Revision of Smoothness Gallery

[rhhs (Richard)]

Background

For quite some time after the smoothness key was introduced, the wiki contained very little guidance on how to tag it. The table did not have the “Description” column yet, and images in the photo column were frequently changed. After gaining some personal experience with smoothness tagging, I tried to write down my own practice by introducing the “Description” column. This was discussed with only a few participants on the wiki talk page and not on this forum (which I didn’t know about then). Independently, a group of mappers in Berlin also developed more detailed guidelines, summarized here. The two sets of guidelines deviate from each other esp. in the intermediate – very_bad range, where I interpreted “usable” in the note below the table as “passable” while the Berlin group interpreted it as “comfortably usable” (also because they aimed at using the tag for describing urban roads that are seldom so bad that high clearance is needed).

I started a thread about how to resolve this divergence of tagging practices in March, here. From the discussion and esp. the poll result here I concluded that a clear majority of mappers agree with my interpretation of how smoothness should be tagged.

I then started editing the smoothness gallery in my own user space here but didn’t manage to complete it and put it up for discussion until now. Compared to Key:smoothness/Gallery - OpenStreetMap Wiki I made the following changes:

Asphalt

Moved Crumbling asphalt path.jpg and Asphalt - bad.jpg from bad to intermediate (they show a surface similar to that in 20201129 114019 (2).jpg, and don’t show any potholes that would force a car to slow down considerably).

Moved Kamieniec Szalejow Gorny road.jpg from very_bad to bad (although the potholes are quite deep, a normal car can pass there, though very slowly. Having high clearance would not help, and an SUV would have to move equally slowly)

Added Asphalt - very bad.jpg and High clearance asphalt.jpg to very_bad

Concrete and concrete:plates

Moved Concrete plates - bad.jpg from bad to intermediate (does not show any potholes, no need to slow down in a car)

Paving stones

Moved Paving_stones_-_bad.jpg and Cracked_pavement.jpg from bad to intermediate (does not show any potholes, no need to slow down in a car) Paving_stones_-_bad.jpg was discussed at length here

Moved Broken sidewalk near the shore of the St. Lawrence River near Aultsville, Ontario.jpg from very_bad to bad (no need for high clearance, looks as bad as 20230418 142605c.jpg)

Sett

Moved Sett - bad.jpg and Nieborowice, Dworcowa 01.jpg from bad to intermediate (again does not show any potholes, and no need to slow down with a car). Moved Very bad sett.jpg from very_bad to bad (no need for high clearance). I am not very happy with the pictures that are now in bad. I think they are on the edge of intermediate and bad: the stones are irregularly placed, but there are no major potholes and a car would hardly have to slow down. Esp. Residential street with irregular sett paving.jpg is at the bad end of intermediate, I think. Better pictures are very welcome! In my experience, the smoothness of sett surfaces varies very little: almost all are intermediate because potholes rarely form in them.

Unhewn cobblestone

Moved Bad Radkersburg Murgasse IMG 0583.jpg and Mukulakivitie Porvoossa.jpg from bad to intermediate (although probably very noisy to drive on, it’s not necessary to slow down to prevent damage to the car, and there are no potholes). Moved 02.Trinidad (59).JPG from very_bad to bad (no obstacles that would require high clearance).

Compacted

Moved all images previously in intermediate to good (they all look very similar to me, and it would not be any problem to ride them with a racing bike. With a car you might want to slow down because of loose stones, but I think this is a property of surface=compacted; the smoothness is similar to a surface=asphalt, smoothness=good road).

Moved Compacted - bad.jpg and Rough compacted track.jpg from bad to intermediate (driving over these surfaces will cause vibrations, but there are no potholes that you need to slow down for).

Added Dirtroadpotholes.JPG to bad (a fine example of a compacted road with potholes)

Can I have your thoughts please? Do you agree with the moves, or not? Why? What would you do differently?

Unpaved

Below the table for paved&compacted surfaces, I created a new table for gravel&unpaved surfaces that shows smoothness columns for intermediate to very_horrible. I moved the gravel and fine_gravel row there and started to add pictures for unpaved roads but did not finish the job. I hope to discuss this later, when I feel it’s ready and after we have reached consensus on how to tag paved ways.

Additional pictures are very welcome, esp. for unpaved bad (these surfaces tend to go from intermediate straight to very_bad as they deteriorate: the surface is often soft enough that it is smoothed in the driving direction, so ruts form before potholes are formed).

[mcliquid]

@rhhs

Can I have your thoughts please? Do you agree with the moves, or not?

Thank you very much for the effort you have put into this.

Generally speaking, I feel that there has been a change at the intermediate level, and that things there had a movement to worse. This is not a problem in itself, but it does widen the gap between the good and intermediate levels.

From a road cyclist’s point of view, for example, it is still possible to ride a smoothness=good path without any problems. However, your changes make it increasingly uncomfortable to use a smoothness=intermediate path.
From my point of view, we now have almost perfect asphalt at good and, to exaggerate, almost no intact asphalt at intermediate right next to each other.

When I think of the more inexperienced users who might use StreetComplete and make this distinction, I see a great danger that only one tendency in one direction would have really serious consequences.

I like the distinction made by our colleagues in Berlin better, because the gradation has been made more gently. Of course, this is completely subjective.

When I look at my frequently used routing profiles, it seems as if they agree with me.

The following two BRouter profiles show that the router (in this case the profile) can still work reasonably well with intermediate, or assume that intermediate can still be used more or less well with a road bike. From bad on, it is rated significantly worse. When I look at the pictures in your gallery, I don’t want to come by on my road bike.

fastbike-verylowtraffic:

assign smoothnesspenalty
  switch smoothness=excellent     1.0
  switch smoothness=good          1.1
  switch smoothness=intermediate  1.5
  switch smoothness=bad           2.5
  switch smoothness=very_bad      3.5
  # unknown, assume "good"
  switch smoothness=              1.1
  # horrible, very_horrible, impassable
  5

ffmbb-long-distance-cycling:

assign smoothness_penalty =
    if smoothness=excellent|very_good|good then -0.2
    else if smoothness=intermediate        then 0.5
    else if smoothness=                    then 0
    else 50.0

[rhhs (Richard)]

Thanks for the effort you put into your reply!

@mcliquid

Generally speaking, I feel that there has been a change at the intermediate level, and that things there had a movement to worse. This is not a problem in itself, but it does widen the gap between the good and intermediate levels.

I think it is inevitable that when there have been 2 different tagging practices and we make a choice for one of them, the ones that used the other one have to move a bit. The Berlin practice has been adapted for city use, where there isn’t as wide a range of smoothnesses as there is between first-world cities and third-world mountain/desert/jungle tracks. If we want to make smoothness applicable worldwide, we will have to have a “steeper” slope from the best to the worst smoothnesses to cover the whole range (or increase the number of values, which I don’t like).

In fact when I’m out mapping smoothness with StreetComplete (ignoring the pictures, which are based on the Berlin practice), I most often doubt between tagging with good or intermediate. The racing bike mentioned as being able to use good but not intermediate is not a very good probe: I’ve ridden many thousands of km with my racing bike when I was young, including many smoothness=intermediate roads that are a bit uncomfortable but passable for a racing bike (i.e. there’s no “risk of damage … or injury”). surface=sett roads are almost smoothness=intermediate by default, but usually quite passable for a racing bike. Many unpaved roads (esp. compacted) are no problem for a racing bike either. So I usually try to imagine how much vibration you would experience driving or riding over that surface, and tag it good when it’s “sometimes” and intermediate when it’s “all the time”.

[waldyrious (Waldir Pimenta)]

Great work, @rhhs! Here are some comments regarding the “Paved and compacted ways” section:

@rhhs

Moved Crumbling asphalt path.jpg and Asphalt - bad.jpg from bad to intermediate

I don’t see this example as in the same class as the other ones in intermediate. It seems to be a much bumpier ride than, say, 20201129_114019_(2).jpg. Compare:

Crumbling_asphalt_path.jpg	20201129_114019_(2).jpg

I can totally see the second one as acceptable (if uncomfortable) for traversing on a bike, or pushing a stroller, but the first one seems like it would require considerably more effort and/or attention to navigate around the more degraded areas.

I understand that pushing it towards bad compresses the scale in that end, forcing it to cover a wider range of surface qualities; and agree that adding more steps to the scale is not desirable, but I’d rather compress the upper end of the scale instead — say, moving File:Strassenschäden.JPG to good, making that category cover a wider range of surfaces:

File:Strassenschäden.JPG	File:Small road with a few repairs.jpg

In the same vein, I would move File:Concrete plates - intermediate.jpg to good (again expanding the definition of “good”); and in this case I do agree with your move of Concrete_plates_-_bad.jpg from bad to intermediate.

The move of File:Cracked_pavement.jpg from bad to intermediate also seems sensible and compatible with the other changes.

In the sett row, I agree with the changes made, and I’d consider also adding File:Calçamento de pedras.jpg as an example of very_bad (honestly, I’ve seen worse still )

Also in favor of the changes in the unhewn_cobblestone row (I’m surprised about the distribution in the current page, actually).

[mcliquid]

@rhhs

I most often doubt between tagging with good or intermediate.

(I have only marked this paragraph, but my answer refers to the entire paragraph)

I agree with your explanation and think that it needs a basic consens as to whether the smoothness tag means that a way is ‘good/easy to travel’ (without restrictions) for a particular mode of transport or whether it means ‘just possible’ (slower, bumpier, sometimes dangerous).

To stick with the racing bike, even if this is only one of many possible means of transport, let me explain my understanding with a few examples.

A typical sett surface: File:Cobblestone pavement of the Alexandru D. Xenopol Street 1.jpg - Wikimedia Commons

In the old towns/city centres in my area in particular, there is still a lot of cobblestone pavement, which is roughly the same quality as in the picture. The stones are quite small (about 10cm?), the gaps are no more than 1cm in size and they are always filled with concrete. None of the stones are loose and the surface is relatively even.
There are also still some mountain passes, especially in southern Switzerland and northern Italy, that are based on cobblestone roads.

One example is the north-east ascent of the San Bernadino from Thusis: bikerouter.de

There is no legal way to avoid the two cobblestone roads in the village of Splügen and in Nufenen. It’s okay-ish on the way up, but on the descent I definitely don’t ride over it at the same speed as on the asphalt sections just before it, otherwise I would definitely risk injury or material damage. And it is noticeably more strenuous for the legs to pedal due to the higher rolling resistance and for the hands to hold on due to the vibration.

Since the smoothness would pass as intermediate according to the current description in the wiki, I would want to avoid exactly this value with my road bike wherever possible. Because at least I want to be able to get around quickly and smoothly.

Another example can be found in many towns in northern Germany: paving stones. (Random example image)

Most of the paving stones that I have seen in urban areas would probably even pass with a good rating according to the reference images in StreetComplete. Therefore, I have adjusted my bikerouter.de profile so that paving stones in general are drastically downgraded, regardless of their smoothness value.

Am I getting across what I want to say?

[rhhs (Richard)]

@waldyrious

I don’t see this example as in the same class as the other ones in intermediate. It seems to be a much bumpier ride than, say, 20201129_114019_(2).jpg. Compare:

I think the two are quite similar. The crumbling asphalt on the left has damage over a larger area that can’t be avoided when cycling over it, but the pothole in the right foreground is less deep than the one in the centre foreground of the right picture (which I took myself). The right road is easier for a bicycle because there is more good asphalt that can be used to circumvent the bad bits. But for a car the bad bits can’t be circumvented. I’d say the left picture is towards the edge with smoothness=bad, but the right one is also worse than average for smoothness=intermediate. We could put the two up for a poll?

@waldyrious

say, moving File:Strassenschäden.JPG to good,

I think the damage it shows is typical for a smoothness=intermediate surface, but there isn’t much of it so we could also say that it “should be tagged with the quality of its best passable part” which is smoothness=good. We could add a caption and explain that.

@waldyrious

In the sett row, I agree with the changes made, and I’d consider also adding File:Calçamento de pedras.jpg as an example of very_bad (honestly, I’ve seen worse still )

I disagree: I don’t see anything that would need high clearance (something sticking out more than 14 cm) so I’d tag it smoothness=bad, and I think it’s surface=unhewn_cobblestone (most of the stones are not rectangular).

[rhhs (Richard)]

@mcliquid

I agree with your explanation and think that it needs a basic consens as to whether the smoothness tag means that a way is ‘good/easy to travel’ (without restrictions) for a particular mode of transport or whether it means ‘just possible’ (slower, bumpier, sometimes dangerous).

I think the consensus from the early days of smoothness is that it’s about “usability” (quote from the proposal, voting ended 2 Nov 2008), which is defined as “that the vehicle can traverse the route without significant risk of damage (e.g. to vehicle undercarriage/drivetrain) or injury (e.g. by falling)” (added to wiki 26 Nov 2008), i.e. ‘just possible’.

@mcliquid

I definitely don’t ride over it at the same speed as on the asphalt sections just before it, otherwise I would definitely risk injury or material damage

I think that describes ‘just possible’ very well: it is possible to use the way, but you have to slow down to reduce the risk of injury or damage to an acceptable level. It isn’t comfortable though (however some others might think it’s fun), so if you want your router to avoid such ways, you might have to tell it that you’re on a racing bike… (Tour de Suisse and Giro d’Italia do pass the San Bernardino sometimes, though, so for them smoothness seems to be good enough).

[waldyrious (Waldir Pimenta)]

@rhhs

I think the two are quite similar. […]

I’m sorry, I couldn’t follow your reasoning in this paragraph. I read it several times but I’m still confused. Can you perhaps phrase it in a different way?

@rhhs

@waldyrious

say, moving File:Strassenschäden.JPG to good,

I think the damage it shows is typical for a smoothness=intermediate surface, but there isn’t much of it so we could also say that it “should be tagged with the quality of its best passable part” which is smoothness=good. We could add a caption and explain that.

I honestly don’t think that criterion would be very useful for surveying. I tend to select the value that matches the average surface quality over the entire way (or split it, if the difference can be clearly demarcated).

But overall, I think it would help if you would comment on my overall suggestion of expanding the breadth of surface qualities covered by “good”. Does that align with your intention for this revision?

In particular, I suspect that the “good” designation can bias us to interpret it as meaning “above average”, but I typically interpret it (and I believe this matches your goal of allowing the scale to apply more globally) as meaning “OK” or “acceptable”. It’s only when I see surface=intermediate that I start assuming that I’ll have to take a bit of care or attention (or expect noticeable discomfort) when traversing it.

File:Strassenschäden.JPG shows a road where the (visible) balance of surface qualities is much more towards good than bad, in terms of area and ability to avoid the “bad” bits (which are really quite mild), hence my suggestion of placing it in the good category.

@rhhs

@waldyrious

In the sett row, I agree with the changes made, and I’d consider also adding File:Calçamento de pedras.jpg as an example of very_bad (honestly, I’ve seen worse still )

I disagree: I don’t see anything that would need high clearance (something sticking out more than 14 cm) so I’d tag it smoothness=bad, and I think it’s surface=unhewn_cobblestone (most of the stones are not rectangular).

You’re right, it does fit better in the bad category — I didn’t consider this carefully enough.

As for being surface=unhewn_cobblestone, my understanding is that sett stones are artificially flattened at the top, not necessarily square or rectangular in shape (otherwise it wouldn’t even make sense for sett:shape=irregular to exist) — whereas unhewn cobblestone is about stones that are not super flat to begin with, and when they are, it’s due to a natural process/erosion making them so, rather than a man-made cut.