of photo collections posted that I had a deeper look into:

• @NicoHood: https://nextcloud.nicohood.de/index.php/s/z6qs58jSQcBXKSw? (CC-0)
• @westnordost: https://drive.google.com/drive/folders/1RPtlFVBATcJ5JH9oFfCiVYMBRfitRbO- (CC-BY-SA 4.0)
• @mcliquid: https://drive.google.com/drive/folders/1X0CU3Yjm02sZSKcgre27k2Oe8dK65ikO (CC-BY-SA 4.0)
• @matkoniecz: https://commons.wikimedia.org/wiki/Category:Mam_Tor_road
• @rhhsm: https://www.dropbox.com/sh/mkjhco2ob60q24l/AABZo9f5bgyA1KwyV-3zxUVYa?dl=0
• @Helium314: https://cloud.disroot.org/s/bWCMMSGigAR7k8s (public domain)

For the photos, too, I took great care with the selection of the photos and supplied reasons why I chose one photo over the other further below

I put both the now selected pictures and the adapted wordings into this document (click!):



Thanks for starting that document, @tordans!

The wordings take a lot from the wiki page on smoothness and the "Zusammenfassung/Faustregeln" from the Verkehrswende/smoothness page, which nicely complement with each other.

FAQ

I also scanned the old PR #3257 for any decisions and reasons for these made there and summarized them:

Why ask for "surface quality" and not simply for "smoothness", as this is also the tag's name?

@smichel17 wrote earlier:

The wiki says smoothness is about the physical usability of a way for wheeled vehicles, particularly regarding surface regularity/flatness. So, your challenge is to communicate that. In other words, the word "smoothness" is jargon; it has a particular definition that is close to, but not the same as, the English definition of "smoothness". You need to keep in mind that the audience only knows the English definition, and make sure the question does not accidentally rely on knowing the jargon definition for context.

Why not use the smoothness values as titles?

@Helium314 wrote earlier:

[smoothness tag values are] quite subjective, a well made sett surface could be called excellent, even if it's bumpy [...].

So, it is better to describe the bumpyness than to describe how "good" or "bad" they are because the latter is connected with the surface type. The literal tag value is judgemental and may only help people both using the US-English locale on their phone and knowing the tagging scheme (well). For the rest, it may actually be more confusing than helpful.

Why not mention how much cars would need to slow down in descriptions? (wiki does)

@smichel17 wrote earlier:

"Would a car slow down to go over it?" is kind of subjective, because it depends how fast the car was going in the first place.

So, if there is a bad surface and a speed limit of 30 kph in a residential area, it doesn't really affect the max speed a car goes. Instead of referring to how much cars have to slow down, we mention that the smoothness (starting with very_bad is potentially dangerous), i.e. caution by whoever uses it is advised, i.e. they have to slow down considerably.

Why show emojis for the vehicle types?

The wiki mentions some vehicles in the "usable by" column. We chose to just display a small emoji with such a vehicle type because

• the emojis are hopefully a useful indicator to better judge for which vehicle type a surface is still usable, but the focus is on the description of the property of the actual surface, so it is a sidenote. The surveyor does not have every vehicle type at hand to try out the surface on-site.
• what's usable or not by any such vehicle type is debatable, so it is avoided to explicitly write that smoothness X should be usable by vehicle Y. It is only hinted at with these icons.
• if there is too much text, people don't read it. Enumerating the vehicle types in prose would be too much text. An icon keeps it short

Why not mention concrete cm-limits for crack widths? (wiki does)

Exact measures for maximum crack widths seems to be an objective measure at first, but it is not that great after all:

A road that has some cracks, has lots of them. Nobody is going to run around and measure every single crack. See e.g. this picture for intermediate asphalt on the wiki. Is there any crack above 2cm in width? Such an exact measurement will make it harder, not clearer or easier to record the right value because the user has to make his decision based on the big picture, not some details at some spot. Let's not fool ourselves, the surface tag is an approximate.

So, one reason the vehicle type emoji is shown is to suggest to users that using that vehicle('s tires) are still OK, and for which tires a crack causes problems is a better measurement than exact centimeters.

What are the picture selection criteria?

• sufficient resolution and quality. The cropped part must have at least 600px in width and have no artifacts
• open license
• similar contrast, no stark shadows or highlights (clouded in daylight are usually the best conditions)
• not include other details (leaves, dirt, grassy knolls, things) or variations (colors, shapes) that can be mistaken to be meaningful
• should show a typical example of that category, i.e. should not show some edge case that also falls under that category. Should show something that is in the middle of that category as there is obviously a smooth transition between the categories in reality
• should each be photographed from roughly the same angle (~ °30), i.e. an angle where you can see both the details in the foreground and the big picture in the background
• must not contradict currently used pictures on the wiki, i.e. put a currently used picture in another category
• a sense of scale if possible to evaluate how big are those cracks for example
• should show the overall picture rather than focus on one detail (e.g. pothole) as the user has to make his decision also on the overall picture, not some detail. Hence the suggested steep angle so that at the same time, such details can be visible

Alternative candidates for pictures

...that I considered. In the end, only one picture could be selected, so here are the (sometimes quite unimportant) reasons why the picture was not selected

Asphalt

excellent

• most 1_excellent_*.jpg from @mcliquid. Current picture shows that slight imperfections should not result in them being classified in a lesser category

intermediate

• Strassenschäden.jpg - might be seen as almost good
• Asphalt_-_intermediate.jpg - used in Verkehrswende/smoothness. The damages shown are a little few, not a big difference to good
• Broken_road,_Castleton_07.jpg and Broken_road,_Castleton_08.jpg - this kind of crumbling paving with no patches but also no other kinds of damages at all is a little uncommon, visually a bit too similar with current picture for bad

bad

• 20201129_114019+(2).jpg - the previous picture for bad but for intermediate in the wiki. IMO it's rather bad, but the fact that it is an example for intermediate on the wiki disqualifies it for bad
• Asphalt_-_bad.jpg - current picture for bad on Verkehrswende/smoothness. Even though it shows definitely more damages than the typical intermediate road, it is a bit on the upper end of bad (compare with bad paving stones/sett etc)
• Typical Torontorian Pothole(s) - close-up photos are little helpful to convey the big picture (and that is what the user has to make his decision on)

very bad

• Potholes_at_the_Level_Crossing,Barrow_Haven-geograph.org.uk-_1621073.jpg - the previous picture. It is not clear how deep the potholes are, difficult to see details due to low resolution
• A bit of bad road - geograph.org.uk - 2236824.jpg - too low resolution

horrible and very horrible:

• Broken road, Castleton 17.jpg - the issue here is that it shows a ruin of a road that is basically impassible. Impassible roads are very likely not even mapped as roads anymore. If only the lower part of the image was shown, it may classify as horrible or very horrible, but then the big picture is missing.

Paving stones

excellent

• 1_excellent_4, 1_excellent_5, 1_excellent_6, 1_excellent_7 from @mcliquid's collection look too much like indoor paving and have some patterns, i.e. would not be the typical outdoor excellent paving
• 1_excellent_1, 1_excellent, 1_excellent_2 from @mcliquid's collection are outdoor, but the granite pattern may be mistaken visually for a rough surface at low resolution
• Sopron-002.jpg (previous picture) had far too much contrast
• 2_excellent, 1_excellent_8 from @mcliquid's collection and IMG_20210916_111822.jpg, IMG_20210916_112421.jpg from @NicoHood's collection and 20210924_151609.jpg are all excellent. I chose 1_excellent_8 because the extremely even pattern is more striking with smaller stones plus in some of the others, the seams are in part darker which may in low resolutions look like the stones are not perfectly even.

good

• all of @mcliquid's collection don't look visually distinct enough from excellent. Sme with most of NicoHood's collection: either too top-down, the paving stones are too non-standard or not distinct enough from excellent.
• my photo (lower part) is too top-down, so the big picture is missing and the seams are difficult ot see
• another of my photos is representative, but again too top-down

intermediate

• Cracked pavement - borders on bad(?)

bad

• racked pavement 1.jpg - does not show the big picture but just this spot. If there were no damages, this might even be excellent or at least good

very bad

• patch_concrete_brick_broken_up_root_force_paving_structure_concrete-1067135.jpg and broken sidewalk - do not show the big picture but just this spot
• Bali, Indonesia - problematic license. Unusual colors and shapes, extensive damages. Doesn't fit well to the other photos in style and perspective
• Broken up sidewalk near the shore of the St. Lawrence River near Aultsville, Ontario.jpg - unusual situation, could have been paved like this from the start. But still best of the bunch so far

Sett

good

• all 2_good_* from [@mcliquid's collection]https://drive.google.com/drive/folders/1WfD6PA-Pnm5C5eDrFztZAMM1kl66X3Xk) are suitable, however the first 5 may set the bar too high cause the seams are not well visible and the last 2 are too close to intermediate already. The rest is a little colorful / showing the granite pattern a little much
• IMG_20210916_112124 from @NicoHood is good but I chose IMG_20210916_111924 because it should show that good smoothness is achievable with large stones too
• Stasjonsallmenningen i Barcode sett mot nord.JPG - (current picture) it looks newly built, it's better to show something in-use

intermediate

• Palace Square setts (SPb, Russia).JPG - too close to good, unusual pattern
• Sett - intermediate.jpg - a little too close to good. Even though they have a rough surface, such stones feel mostly the same as "normal" concrete paving because they are very regular and the granularity of the surface too fine that it would matter much on anything except the tinyiest wheels

bad

• IMG_20210916_112506.jpg and IMG_20210916_112601.jpg from @NicoHood's collection fit but due to the top-down perspective the photo was taken, it's not well visible how deep the gaps and how chiclet-like the stones are but these are major factors
• @mcliquid's images are mostly fine, however 4_bad_1 and 4_bad are a little untypical and might even be intermediate even though they look a bit broken, 4_bad_3 and 4_bad_4 are difficult to see/blurry and 4_bad_2 is good but not better fitting than current picture or the ones from @NicoHood
• Surface sett very bad.jpg - has been a proposal on the Verkehrswende/smoothness page for very_bad but IMO looks more like bad

compacted

intermediate

• both Compacted - intermediate.jpg and 3_intermediate_2 - could be mistaken for fine_gravel (maybe prompting users to say that the surface displayed is not the correct one), current picture is more inclusive in that it is clear that more loamy surfaces are also compacted

bad

• there are some in rhhsm's collection such as IMG_20210530_134448912_HDR.jpg or IMG_20200927_113529403.jpg (too much contrast though), however current one fits best because it shows various things like bumpyness, loose larger rocks etc

horrible

• The way home from Christmas dunes - two and ahalf hours to go 80 km - killer ruts (16058640321).jpg - could be mistaken for dirt/ground roads. Compacted roads usually have larger gravels below the finer gravel surface, so if they become damaged erosion, the bigger stones surface

gravel

intermediate

• Tamsa.JPG - unusual perspective, blurry in the front, very similar to the picture for bad
• 3_intermediate.jpg - looks a little to uniform/clean, size of gravel is difficult to ascertain from that perspective

bad

• 4_bad.jpg from @mcliquid and Štěrková cesta u Pláštíku.jpg - size is quite regular and it seems different opinions exist whether this would be bad or very_bad

very bad

• Štěrková cesta u Pláštíku.jpg - size is quite more and it seems different opinions exist whether this would be bad or very_bad

horrible

• 6_horrible.jpg - visually not distinct enough from very_bad

What could be improved

Note the picture guidelines.

Would be good:

• Find pictures of really bad paving stone surfaces (including very bad). bad and below paving stones are usually damaged, cracked, displaced, uneven, rough ones etc. - current ones are so-so
• intermediate paving stones: photograph from closer to the ground at steeper angle like with most pictures from Verkehrswende/smoothness) OR find a different picture that fits. Loose-looking stones are a bit uncommon. Current photo is from @Helium314

Not that important:

• excellent paving stones: photograph from closer to the ground at steeper angle (like with most pictures from Verkehrswende/smoothness). Current photo is from @mcliquid
• good sett: photograph from closer to the ground at steeper angle (like with most pictures from Verkehrswende/smoothness). Current photo is from @NicoHood
• bad sett: current picture tends towards intermediate (visually). A similar photograph (same perspective etc.) but with some irregularly shaped, more weathered and/or broken stones mixed in would be a better picture.
• very bad sett: use different perspective - from closer to the ground, down the street. Current photo is from @westnordost

[rhhsm]

In a nutshell, "usable by" as an objective and primary measure to judge the smoothness is more subjective as it sounds like, because what is usable and what not is in itself highly subjective, including when a certain surface quality forced people in vehicle to slow down or drive more carefully. Looking for specific damages and properties (as described and as shown in pictures) that result in that a surface may not be eligible for certain vehicles anymore is more objective.

I think it's the other way around: the use case is that we want map users to be able to decide whether a way can be used with their vehicle, or should be avoided. "Smoothness" is a measure of usability, not the other way around. Usability is unfortunately rather subjective, so I tried to make it less so by adding the descriptions column in the wiki (where the appearance of the surface is just one of the criteria, together with need to reduce speed and ground clearance). I think if we have a choice between a tag that is very useful, but subjective, and one that is objective, but useless, we should go for the first.
I also think that SC should follow the consensus reached on the OSM wiki, and not set its own standards. If we want to change the consensus on a certain tag, this should be discussed at the wiki, not here.

[westnordost]

@rhhsm
Sure, in the end, for data consumers, what matters is a measure how usable it is (for different vehicle types).

I understand you added the descriptions column in the wiki only a year ago. (Which was even after the Verkehrswende/smoothness page was created and discussed.) But as you wrote yourself in your last comment, you added it to make the selection less subjective, i.e. less dependent on what each individual would classify as "usable". And you certainly didn't pull this description out of your ass, you carefully thought of which objective criteria could be named that makes a surface being properly usable by a certain vehicle type.

What is being refined here is to mention surface-specific properties in the descriptions such as the gaps between and shapes of the stones (for paving stones, sett), damage like potholes and ruts (for asphalt), erosion damage (for unpaved surfaces) etc. to make it even more objective and clear. Same for pictures.

This doesn't make smoothness "objective but useless" because your descriptions, Verkehrswende descriptions an our descriptions now are derived from the evaluation what vehicle type will still be able to use a surface.

So, in short, I fiercely deny your claim that this implementation doesn't follow consensus or that it sets its own standards only because it does not mention "usable by XYZ". As I (in the FAQ) and others wrote before, for an on-site contributor, "usable by racing bike" is a relatively useless hint unless the particular surveyor happens to be on or at the very least owns a racing bike and ranks average on the "careful-venturesome" scale.

[westnordost]

Here are two tests trying to incorporate the "usable by" column from the wiki more, in a subtle way:

🚀	👀

Exact wordings aside (in hindsight, I think "sufficient for" is better than "no issue with"), 👀 really adds a lot of clutter in the UI, the texts become borderline so long that I fear users will tend to not read it at all anymore and solely base their decision on the example image.
🚀 I find okay because it doesn't fill up much space and users can scroll past it when they make their selection. Also, it is one of those well established "hint" texts users will need to read once and will go unseen once they internalized it.

I chose emojis to refer to these tests so that you can give feedback on it via a reaction (use 🎉 for "neither, keep as is!)

[rhhsm]

I think we have to accept the fact that to be able to tag this correctly (objectively and consistently), some reading is required. Would it be possible to add an introductory text (or a reference to the wiki) the first (few) times a SC user attempts to solve this kind of quest? Maybe it could be deselected by default, and to be selected, the user has to confirm that he understands the details of the tag and has read the wiki (similar to the speed limit quest)?
I prefer 👀, especially if the icons are replaced by text (for instance "usable by SUV"). The icons add to the clutter, it would be better if it was just text.
The "very bumpy" (very_bad) picture is not bad enough I think: it is still usable by a normal car and doesn't need high clearance.

[westnordost]

The "very bumpy" (very_bad) picture is not bad enough I think: it is still usable by a normal car and doesn't need high clearance.

It's from this picture (click).

That was the worst picture of a road I could find that wasn't totally broken (i.e. closed off). I think the photo fits very well though. No sedan user in his right mind wouldn't slow down to very slow for passing these huge potholes. SUV users might have more confidence. A road that looked like this continuously would certainly not be safe for normal cars.

I know, "usable by" can be widely interpreted. What seems to have emerged from the long community process of selecting the pictures (I mean both in the last ticket and in the Verkehrswende wiki), is the interpretation that this means it is OK to use. I.e. maybe not entirely without issue, but still in the range of normal usage. smoothness=bad can be used by city bikes or even racing bikes after all, but they'll avoid it if they can. If "usable by" was to be interpreted in the strict sense like "theoretically usable by", both
and would need to be classified excellent because someone on roller skates could avoid all those damages if he tried hard enough. Certainly, this is not how the tag is intended because it wouldn't make sense.

[westnordost]

I went through the whole PR and its contributions again, uploaded all the pictures I found most representative to the openstreetmap wiki that weren't yet and weren't yet on wikimedia commons and created this page:

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

This page not only includes the pictures chosen to be displayed in the app, but many more that have been contributed along the way. I thought it would be a waste to throw this away. Feel free to extend or refine it. Many surface types are missing, some surface + smoothness combinations could use some more example pictures on variations.

[rhhsm]

The "very bumpy" (very_bad) picture is not bad enough I think: it is still usable by a normal car and doesn't need high clearance.

It's from this picture That was the worst picture of a road I could find that wasn't totally broken (i.e. closed off). I think the photo fits very well though. No sedan user in his right mind wouldn't slow down to very slow for passing these huge potholes. SUV users might have more confidence. A road that looked like this continuously would certainly not be safe for normal cars.

You would certainly need robust_wheels to do this road, but high_clearance doesn't help. I find it hard to image that there are roads that need high clearance to navigate but can still be considered paved, and the difficulty in finding good pictures of such roads is evidence for this. That's why I started the description of very_bad in the wiki with "Unpaved roads with ...". Most of the pictures in the very_bad column of the wiki gallery (thanks for adding!) don't need high clearance either (only the pictures of General_Wade.., smoothness_very_bad and Mountain_path... show some large rocks that could really kill the oil pan of your normal car).

I know, "usable by" can be widely interpreted. What seems to have emerged from the long community process of selecting the pictures (I mean both in the last ticket and in the Verkehrswende wiki), is the interpretation that this means it is OK to use. I.e. maybe not entirely without issue, but still in the range of normal usage. smoothness=bad can be used by city bikes or even racing bikes after all, but they'll avoid it if they can. If "usable by" was to be interpreted in the strict sense like "theoretically usable by", both would need to be classified excellent because someone on roller skates could avoid all those damages if he tried hard enough. Certainly, this is not how the tag is intended because it wouldn't make sense.

It's why I suggested in the wiki that a mapper should consider whether he would recommend the way to a friend, or recommend to avoid it by taking a detour that is 30% longer. I also proposed that only 4-wheeled vehicles should be considered, because for a bike you need only 20 cm of good surface to make it usable. Potholes etc. are much easier to avoid on a 2-wheeler, so what matters is if they are avoidable or not for a 4-wheeler. However there was no consensus for this.

I think we have to accept the fact that to be able to tag this correctly (objectively and consistently), some reading is required. Would it be possible to add an introductory text (or a reference to the wiki) the first (few) times a SC user attempts to solve this kind of quest? Maybe it could be deselected by default, and to be selected, the user has to confirm that he understands the details of the tag and has read the wiki (similar to the speed limit quest)?

What do you think about my suggestion above?

[westnordost]

What do you think about my suggestion above?

The smoothness wiki page does quite a bad job to be helpful for someone on-site to decide on the right category. Anything helpful from the wiki page has already been incorporated in the app. Note that users of the app don't even see the smoothness tag value, because this in itself is confusing as long as you've not read the whole wiki article and internalized it, which StreetComplete does not require for any quest.
Defining/Interpreting "usable by" as "recommending to a friend" doesn't make it any less subjective, it only adds even more text to this struggling wiki page, same with any additional paragraphs explaining the 2-wheeler/4-wheeler problem and more controversies.

What really helps in making contributors use the tag more consistently/objectively is a unified scale with descriptions and example pictures where the roughness/usability of each example picture of different surface but same smoothness is comparable - the Berlin/Verkehrswende/smoothness wiki page and forks of it are hundred times more useful than Key:smoothness. And hopefully in the future when enough eyes have looked at the Key:smoothness/Gallery page and refined it, it can be put more prominently directly on the Key:smoothness page.

[westnordost]

thanks @1ec5 for finding some missing fitting pictures for very good and very bad concrete

[naposm]

thanks @1ec5 for finding some missing fitting pictures for very good and very bad concrete

How about also adding some text like:

• You could skateboard here without any issue
• (some catchy phrase for scooters)
• Might hurt a bit, but you can cycle if you pay attention
• You have to zigzag a little, but you can easily get through with a car
• Even with a car you'll have hard times

(take them as an example, I'm not that good at copywriting 😆)

The essence is telling the user what the icon is intended to represent without being boring (in StreetComplete-style). I think most users wouldn't care (or even notice) the link between the emoji icon and the surface for some time before they actually pay attention to it, so this could help at the beginning, then they will just look at the icon to identify what each surface is intended for.

[westnordost]

@Helium314 you can remove my permission now