Talk:Accepted features/Smoothness

From OpenStreetMap Wiki
Jump to navigation Jump to search

Ok, I agree with the smoothness tag as proposed in you'r tab, because I need something like that for mountain track roads ( saying which vehicule you need at least to drive on it ) Sletuffe 16:28, 16 May 2008 (UTC)

Even being new at OSM: from my view as a biker and inline-skater smoothness is a very practical solution in conjuntion with surface Pmurk65 27 May 2008

smoothness and surface

I dislike the dependencies between the proposed smoothness and the surface key. If I read your list correctly, a way with surface=cobblestone and smoothness=bad is illegal? I'd certainly like to distinguish between various grades of cobblestoned road. In my opinion, smoothness without surface is pointless, hence I'd rather a smoothness key refined surface, so you get excellent or good or intermediate paved roads, good to catastrophic cobblestoned roads, etc. So how about a key smoothness=0 (default), +1, -1 to refine a base smoothness for the type of surface? Robx 07:11, 30 May 2008 (UTC)

This dependency was not my intention. The surface descriptions in the second column are EXAMPLES, not used for definition. The smoothness of a way should be assessed solely based on the usability of that way by the vehicles I mention, NOT based on the surface. This is my whole point: as a user of a certain road or path I am only interested wether I can drive on this road or not. If I sit on a racing bike, I am not interested wether the surface of a way is cobblestone or mud, just wether it is smooth enough to be used with a racing bike or not. I changed the proposal to make that clear (hopefully). With your proposal of a refined surface key, it gets very difficult to draw usage-specific maps (e.g. for racing bikes), since there is a potentially infinite number of surface values. Imagine a piste on a salt lake -- with the surface key it would read "surface=salt", which would have no meaning before the key is integrated into all the renderers and Garmin type files and whatever. With "smoothness=excellent", it would instantly show up everywhere as usable with a racing bike, which is my intention. --Chrischan 11:05, 1 June 2008 (UTC)
I'm beginning to like the proposal. It doesn't store all relevant information, but it's a good start, and appears to be clearly defined. I've started tagging some roads around here, the better cobblestoned roads with smoothness=intermediate and the awful ones with smoothness=bad. Also some really smooth paved roads with smoothness=excellent. I'd suggest adding some default values: A road should have a default of smoothness=good given its surface is paved or unspecified. A road with surface=cobblestone should default to smoothness=intermediate. A track should probably default to intermediate also. Robx 20:09, 2 June 2008 (UTC)
Perhaps an additional value of "unknown" would be best as a default? I've even been on an interstate highway (motorway) which is definitely smoothness=bad, or maybe smoothness=intermediate if you're feeling generous. Just anecdotal, but I think it's a bad idea to imply the smoothness based on the highway (or even surface) classification. --Hawke 21:26, 2 June 2008 (UTC)
I'm sure those exist, but they're certainly the exception. A default of unkown (or rather, no default), includes in the smoothness tag also the information whether the way has been surveyed for smoothness, which may be a good idea. But I'd rather not add thousands of smoothness=good tags and rather state somewhere that I've surveyed a given area for smoothness. Robx 18:37, 8 June 2008 (UTC)

Couldn't this proposal been covered by surface=* to begin with??????????????? --Skippern 22:18, 2 December 2008 (UTC)

Many have tried hard, but surface as never been able to answer "what vehicle do I need to go on that road" As far as I know no renderer is able to take advantage of it, and routing program do only use surface=paved & surface=unpaved at most. Also please keep on reading this page wherever surface is mentionned, and you'll see the problem Sletuffe 22:52, 2 December 2008 (UTC)
If the tool is broken, fix it, don't make more that can break down! The problem isn't the tagging, but the renderers and the routing programs according to what you say. It will not make a difference for the routing program if you have smoothness tagged as well if it only differs between paved and unpaved. You should have focused with on getting the programs accept more surface alternatives from that WAY TOO LONG LIST of surfaces, than adding this IMO pointless feature. --Skippern 10:47, 3 December 2008 (UTC)


I've never like subjective tags such as this. This is why tracktype=* never got anywhere, as it is not obvious what the difference between "bad" and "intermediate" is. For instance, the roads where I live are terrible, but saying as they're all like that, should it be smoothness=bad or smoothness=intermediate? Bruce89 14:19, 30 May 2008 (UTC)

I like it, as long as there are usable guidelines for how to tag. "suitable for roller skates" and "suitable for mountain bikes and 4-wheel-drive automobiles" is quite good, IMO. --Hawke 15:16, 30 May 2008 (UTC)
In my opinion, tracktype=* never got anywhere because it does not describe a singular property of a way, but a combination of properties. Smoothness, although as far as I know not being a standardized physical property, could even be measured physically (I think of a wheel of defined size, measuring the vertical deflection while being dragged along the way). Obviously, this would be very unpractical. But when you honestly try to classify the roads you know into that scheme (based on wether you would drive on it with the vehicles I mention), you will see that the classification is pretty unambiguous in *most* cases. But I think this is very similar to the width of a road: although it is a physical property, few people would actually bring a rule and measure the width, most people would guess. Still it is very good to know wether the width is 2m or 20m, or wether the smoothness is excellent or horrible. --Chrischan 11:36, 1 June 2008 (UTC)


I have to say that I'm not so sure about "catastrophic" for the lowest value, it doesn't seem to fit with the others. I suggest "horrible" as an alternative. I'd support the proposal though. --Hawke 15:20, 30 May 2008 (UTC)

I know what it means, I just don't think it's a very good word to use for a value here. --Hawke 17:06, 30 May 2008 (UTC)
Hawke, why do you think "catastrophic" does not fit with the others? I actually do not care about the values at all, I just want the schema ;-). But maybe I should add Alvs idea to have one more tag essentially saying "no wheeled access possible" beyond the 4wd access value. --Chrischan 11:42, 1 June 2008 (UTC)
It just doesn't fit with the rest, in my mind. "How was that sandwich you ate today?" "Oh, it was catastrophic!" doesn't work. Or -- "Hey, I want to go for a ride on Foo Lane; how's the surface quality?" "Oh, I wouldn't ride there, it's catastrophic!". "Catastrophic" to me would give a degree of badness to something that was already bad ("I had a catastrophic accident on my bicycle today").
I do like Alv's idea of a sixth grade beyond the 4wd level. Perhaps "impassable" would work, especially if this is intended to apply to wheeled vehicles only. Bipeds are quite good at navigating obstacles, and if it's impassable by them, we probably shouldn't be mapping it as any sort of routeable type.--Hawke 16:26, 2 June 2008 (UTC)
How about "very bad", then catastrophic as the lowest value. Otherwise it's too big a jump from grass to mountain tracks. --OliverLondon 16:28, 2 June 2008 (UTC)
I started mapping tracks as I was hiking this week end with the smoothness tag in addition of the highway=track tag, and it turns out that I mapped a lot of tracks with the lowest possible quality ( smoothness=catastrophic ). But I doubt any 4wd could pass the tracks I went to so I missused that tag because It lacks "above" (or below ;-) ) catastophic. So I wonder why those tracks were there If no 4 wheels vehicules can use them ? The answer to my question came 10 minutes later as we crossed their way : 2 differents 4 wheels vehicules, a tractor and a "Quad" ( in french ). So, I am in favour or adding a "horrible" or whatever we should call it level to increase to 6 the number of smoothness tags in the goal of mapping what can drive on after a 4wd. To my mind, Mountain Bike can drive where 4wd ( not the army ones) cannot, the "horrible" tag should fit for 4wd and the "catastrophic" tag for Mountain bike & Tractors & "quads" Sletuffe 13:34, 3 June 2008 (UTC)

impassable ?

ok this looks better because it's less subjective than catastophic, but I don't think it's ok either. Why ? Answer this : how could a track be impassable by 4 wheeled vehicules ? It cannot, by the fact that it was created as a track in the goal to make passing possible what's worse than horrible ? I'm not a native english speaker ? or could we insert a "very_bad" between bad and "horrible" ? Sletuffe 15:43, 4 June 2008 (UTC)

I think impassable wouldn't be applicable to tracks. On a footway, it would say that it wouldn't be passable by mountainbike. It's not a value that would be used much, mostly there for completeness. Robx

Laterally varying smoothness

In fact, it doesn't need to be specified. The proposal is quite usable the way it is right now. I'd suggest voting on this as is, and then considering refinements such as smoothness=bad plus smoothness:left_border=good plus left_border:surface=painted_paving_stones. If you encounter a way that doesn't quite fit, add a note or invent some appropriate tag. Robx 18:32, 8 June 2008 (UTC)

Vote? (and what about sand)

Since discussion seems to have died down without major objections, how about taking this to vote?

I did come across a minor problem: In the local forests, there's some deep sandy tracks (partly bridleways). Since they're barely passable by wheeled vehicles, I'd tag them smoothness=bad, but arguably they're really quite "smooth". Robx 07:42, 30 June 2008 (UTC)

  • Same as you, I'm using it for a while now, and would be sad to see it dropped ;-)
For your remark about "smooth sand" but passable ( only by 4wd ? then I would have tag it smoothness=horrible ), I forsee some confusion just because of the name "smoothness". If I understand this proposal correctly, and as I was saying here before Chrischan transformed it into a nice proposal, the real goal of it, as mentionned here in the proposal : The smoothness of a way should be assessed solely based on the usability of that way by the vehicles mentioned above, NOT based on the surface properties
In clear, even if it is a perfectly smooth surface such as a wall, because it isn't passable then it shouldn't be tagged with smoothness
Ok my example is bad because we don't tag walls, but I'm sure every one get the point Sletuffe

Accessibility / Usability?

I like the idea behind this proposal: "This is my whole point: as a user of a certain road or path I am only interested wether I can drive on this road or not." What I do not like are the names. If what you are after with with this tag is whether you can use a road with a particular type of vehicle and not how smooth the surface is (this is just a means to indicate whether you can use the road) than I think the name should be rather accessibility or usability. This is also why I proposed to extend the access tag to cover the semantics but it seems people prefer to reserve this tag for legal access restrictions. Something like accessibility=(car|hc|4wd|tractor) would work perfectly for motorized vehicles and is actually the way how it is done in most maps that I know. It has also the benefit of largely resolving the sand problem as well as the discussion about subjective scales (what precisely is the difference between bad and intermediate etc.). Above all the meaning will be clear for the users of the map (the meaning of accessability=HC is much more obvious than smoothness=intermediate). The only problem I see is that things like roller blades only interested in excellently smooth pavement will not be covered anymore. But such "vehicles" are really rather interested in the surface property and should be covered by the surface tag. Ukuester 15:58, 4 July 2008 (UTC)

I agree with you too, what we really need more than "is it made of grass, sand, cobblestone, stones, asphalt, smooth asphalt, asphalt ( that the surface tag can handle) is the usability of a way, and I feel that "smoothness" doesn't serve it well ( anyone comming very fast to this proposal would think it's intended for roller, skate or jogger ). Also I don't want to forget them, it's to my mind a secondary need.
You are saying it would bann "roller blades" why not ?, if I continue your idea, why not use accessibility=roller_blades ?
(by the way what is an HC ? )
I would prefer usability rather than accessibility, and would propose something like :
Proposal usable by minimum supposed surface
usability=roller roller blade/skate board and all below smooth asphalt or equivalent
usability=sport_car sport car/racing bike and all below asphalt or equivalent
usability=car city bike/normal car and all below none
usability=4wd 4wd and all below none
usability=tractor tractors/quads/tanks/mountain motorcycle/Mountain bike none
When reading my table, I might be agreeing with you that the 2 first might well be dropped and returned to the surface tag. But not sure because maybe someone will imagine a different surface type than asphalt where roller could still go.
The only thing I am not happy with in this table is the fact that the tag refers to one and only one vehicule while I'd like to refer to a "group of"
example, my "car" would better, but longer, be "any_vehicule_group_that_can_drive_where_a_car_can_drive"
Usability Usability ! that what counts.
using surface=asphalt and saying rollers can use it will fail when later someone add à surface tag that is different but still usable.
Is it easy to see that I hate the surface tag, don't use it, and don't understand the need for it ?
My point of vue might also be transposed to the way the access tag is used. We should think in terms of "group of vehicule sharing some properties" and "group of surface type sharing some properties" or else a routing program ( isn't it what we want in the end ? ) is unable to take into account all tags and variation that are added all the time to those tags.
HUGH !, this is the friday thought

Sletuffe 18:51, 4 July 2008 (UTC)

HC stands for "high clearance". For most tracks that a passenger car can not navigate you wouldn't need 4WD but simply a sufficient high clearance which some 2WD Jeeps have. The typical hierarchy for unpaved/dirt roads in the US (where this type of road is much more common than in Central Europe is "any passenger car can use" - "requires high clearance" - "requires four wheel drive" (example). I'm fine with using "usability" instead of "accessability" to make a more clear distinction to "access". I like your scale except that I would want to have hc added between car and 4wd. The fact that groups of vehicles are meant instead of particular vehicles should be made clear in the documentation of the tag. I would like to see a comment by Chrischan though. :-) Ukuester 11:58, 7 July 2008 (UTC)

This sounds like a totally separate proposal from Smoothness. You might want to create a new proposal and move this discussion over to it. --Hawke 16:38, 7 July 2008 (UTC)

I disagree. According to the proposal the rationale behind smoothness is "This is my whole point: as a user of a certain road or path I am only interested wether I can drive on this road or not." I tried to make this more explicit than it was in the original proposal. Please explain why you think that this is totally separate. Right now we have a proposal surface, another proposal additional surface values, a third proposal tracktype, and a fourth proposal smoothness which are all closely related. I don't think a fifth proposal usability will make thinks any better. Ukuester 07:34, 8 July 2008 (UTC)
I was mostly referring to Sluteffe's table, which is effectively another proposal. Your "accessibility" tag is also another proposal. --Hawke 06:01, 10 July 2008 (UTC)
I also think this discussion should stay here, because indeed it is what I intended with this proposal. Actually something like this was one of my first thoughts. But then I decided to propose it the way I did, because I think "smoothness" is exactly what I want to describe. With the tagging scheme above, I see two problems:
  • as mentioned above, I would like to address the "smoothness" independent of a particular type of vehicle. Ofcourse it could be mentioned in the documentation that a way for racing bikes should be tagged as "sport_car", but people only into biking would probably not start using this tag. Likewise, people driving in cars would probably not start tagging something as "roller_blade", because they think this is something for the roller blade crowd.
  • More important, we get into conflict with other tags. Many people might imply that they can use a way with "Usability=sport_car" with their sports car, although it might be a road without car access ("access=no"). It would also be very strange to tag a 50cm wide path in the alps with "usability=4wd" (although it definitely can't be used by a 4wd), only because I want to make clear that I can use it with a mountain bike. -Chrischan 20:29, 8 July 2008 (UTC)
The 50cm 4WD path is a point well taken. Unfortunately, you can also turn this argument against your original proposal. How do you tag a 2,50m wide track which has two very smooth drive traces but requires HC because the cross cut profile of the track looks like this: __x^x__ (smooth traces left and right but a something in the middle that would cause every passenger car to hit the ground)? I'm lacking the proper English terms for this but such tracks are not uncommon. If you tag it with smoothness=bad people driving in cars will know they can't use it but people riding city bikes will assume they can't use it either, although they actually might be well able to do so. To me it looks as if usability by bikes and cars (not to speak about other types of vehicles) might actually require different scales. :-/ Ukuester 08:20, 9 July 2008 (UTC)
I agree with what Chrischan says, and disagree that you've made that argument against the original proposal. IMO That would call for another access tag (high_clearance=* or access:high_clearance=* perhaps? I think naming the values after particular modes of transport, such as accessibility=sports_car is nonsensical because of the complications that arise when you see a way tagged like that but which you can't take a sports car down. "bad" doesn't have this problem, and as long as objective standards are used for deciding what a "bad" or "good" smoothness is, it doesn't have the ambiguity that the original post was concerned with. --Hawke 06:01, 10 July 2008 (UTC)
The question the proposed tag is supposed to answer is whether I will be able to use a particular road/track/path with a particular vehicle. Now how do you use smoothness to tag the track described by me above and make clear that it can be used by pretty much any type of bicycle but only by cars that posess high clearance. According to the table on the definition scale its impossible (the reason is that there is no complete ordering among vehicle types such that one vehicle can use all tracks usable by a "larger" vehicle). Of course a car specific tag clearance=* would do the job but the need of an additional tag suggests that the proposal is unable to really answer the question it is supposed to answer. It seems to me that usability by roller skates and racing bikes (and probably trecking bikes) could be handled resonably well by the surface=* as well as the smoothness=* tag. To know whether I can use a footpath with my mountain bike I would need to know the maximum steepness (not covered by neither smoothness=* nore surface=*) and the amount/existence of "steps", big rocks etc. (somewhat covered by smoothness=* but rather not by surface=*). Actually, for mountain bikes some scale denoting the technical difficulty similar to the proposed hiking scale might be usefull. For cars I need to now whether I need 4WD (related to the steepness and the surface) and whether I need HC (related to the existence of clefts, rocks, etc.). Both is not fully covered by neither smoothness=* nor surface=*. To me it seems that all these aspects cannot be squeezed into a single tag because the question of usability is multidimensional. The question now is, whether we are better of with smoothness=* (and some additional tags) or with usability=* with multiple values listing the vehicles able to use a road/path or with splitting all the information about a highway in multiple tags as proposed below. I agree that in any case we should and would not tag a smooth footpath with usability=sportscar and that we would not list all vehicles able to use a particular highway but make reasonable defaults/assumptions (e.g. any highway usable by a passenger car can be used by a trecking bike). Ukuester 10:02, 10 July 2008 (UTC)
  • The argument here seems to be that the proposed smoothness=* isn't enough to tell whether a given way is usable by a car with a certain amount of ground clearance. My proposed solution:
  1. Reword the definition of smoothness=* so people don't think it's the be-all and end-all in usability-tagging. It just says whether a way is smooth enough to drive with a given class of wheeled vehicle.
  2. Propose a separate tag clearance=* or whatever that states how much ground clearance a four-or-more-wheeled vehicle needs to use the given way.
In my opinion, smoothness=* is well-defined as is, providing a nice balance between detail and simplicity. It's a usable tag (I've been using it a lot already), and if there's some special cases it doesn't cover, so be it. Robx 10:53, 14 July 2008 (UTC)
  • Ok then, we have to move forward. I'm ok with the name "smoothness" also I'm still scared about confusion, but a nice description might do the work. ( I also suggest to remove the "surface examples" colum to minimise confusion ).
Ok also not to use "vehicule centric" tags and stay with excellent/good/intermediate/bad/horrible
I also accept that we cannot "as is" deal with left/center/right changing smoothness, but other tags might be created for that purpose
If the global idea is ok for most people here, I think we could move to voting
I however d'like to point some refinement :

Sletuffe 15:53, 15 July 2008 (UTC)

Fair enough. I added a lot of pictures that can be used for illustration (also for the limits of this proposal). Ukuester 18:36, 20 July 2008 (UTC)

Missing vehicules

Also I'm globaly happy with the main table I want to add refinement in the goal of not forgetting some vehicles while keeping actual compatibility.

  • Has Ukuester mentionned a car with High clearance cannot fit in any tags but is able to pass ways where normal car cannot while it is not able to pass where a 4wd can.
  • I also think tractors/quads/Trials are not taken into account and we might loose, in the mountains, special usability ( maybe also contributors from forest managers ) ONF (france )

Here is my proposal :

Proposal usable by :
smoothness=excellent roller blade/skate board and all below
smoothness=good racing bike and all below
smoothness=intermediate city bike/sport cars/wheel chair/Scooter and all below
smoothness=bad trekking bike/normal cars/Rickshaw and all below
smoothness=very_bad Car with high clearance/ Mountain bike without crampons and all below
smoothness=horrible 4wd and all below
smoothness=very_horrible tractor/ATV/tanks/trial/Mountain bike
smoothness=impassable ??? no vehicles ?

Comments :

  • I still have problem understanding what impassable will be used for
  • I think mountain bikes can go where 4wd have to stop ( dispending on pilote's legs ! )

Sletuffe 15:53, 15 July 2008 (UTC)

I guess impassable would mean no vehicles. I added a lot of example pictures (that I would like to see how they would be tagged) and one of them cannot be navigated even on a mountain bike. If you are very good, you might be able to go it down (although it's pretty tough) but certainly not up. Ukuester 18:33, 20 July 2008 (UTC)

Don't try to fit the surface description into one key, use more objective tags

I don't think this tag alone describe the surface better than the other tags for surface and tracktype. This may fit for wheeled vehicles somehow, but why use the subjective keys and don't use directly the size of wheels that works on this surface?

I suggest to use a combination of tags to describe the surface. Some of the tag may even be subjective. The usability tag above is a start. We may extend the surface tag from paved/unpaved to describe better if it's sand, asphalt, concrete, stone, grass.

  • example: An old shabby residential road
 highway=residential, width=5, surface=asphalt, smoothness=coarse, usability=bike, clefts=yes, minwheelsize=25cm
  • example: A narrow track with paved surface in good condition
 highway=track, width=3, surface=paved, smoothness=fine, usability=roller, clefts=no, minwheelsize=5cm

Of course you don't need all those tags for each road. You can assume some defaults. The last example don't need the usability tag, it should be clear from other tags that the surface is usable for skates --Andy 14:20, 9 July 2008 (UTC)

  • Nobody's claiming that smoothness=* alone describes the surface of a way completely. It can be used independently of the current tracktype=*. You're free to propose additional tags for capturing the type of surface more exactly. Robx 10:57, 14 July 2008 (UTC)
  • Ok, see, this is maybe a case where someone missunderstood the smoothness=* tag. smoothness=* is not intended to describe the surface of a way. It's just only about usability.
I however note that your idea about a tag called minwheelsize=* + using width=* could maybe solve our problem
But I am not completly sure that size of weels is enough. Maybe a F1 with big wheels while not be able to drive on a track. Adding the surface would help, but then we would loose the easy use of smoothness=* describing usability in one tag

Sletuffe 16:05, 15 July 2008 (UTC)

Most of the vehicles mentioned in the table above (sports cars, racing bikes, mountain bikes, 4WD cars, ...) have the same wheel size. Thus, I don't think wheel size is going to help much. But your comment "smoothness is not intended to describe the surface of a way. It's just only about usability" highlights that surface is what people think when they read smoothness. :-/ Ukuester 18:24, 19 July 2008 (UTC)

Don't invent too many similar tags

I agree we really need a more general tag describing the surface conditions of the highways like this smoothness.

  • I would preffer, if it would deprecate the tracktype. Would be a mapping reasonable like grade1->good ... grade5->impassable?
I'd like to deprecate tracktype too, but I'am scared there is no mapping at all between smoothness & tracktype and not the same goal in the end.
I don't personnaly understand the usage of tracktype, and it seams to me many are not using it the way it was intended to be used by using it as a scale to describe how hard is it to drive on a track. Usage which should be better covered by the smoothness tag.
If you look closely at the describtion, you'll find grade1 is duplicate of highway=unclassified
and that grade3 to grade5 is very vague with many terms like "A mixture of", "quite compact", "tire marks" compared to "subtle tire marks". All that seams very confusing and subjective to me.
For exemple, should I tag a track with grade4 in winter and then grade5 in summer when grass has grown ?
For me, tracktype is an "in between" surface tag and smoothness tag Sletuffe 11:19, 10 August 2008 (UTC)
  • What about the hiking tag? Shouldn't there be only one tag describing the usabillity for wheeled vehicles and for use by foot?
I don't think so, we shouldn't bother people who don't care about hiking by creating a single global tag Sletuffe 11:19, 10 August 2008 (UTC)
But tracks are quite often used by vehicles and by foot. So you need two different tags for describing the same condition of the same way. I would preferre a mapping of the hinking values to the smoothness and probably extend the smoothness for ways where riding is not possible any more but where we can still make a difference for usage by foot.--De muur 08:41, 11 August 2008 (UTC)
  • And what default value shall be assumed, if no smoothness tag is set? Default value for surface is paved, so it could be either Excellent or Good.--De muur 11:39, 7 August 2008 (UTC)
I suppose something like "bad" for a track, "good" for other highway and "excelent" for cycleway should be enough Sletuffe 11:19, 10 August 2008 (UTC)


Some examples to discuss and illustrate the taggings.

Quite "smooth" can be used by any type of bike. Obviously not wide enough for motorized vehicles.
Wide smooth path
Same path as above a little down the road. Still quite "smooth" but much narrower. Can still be used by any type of bike. Obviously not wide enough for motorized vehicles.
Still same path as above which crosses a meadow now. Not very "smooth" anymore. Could be used by mountain bike (not comfortable though) and probably with HC cars.
Another track covered with grass, but short and thus quite smooth. Can be used with a passenger car and a trekking bike.
Path covered with medium rough gravel (3cm pieces). Maybe ok with Trekking bike and above.
An example of an impassable (except on foot) path. Heavily washed out, exposed roots and rocks. Cannot be navigated uphill even on a mountain bike, but no problem downhill. (With 500cc dirtbike that whould be fun!)
An example of a pretty rough track. Quite eroded, exposed roots and rocks. Can be used with a mountain bike and a high clearance vehicle. Depending on the steepness and the weather conditions the track may additionally require 4WD.
Another example. Not eroded as the one above, simply covered with very rough gravel/rocks. Can be used with a mountain bike and probably even a passenger car, not very comfortably so though.
This track is smooth in the direction of use (can be used by a trekking bike) but has an eroded cross profile. The left, downhill facing side of the track is maybe 30cm lower than the right half. This track requires exceptional high clearance. This becomes particularly obvious if you look at the far end where the track turns right.
The same phenomenon, but a very different type. This track is smooth in the middle and can be used by any type of bike. On the outskirts it is muddy, thus requiring high clearance if navigated with a car.

Another proposal

I think the original proposal has not enough values. And for non-english speaking people it is hard to find the difference. I'm very interested in this probosal, because our car has a very hard suspension, and it would be nice to find a route with good streets. I travel to ukraine sometimes, and there it can be really horrible if you choose a bad route. In germany we have some streets which have a very good asphalt surface, and it looks like an excelent road. But with the car they are very uncomfortable, because the asphalt is on top of a very old cobblestone road. (after reunion of germany they tried to repair the roads in east germany very fast) This roads could be good with skates, but horrible with cars. I think for skates the surface material is an information which is more useful.

Here is my proposal :

Proposal Description:
smoothness=1 Very comfortable road
smoothness=2 In germany this would be a usual road
smoothness=3 In germany it would have a sign "Achtung! Straßenschäden" (Attention! Street-Damage). This road would be a littlebit uncomfortable with car ... In berlin many streets have this sign .. This can also be good cobblestoned roads ..
smoothness=4 For example a usual cobblestoned road. Or a very bad asphalt road. This are usual streets in ukraine. Feels uncomfortable in car.
smoothness=5 For example a very bad cobblestoned road. Feels very uncomfortable in the car.
smoothness=6 Very bad road. In the car you must drive very slow, because the suspension would be damaged.
smoothness=7 You need a 4wd car.
smoothness=8 Not usable by car. But for walking it's ok.
smoothness=9 Only for mountain climbers. :)

Please say what you are thinking about this proposal.. --TEL0000 23:36, 8 August 2008 (UTC)

I don't like the numbers, as they are harder to remember. The "level 8" and "level 9" are pointless, at this point it's not a road (9 is a joke?). It also is very subjective, more so than the original proposal. e.g. "feels uncomfortable" depends a lot on the person and indeed the car. --Hawke 00:18, 9 August 2008 (UTC)
For me only 1 to 7 are important, because i'm driving in car. For me the numbers are more easy to remember, than the english names, because my english is very bad. I agree with you, that it depends a lot on the person, and the car, but i don't know how i can explain the smoothness better. --TEL0000 22:38, 9 August 2008 (UTC)
I don't like numbers either, because we cannot extend it later if we find a need between 2 and 3 ( 2.5 ? ;-) ) 1 and 2 ? are german road uncomfortable ?
for the rest, you want to distuinguish the comfort of a road, then the surface tag is made for that
because smoothness cares about the usability of a road Sletuffe 11:34, 10 August 2008 (UTC)
I agree with you, it is hard to extend it later. So maybe words are really better. But anyway we need more words, than the original proposal. :) The most german roads are comfortable, but in east germany there are still many uncomfortable roads. And in Berlin they can't repair the roads enough fast...
You are right, i was thinking about the comfort of the roads. I thought smoothness is the right tag for it. I think surface is for the material of the surface (cobblestone, concrete, asphalt, dirt). But cobblestones can be comfortable, or very uncomfortable, depending on the quality. The same with concrete or asphalt. --TEL0000 19:31, 10 August 2008 (UTC)

Moving to vote period ?

As no one has raised comment to improve or refute the original proposal updated by my and Ukuester's version. I'd like to propose moving to a vote period. Starting it the 1st september ( to let people come back from holydays :-) ) Sletuffe 12:39, 13 August 2008 (UTC)

Now we have September 18th. Why did you not start the voting? --De muur 18:46, 18 September 2008 (UTC)
because I forgot ;-) ok, let's do it ( as soon as I understand how) Sletuffe 16:07, 20 September 2008 (UTC)

Moved from another page - quantizing status

A fella with bad speling

you know what ? i'ts even worse in my own language ;-) Sletuffe 00:28, 24 August 2008 (UTC)

kindly advised me to move over here, so here I copy myself from there, which would apply to smoothness. The significance of my copy is that (there) I tried to define the meaning of the - abovementioned subjective - terms. We definitely need this, even if it's not consistent to the exact level all over the world, because route planning could put this variable in good use.

This is not a final design, just trying to show how to quantize a bit more general than telling "you can go on this way by 3 wheeled yellow skateboards if you are medium experienced", whatever that should mean. ;-)

state description
excellent (smooth) smooth surface, like newly made highway, cycleway, compact footway, etc.
good the normal worn-and-used state of ways, some bumps and cracks as normal
average (wave) as highway/cycleway: patched and a bit cracked, corners may be degraded, starts to limit maximally possible speed; as track, it have smaller bumps and holes; generally a way which could use some fixing, but still pretty useable
bad (rough, holey) as highway: broken surface, holes, cracks, corners broken, surface sink or raised, wavy; as track it shows definite wear, holes, or other surface problems (mud, water-craved little holes, etc); either case it severely needs fixing up, definitely limits usage by faster vehicles (or fast feet ;))
very bad really broken, like a paved highway with large unpaved parts, or stone road with large dirt/mud patches, big holes, etc; cars cannot go faster than 40Km/h as highway, and as a footway running people may broke their ankles. Generally it's been some official's joke to call this road "highway" or "track". Needs rebuild.

Others tried to mention other "possibilities", but please realise that most of them seems to be designed for tracks (cycleway, footway) and not highways.

If I have repeated old arguments, apologies. If I offended someone, apologies. Generally forgive me for being existant. :-) But comments are welcome nevertheless. --grin 10:54, 23 August 2008 (UTC)

Maybe what you want in the end is, like it was said near the end of : Proposed_features/Smoothness#Another Proposal is to describe the "comfort" of the road. What (I think) we want here, is the "usability" or "passability" of a road. Even if I need to drive at 5km/h to go with my car, that's a possibility my routing program should take into account.
I like this new proposal. But if this page is about the usability or passability, is there any other proposal-page for the comfort of the road? There are so many resembling proposals now, that i don't want to open a new proposal again... --TEL0000 02:33, 24 August 2008 (UTC)
Same issue from a roller-blading point of view. A road with 'excellent' smoothness would be great (a fun place to go to especially for skating perhaps), but then a road with 'good' smoothness might just be more uncomfortable on roller-blades. Seems like just a problem with the descriptions. Maybe 'uncomfortable for type x' bits should be added the description. -- Harry Wood 15:39, 22 September 2008 (UTC)
If we introduce a concept of "uncomfort" then It would become even more subjective, and I don't like that, however your point is interesting, good and excelent are too close in description, and it's quite clear that you can be roller-blading where a racing bike can drive. Maybe we should drop excellent by merging it with good. Sletuffe 21:41, 22 September 2008 (UTC)

In you'r proposal, the advantage I see for routing is to decide between different roads wich is faster, then you'r "maxqspeed" could do the work Sletuffe 00:45, 24 August 2008 (UTC)

Oh, And this should be relative to the material of the road. So if the road's made of dirt (is there a material tag?) or unpaved then "good" means a good unpaved road, even if it's not "good" for a Porsche. --grin 10:56, 23 August 2008 (UTC)

I hope the surface=* tag will evolve to take many other "surface values" into account
'relative to material' Mmmm, sounds a pain for routing algorithms Sletuffe 00:45, 24 August 2008 (UTC)

And the most important may be not obvious: this should work for highway=motorway, highway=primary, highway=secondary, highway=residential, etc. too! --grin 11:02, 23 August 2008 (UTC)

Yes it does ! "Apply to linear" so any way can get the smoothness tag. But a reasonnable defaut for all ways should be choosen to avoid "over tagging"
By the way, sorry for "inserting" my comments inside yours, but it was quite long and I wanted to multi-part comment ;-) Sletuffe 00:46, 24 August 2008 (UTC)

Weather conditions

Reading this one could think that all your tracks are inside a mall. We need a system that considers dry, wet, snowy and frozen ground differences. For instance some kinds of paved ground can be very slippery when wet for inline skates. Other ways turn impassable when muddy. We should not start a new system that does not consider weather conditions. We have enough systems that don't care for the weather. --Lulu-Ann 15:03, 22 September 2008 (UTC)

I don't agree, for now we have NO system that deal with usability at all. So even if that proposal doesn't deal with wet/snowy condition that's still a step in the good direction, we could later increase it with a system like : smoothness:wet=very_bad or smoothness:snowy=impassable. But even then, that would become complicate, will you map every possible snow thickness between 1 and 20 cm ? Sletuffe 21:52, 22 September 2008 (UTC)

Mountain bikes vs. tractors

I'd imagine that a mountain bike is normally less able than a atv or a tractor. Naturally some sporty people might do "parkour on bike" just like they even compete with trial bike, but I wouldn't consider that "usable for travel". Factors limiting travel by tractor and mountain bike are (approximations here, feel free to adjust and add what's missing) Alv 09:08, 4 November 2008 (UTC)

element Tractor ATV bike
Mud surface pressure smaller, "wins" tires sink into mud first, "lose"
High middle strip clearance of > 50 cm, tie clearance of > 30 cm or unaffected because of small width no effect, tie
Consecutive deep holes maximum diameter > 40 cm, winner max d. 30 cm. at over 60 cm they start to become unobstructive "valleys" max d. maybe 40 cm
Fallen trees, roots maximum diameter 30-40 cm, winner max. d. 20-30 cm max d. 20-30 cm
Pointy and/or loose rocks maximum diameter 40 cm, winner max d. 20-30 cm max d. 20-30 cm

early approval and last unilateral modification

This Proposal has been Approved by majority (18:6 on 2008-11-02)

  • Created Key:smoothness
  • Removed very_* because no-one can guess the difference between horrible and very_horrible.
  • Instead of impassable access=no should be used!

--Phobie 15:26, 2 November 2008 (UTC)

Hey ??? how can you take such unilateral decision of that kind without discussing it first on the talk page ?? I'm moving it to the talk page and undoing your changes Sletuffe 17:16, 2 November 2008 (UTC)
Let's talk about it first if you wish :
  • Approval, ok, if no-one is voting on it anymore we could move it to approved features
  • remove very_* : "no-one can guess" and so what ? is the feature page at all guessable ? no. There are descriptions made for that
  • access=no Instead of impassable : well, then you didn't understood this proposal or didn't read it. smoothness is intended to be a physical propertie, while access is a right propertie. nothing forbids a way from being impassable AND access=yes Sletuffe 17:23, 2 November 2008 (UTC)
Other than perhaps common sense. A right of way that is impassable is by definition not a right of way. Chriscf 09:33, 3 November 2008 (UTC)
Other countries might not and don't have a concept of a right of way, in the sense used above. Alv 10:04, 3 November 2008 (UTC)
This is moot. It's generally accepted that a right you cannot exercise is no right at all. access=no for ways that are impassable at all times makes sense. Remember that highway=steps already implies motorcar=no, which equivalent. Chriscf 14:03, 3 November 2008 (UTC)
Well we don't have a such rights at all, we have the freedom to travel, so claiming that anything with highway=* requires a "right of way" is just not valid worldwide. (Here only building a road on others' properties is a right that can be given, bought or acquired.) Alv 09:08, 4 November 2008 (UTC)
Impassable = access=no. End of discussion. Chriscf 11:04, 4 November 2008 (UTC)
(I haven't been pursuing the tag nor the value impassable, my comment was solely on that right-of-way remark...)
As the definition used in the proposal now is and has been, admittable. Probably what made me refrain from voting, was just the implications and the lack of intertwining this with access definitions. Now anything with smoothness=very_bad to very_horrible could imply motorcar=no but it even doesn't... If I were to want to find where access is only physically restricted or difficult but not forbidden (say, to find best places to propose that the city takes measures to improve the connectivity), they'd all be motorcar=no when I'd benefit from some of them being motorcar=obstructed or the others being motorcar=forbidden. Or I wanted to make a routing software for situations justifiable by necessity, I'd need to know what is just forbidden but faster and what is a physical dead end. But it's not (yet) in my interest to pursue such changes. Alv 12:10, 4 November 2008 (UTC)
  • Approval: Normal votes takes 2 weeks and vote-time gets extended if you get too few votes. This vote was set to 12 weeks! After 6 weeks acceptance by the community was more than obvious...
  • Remove very_*: Yes, this was a arbitrary effort to create "key" which is goot for OSM. I will add those senseless values to Key:smoothness if you need that.
  • I read everything here. While the key has been approved, the values impassable and very_* seems not to get a clear approval. Since smoothness is a generic key only minefields and active volcanos are impassible by all kinds of vehicles, horses and passers-by! Things like motorcar=no has always been used for ways which are impassible by cars. It would make more sense to use motorcar=impassible and have a vehicle-specific value than using smoothness for that!
--Phobie 00:48, 4 November 2008 (UTC)
  • no problem for approval for me
  • very_* : yeah, thanks, I'll do it myself
  • As you said earlier, there are 24 votes on the proposal which include very_* and impassable. Don't take a shortcut saying it was not clearly approved while it was. I won't say impassable will be useful, and I personnaly would have dropped it and won't use it, but HERE, we respected a democratic process, and because the guys who added impassable have their reasons I ignore, I respect the process and will keep impassable. Create a proposal deprecation of very_* and impassable if you which, but don't change the content of this proposal after the vote. Sletuffe 09:11, 4 November 2008 (UTC)
I personally didn't even vote on this because I still plan to have a stack of photos to classify by several different surface criteria and only then make up my mind about the relevant factors. But since this is likely to be used, I think that it should be with values that offer reasonable benefits over the old tracktype=*: without the very_* it's impossible to differentiate between
  1. tracks that have gone bad enough that a normal modern day car is likely to scrape it's undercarriage but where an older car is very much usable from tracks where both are usable. Yet getting a rickshaw over the bumps of the worse class can be most difficult.
  2. tracks that are very much usable by an old and slow car with high clearance - or some SUV's - and other's that require a Jeep with raised suspension, off road tires and lockable differentials - say a road to a remote village in the jungle. Tractors are on a par with ATV's but much more able than any non-off-road-equipped car and that difference is easy to spot. Alv 09:08, 4 November 2008 (UTC)
And as to impassable, I think that it would be usable later when a tagging for seasonal variations is agreed upon; say smoothness=horrible + surface=sand + seasonal:spring:surface=mud + seasonal:spring:smoothness=impassable. Alv 09:08, 4 November 2008 (UTC)
Alv ! You enlighted my mind ! I now know for what purpose impassable could be used for : In france we have (a quite special case) a tertiary road that is exactly 1 meter above see level. Due to [tides] this road is usable only 12 hours per day and the 12 other hours it is under 1 meter of water and so unpraticable by any wheeled vehicule. While we don't have the time variation smoothness, I'll certainly tag it with impassable Sletuffe 10:01, 4 November 2008 (UTC)
"Unlikely navigable even by 4wd or a mtb, but might be possible" was the original draft wording for the last value. That'd then be "Normally no wheeled vehicles, but they just might get through depending on conditions." Alv 10:08, 4 November 2008 (UTC)
I can't remember of that wording, that would have helped me much into understanding it. Now I see much it's potential (even still rare) use. I'd like to change the description, but since it as been accepted "as is" (="no vehicules") I'don't know what to do Sletuffe 10:39, 4 November 2008 (UTC)
Democratic? I read OSM was meant to be anarchic! All those proposals are only there to create a best practice guide. But still everyone can use his own tagging scheme. If you really think smoothness=impassible is usable it should be kept. But keeping it because of some wikilogins posting {{yes}}-tags (without comment and without adding anything to the discussion) is not the best idea! I think motorcar=impassable and motorcar=forbidden is a much better idea! --Phobie 20:11, 5 November 2008 (UTC)

elaborate on 4wd

Somebody please elaborate on the meaning of "4wd" as used in the tag definition: a 4wd Audi all-road & high clearance Subaru Forester (4wd) - stock 4wd pickup - stock 4wd Jeep - 4wd Jeep with raised suspension, off-road tires, limited slip differentials etc.? Alv 09:08, 4 November 2008 (UTC)
This is a very clever question ! thanks for asking it ;-) And the problem is not that simple, the definition asks for an "image" we have in our head of a 4wd, and unfortunetly, there are so many on the market, that my basic 2 traction car is sometimes better at going on a track than a 4wd Panda "plastic toy" is. However, since the proposal talks about a scale, we have to assume by reading it that a 4wd is "better" at going on crappy roads than any other transport means listed before. So I would say a 4wd in this proposal is something like that :

[4wd in snow] or [in mud] Sletuffe 09:19, 4 November 2008 (UTC)

These two pictures appear to be the same. :-) Certainly a better name than "4wd" is needed, since this is a specific technical term that refers to the drivetrain of a vehicle, meaning only that all four wheels are powered. This includes standard road vehicles such as the Audi A4 and the Ford Mondeo. Chriscf 09:51, 4 November 2008 (UTC)
Yeah, clearly the same, because in my mind, there is only one type of 4wd, but that's because in french we call a 4x4 a car that is able to drive on very crapy roads. And we almost dropped it's original meaning of having four powered wheels. The Audi A4 is, to my mind not a 4wd while unfortunetly, it is.... I then agree 4wd is unclear... what else could we use ? All terain vehicule ?Sletuffe 10:12, 4 November 2008 (UTC)
"All-terrain vehicle" has a specific definition - a subset of quad bikes. Chriscf 13:59, 4 November 2008 (UTC)
"offroader" --Richard 16:43, 4 November 2008 (UTC)
Wasn't aware of the name, so an "off road vehicule", I'm changing the description ! [1] Sletuffe 16:47, 4 November 2008 (UTC)

Talks about default values

At the beginning of this proposal, their was no default values so, in the final accepted one, there still are no defaults values. However, could some be supposed in many case to ease the pain of mappers ? I would propose :

  • Without a smoothness values,
highway tags are supposed to be tagged smoothness=good
except track that is supposed to be tagged smoothness=bad

Sletuffe 09:41, 4 November 2008 (UTC)

I think default values are a stupid idea, because the tagging in OSM is typically incomplete. Rigth now we have many thousand ways without a smoothness tag. Assuming one single default value for these ways is just senseless.
And when you define a default value, there is no way to differentiate, whether a tag is still missing or whether someone omitted the tag, because it is already covered by the default.
--De muur 11:50, 4 November 2008 (UTC)
That might be a "stupid" idea, but that's allready in use in many case. Either we have to define a new way to tell something is missing, or not mapped (I'm working on it if you want to help : see Proposed features/internal informations between mappers or using defaults until that.
Wether or not we explicitly say there are no defaults, there will be in the developper's mind that makes a navigation software and that's allready the case for other cases. Sletuffe 12:12, 4 November 2008 (UTC)
Without a default value, a missing smoothness tags means, that we do not know the smoothness of the object. And for all the existing items in the data base this is the only correct assumption.
Since in reality each object has a smoothness, there is no mixing between not present and not known for this tag. So you do not need default values and you do not need to define mark for a missing smoothness tag.
--De muur 12:00, 5 November 2008 (UTC)
We are talking about that on the talk@ list, and as I am constructing my mind, I now think that defaults, for any kind of feature is bad juste because, like you said, we cannot tell for sure, wether the mappers decided not to set one because he doesn't know or because he knows the default was enough. I'm ok to drop defaults on this feature, renderers and routing software will decide on themself Sletuffe 14:50, 5 November 2008 (UTC)

Next step : deprecation of tracktype ?

I fear there's gone a be a lot of opposition, that's why we didn't mentionned that in this proposal, but that's the next step I think we need to take. The tracktype tag has many problems I mentionned on the surface and tracktype page, and I think people are notmiss-using it because it's not clear and not efficient enough. Only future will tell, but I have great hope that people will find the smoothness advantage over tracktype. So I'm going to start a separte page ont this idea of deprecating it Sletuffe 11:15, 4 November 2008 (UTC)

When you have objective criteria and a less stupid value set. Not before. Chriscf 11:30, 4 November 2008 (UTC)
Usage will tell, I didn't say I'm going to deprecate tracktype, I said I'm going to propose to, by arguing about tracktype being worse than smoothness in many regards. Nothing is ever perfect, but it's clear to me that it's now a bit better Sletuffe 12:06, 4 November 2008 (UTC)
Have you thought of better tag values yet? It remains the case that we cannot have the values listed here in Map Features. They're subjective, and don't tell people anything useful. Most importantly, new mappers can't learn anything from it. They see a road tagged with "name" and "ref" tags, they can guess what they need to tag in their own ways - they can't do that with this. Any scheme for this property must be intuitive. Chriscf 17:35, 20 November 2008 (UTC)
"We" most definitively can have, it's not in your power to tell those bastards that voted on this that they cannot have it. Alv 18:02, 20 November 2008 (UTC)
It clearly is in my power, because I just did it. The existence of a vote does not bind us to stupid outcomes. Chriscf 12:25, 21 November 2008 (UTC)
Just like walking on a motorway is in your power. Likewise nothing's stopping you, but numerous others will remove you from there as numerous others will keep reverting your reverts. Alv 14:56, 21 November 2008 (UTC)
Usage will tell if people find the Tag useful, having them in the Map Features is just one step towards wider usage. Concerning the subjectivity: Unless someone proposes better values and definitions, I find it a rather useful compromise. It's just pretty hard to define an objective tag describing the surface quality, since judging the surface quality is already subjective. Or how could you do it, aside from counting every bump? --Driver2 20:07, 20 November 2008 (UTC)
I must admit you are quite right, yeah those values are not quite intuitive, but we have all search for, and no one found any better. Because, probably, like Driver2 said, it is a bit subjective, but just a bit : If you are using rollers, then no chance for you to drive on a mountain track. But any how, we have come to a point that we thought (at least those who voted yes) that having it was better than not. And in regard to tracktype=* gradeX are not much intuitive, but more than that, to me at least, the description is not intuitive either. And guys from the surface=* have tried hard to cover all possible case of the nature of a track or road without being able to answer what many people wants to know : "Can I go there with my car". I don't say smoothness=* answers this perfectly, because where someone would stop by fear of damaging his car, some other, with the exact same car, will try and perhaps manage to do it. I would end by saying that not only does a tag can't solve it, but not any explanation of anyone will either, so we try to keep in this tag the maximum description to what people await from a "passable by passenger car" while still knowing that so many other indescriptible factors will probably make it fail. Sletuffe 01:25, 21 November 2008 (UTC)
"Least worst" is hardly a justification. If you can't define the values for this tag objectively, we can't map it. Mapping is a process of recording facts, after all.
I didn't say we haven't define this tag objectively, neither I said it was subjective, this is not binary, I would say this tag is "quite objective" or, if you prefer "a bit subjective". If I take a track with very_horrible example for instance, this is near 100% objective to say that my racing bike cannot drive on it, and that is an usefull information. The description is not 100% objective in smoothess, because we can't extand it to cover every vehicules in the world. But If I follow your "Mapping is a process of recording facts", then I suppose you agree to deprecate tracktype right ? Many tag in osm have always a bit of subjectivity, tertiary VS secondary ? width ( did you take a rule to measure it ? ) and many others, so mapping is not, for me, of recording only 100% objective facts, because 90% of objectivity (even if figures cannot be applied to objectivity) is sometimes good enought to get an information from something. Sletuffe 13:46, 21 November 2008 (UTC)
We have ca. 300.000 tracks in the OSM data base of Europe and ca. 200.000 times the tag tracktype is used. I think this is quite a good quota, and I don't expect smoothness to reach such a level in the foreseable future.
tracktype, smoothness and surface all describe different factors of the same way. They are somehow linked to each other, but no one can substitude for the other ones.
--De muur 12:12, 4 November 2008 (UTC)
Maybe not in the near futur, but I bet a bear with you it will ;-))) Sletuffe 12:17, 4 November 2008 (UTC)
I predict that well see lot's of values besides the ones listed: entermediate, great, smooth, horible, horryble, verysmooth, verybad, veryexcellent, bad-, bumpy, gravel, ... But well see. Alv 12:22, 4 November 2008 (UTC)
Thats the case with every single Tag out there. There are always a handful values being used that are not documented. Either because people think more values are needed, just mix them up with something else or simply spell them wrong. Presets in popular editors can prevent many of these. --Driver2 20:07, 20 November 2008 (UTC)
You are not the first person who proposed a tag, saying nothing about deprecating existing similar tags or minimizing the risks of conflicts or confusion, got a positive vote from 15 people and try to impose the result to thousands users. The same happens with highway=path supporters which are now trying to deprecate footway, cycleway and bridleway. But when such activism leaves the wiki and comes to a wider public like the ML and renderers developers, the opposition is much stronger. So, please, leave existings tags, watch the statistics and wait 6 months. If your tag is really better, it will be voted by its popularity. -- Pieren 10:27, 21 November 2008 (UTC)
I have changed my mind about writing a proposal to depracate tractype, I think it would deserve smoothness as I won't be objective about it, and it will look just too much a propaganda. So I'll let usage make it's away. But still showing, whenever someone need it, that it exists. However usage is not only a matter of quality, If no one knows something exists, it won't be used, even if better ; so hiding it like chrisfc tries is not IMHO a solution either. Whenever a newcommer want to tag something he will use what is in the JOSM presets or will look at the map features. And then, he also watch at the renderers to see if his work is used. And wether they want it or not, as I said on the talk list, developers (of renderers and editors) have the power to influence mappers and they might be tempted to do so based on their own belief, and I belive that is bad. Yes it's a last security against bad tags, but refusing the wiki approach or any other means of communication is just somehow considering that their ideas are better than other's. I'll stop here or else I might well put another page of text that is outside the scope of this talk page. Ugh ! yeah, it's friday ! Sletuffe 13:04, 21 November 2008 (UTC)
As of the "path" example you gave, I have first voted no to it because It would mean I have to change my way of tagging and I am lazy. Then I came across many case where "path" was much better, then I revert my vote to yes and still no to deprecation. And after 6 month of usage, and at last mapnik of osm rendering it, I will now revert my vote and say yes to "path" and yes to deprecation. Time helps things. Sletuffe 13:15, 21 November 2008 (UTC)

I would support deprecation of tracktype. I think its values were extremely poorly thought out. Chriscf has concerns that the values for smoothness are too subjective, but tracktype is far worse in this regard. --Hawke 13:49, 21 November 2008 (UTC)

Which is not a valid argument for this scale. We can't include this on Map Features - it's a really, really, poorly thought-out idea, and if it goes on Map Features people will start using it without understanding it. So, for now, it will be hidden on Map Features until such time as someone sorts the damn thing out. This is not a voting process, it is an approval process, of which voting is one part. It's also not binding until there is empirical evidence that it's a good idea. People not knowing and not tagging is infinitely superior than people not understanding and tagging inaccurately. Valid concerns were raised, which were ignored and not addressed. Chriscf 15:10, 21 November 2008 (UTC)
Looks like we are going nowhere, you talk about "Valid concerns were raised" but which ones ? have you read what I said ? can you tell me what's wrong with my point of vew ? have you anything else in your bag that "stupid tags, stupid ideas, poorly" ? Sletuffe 15:20, 21 November 2008 (UTC)
It's not a valid argument for this scale, but it's a valid argument for deprecation of tracktype, which is what this section is about. Your valid concerns were addressed:
If you had other valid concerns, you did not raise them. --Hawke 16:08, 21 November 2008 (UTC)
[2] [3] [4] [5] [6] [7] [8] [9] Chriscf 09:45, 24 November 2008 (UTC)
You are also citing me and my valid concerns, but did you read it? I talked about anarchy, the philosophy of doing good things without having laws. The editwar you started on Template:Map_Features:smoothness is not one of those good things! --Phobie 04:10, 29 November 2008 (UTC)
Heh. Almost none of those are objections. The actual objections are here [10]. They consist of:
  • It would be better to improve tracktype=* than use this: If tracktype=* is improved to the point where it covers what this does, we can drop this one and use it instead. Of course, that'd entail finding another replacement for (since tracktype=* is bad just because it only covers highway=track and not other highway=* tags.
  • It doesn't cover varying weather conditions. It's not meant to, and would complicate things excessively to try. Basically, this is not the place for that.
  • All highway=* should be smoothness=excellent, everything else can be inferred from "gradetype tracks", whatever that is. No idea what "gradetype tracks" are, but it's clearly the case that not all highways=* are suitable for, say, inline skates.
  • Here's your objection, chriscf (and the only one you linked to in your list which was actually an objection): "the lack of criteria for differentiating between the different classes". Sorry, but the criteria are pretty clear: "usable by (vehicle type)" where usability is defined at smoothness=*.
Then there are a couple of opposing votes with no reasons given. None of them are showstoppers, and nothing has happened to give you (or anyone) veto power over the tag approval process. So please, just stop. --Hawke 23:54, 29 November 2008 (UTC)
You evidently understand neither the objections nor the process. Chriscf 17:35, 1 December 2008 (UTC)
I understand both. It is clearly you who doesn't understand the objections, when the people who made them (see Phobie, above) object to you citing them! The only serious objection in the list that you cited is subjectivity. That one is a matter of opinion, not a clear-cut problem. In any case, repeatedly hiding a feature which was approved in accordance with the process is not a part of the process. --Hawke 19:32, 1 December 2008 (UTC)
If you keep talking about "approved in accordance with the process", you evidently do not understand the process. For one, the official process does not depend on a majority vote. I'm not prepared to repeat myself again, so you can expect me to stop removing broken features from Map Features when idiots stop adding them, and not before. Chriscf 09:36, 2 December 2008 (UTC)
The Approval process says "8 unanimous approval votes or 15 total votes with a majority approval". No part of the process is "Chriscf approves it". You fail at understanding the process. --Hawke 21:09, 2 December 2008 (UTC)
Hawke, my dear, I am so sad to see you spend your time in vain against such a stubborn guy. We'd better spend our time on new proposal that are very hard to solve, such as the new access namespace you proposed. He is pointless, without arguments, without new ideas. Let's keep undoing his changes until some administrator takes the approprivate measures. After reading much on the mailing list (on which he didn't want to take part), people raised rather good comments (not too much objections) on which you answered perfectly well. I start thinking that if people want to tag perfectly and not subjectively a track or ice_ring, or else, they do as they want, they can propose if they wish 20 additionnals tags to smoothness to try to reach perfection. The result is obvious, noone will (except chriscf here present, and 5 other brave guys ) use it. In the end, smoothness will impose itself as beeing better, not perfect, but better to mappers.
We are on the bright side of the force Sletuffe 16:58, 2 December 2008 (UTC)

Confucius say: man who cannot spell "deprecation" should not propose it. --Richard 10:02, 24 November 2008 (UTC)

No Confucius says: Man who cannot spell "deprecation" could probably spell it in his mother language. Let's change the default speech of the wiki to Spanish! --Phobie 03:45, 29 November 2008 (UTC)

Limit Smoothness to road vehicles?

Let me give some use cases to say why I don't think smoothness is usable for other means of moving. Considering a snowmobile. Any way if there is no snow will be unusable. If there is 1m of snow, most places will become passable, however for some places maybe there is a need for 2-3m of snow until it's possible to go on a snowmobile. Clearly, for a snowmobile this specification would not work at all. For snowmobiles other factors would become important, like snowdepth, temperature, avalanche danger, etc.... For a skier it's not so drastic but also clearly unusable. I think for both cases there is clear agreement that another classification must be setup. --Extremecarver 00:58, 20 November 2008 (UTC)

smoothness is clearly not for snowmobile, ski or hiking, the proposal talks about wheeled vehicule only and if you need it, create a new proposal for them Sletuffe 14:28, 20 November 2008 (UTC)
I don't need them (yet), but just wanted to give it as an example.--Extremecarver 14:57, 20 November 2008 (UTC)

So let's come to mountainbiking that is included in the proposal for now, but had voices in the talk page above about inability to transmit all important information. The biggest differentiation must be made between uphill and downhill. While on uphill my technical riding level will matter to a certain extent, this extent is not as big at all as for going downhill. Therefore there are systems setup for mountainbikers, notably the IMBA in the english speeking world, and the singletrack skala in the German speeking world, maybe other language regions usually use even other classification systems, using such a classification, a much better differentiation can be achieved and will be asked for by participants of the sport. I see this as a chance for OSM to be able to cater for a wide range of different expectations, not only people who look for a streetmap.--Extremecarver 00:58, 20 November 2008 (UTC)

I clearly agree with you here, the mtb part of smoothness might not be well suited, there are talks somewhere I'll move to help creating a mtb scale Sletuffe 14:28, 20 November 2008 (UTC)

I would not see the key smoothness as bad, but I think diversification should be held up more than simplification. In the end there exists no acces=no or impassable, it's just about the means you need to use the way. If many sports develop their own tagging scheme, after a while renderers will be able to see the similarities and use it for the better. There is a myriad of different tags related only to cardriving, other means of transport shall describe the ways they need for their needs with the same scrutiny. I support the smoothness approach for vehicles with 4 wheels, but I don't think it will be successful, because the needs are to different to put into one bag.--Extremecarver 00:58, 20 November 2008 (UTC)

I clearly agree once more, we "could" depracate the mtb part of smoothness and stay with motorized vehicules or with 4 wheeled vehicules. and create alternative scale for any sports
BUT ! I don't necessarly see it bad to include a thought in smoothness about mtb because, think of that : who will take care of setting tags for mountain bike only ? well, easy answer : mtb users only. And that is some loss of information : I am personnaly tagging with smothness in the goal of tagging for 4 wheeled vehicules and I don't care that much about mtb (well, at list for now, 'cause I have just bought one ;-) )
The final result you are expecting by removing the mtb part of smoothness will probably end with the opposite : less informations added about mtb by people who don't care much about mtb.
smoothness=horrible or very_horrible is a very poor information for mtb, I admit that, but it might be added by people who don't care about mtb and in such, give information to mtb users.
If you want more precision, there is no problems in adding a mtb scale, but fewer people will contribute to that scale than to the smoothness tag, keep this in mind. So, in the end, I don't see problems in smoothness being somehow a sub-set of the mtb_scale but still independant. Sletuffe 14:28, 20 November 2008 (UTC)
I can accept that point of view, as long it is made clear on the front page, that for hiking, mtbiking etc there too exists more specific tags. For a non mtbiker the picture that a way is usable with a mtb, quad etc, might help in deciding which smoothness grade to give.--Extremecarver 14:57, 20 November 2008 (UTC)

To my mind, the smoothness of a way would have to be set with reference to the primary traffic type of the way - so for a highway=primary,secondary,residential et al, the reference point would be a car. For a cycle track, the reference point would have to be a bike. Ditto for a footway (person) and a bridleway (horse).

That would substantially reduce the list of possibilities when tagging the smoothness. If you are tagging a highway=tertiary the options pretty much come down to 'smooth' (all standard vehicles), 'bumpy' or 'rough/4WD' (or your choice of term for a proper 4WD, by which I mean a high clearance vehicle supporting either centre diff lock and/or low range gears). Pretty much every other possibility in between is really covered more by the surface tag (grass/cobbles implies you might need a decent clearance, sand implies smooth but still really only for 4WDs).

Mmm, your "pretty much come down to" looks a bit too restricted to me, what, if, in my country, there is a very smooth like glass tertiary road ? do you expect me to tag it with surface ? how do I say roller would be more than happy to roll on it ?
What is there to gain to drop to just 3 values ? ( use smoothnes++ smoothness-- or smoothness== ? ) appart from lazyness ? and missing the capture of some rare case ? Sletuffe 12:58, 27 November 2008 (UTC)
It is deliberately restricted to a small number, because I feel the fewer the number of options, the less subjectivity involved. You could have seven levels like the current proposal, but then you're basically back to square one with regards to subjectivity. A "smooth like glass" road and a very smooth road (but perhaps with a higher stone vs tarmac content) are in real use identical for the purposes of a passenger car/track/4WD. Your roller blade requirement is a secondary concern, as roller blades are NOT the primary use of highway=secondary. If you want to tag the way with reference to smoothness for a roller blade, then use something like smoothness:rollerblade=smooth. That way the primary vehicle type is applied for the smoothness tag, and other vehicle types (rollerblades, bikes, skateboards, segways and snowmobiles) can be tagged sparately as a relevant smoothness. -- gaffa 14:03, 27 November 2008 (AEST)

For a bike path, then you might have 'smooth', 'bumpy' (ok, it's a bad word for it) and 'rough'. Same words, different meaning, but correctly referenced to the track type it refers to. Smooth works well for road cycles, bumpy means your probably need an MTB, and rough means you actually have to be competent on the MTB.

Again, same for walking track - 'smooth', 'bumpy' and 'rough'. It's all still valid when you are walking, and again the surface tag (rocks, grass, marsh) adds the additional meaning, while still allowing the individual smoothness tags to actually be meaningful withoput necessarily having the surface tag.

Trying to come up with a single list of values for smoothness grades for all track types and all modes of travel in one set makes it pretty complex. I agreed with the track type deprecation - the tag values are meaningless. As a fairly competent 4WD driver, what I regard as a grade 3 track, others might regard a grade 5 (or vice versa). Personally, I think most drivers could discern the difference between smooth, bumpy and rough (or whatever terms, but with as smaller set of options as possible).

Given that the majority of maps I have have two "grades" - 'normal' and '4WD', with the surface type as a separate value, I reckon we could get away with a set of three suitably named values for smoothness, as long as the reference is the to the primary traffic type. --gaffa 23:15, 27 November 2008 (AEST)

Overall, I don't get you tagging idea, it looks quite well but I don't see... HOW ? can you drop a real life example ? and the tags you expect ? Sletuffe 12:58, 27 November 2008 (UTC)
I think the tags need to be both limited and easily to differentiate between. They should also definitely ONLY be in relation to the primary vehicle type designated for that way. Use sub tags (smoothness:skateboard=bumpy) to handle alternative vehicle types (and that way when someone says "but what about my tractor? It can go on impassable roads", it's easily to say, "well tag the bloody thing with "smoothness:tractor=whatever". Everybody is happy - the standard smoothness value represents the smoothness for the default vehicle, but rollerbladers, skaters, snowmobilers and tractor drivers can all have their own set.
In terms of real world examples, using the photos posted further up this talk page, I would tag them as (in the order that they appear): smooth, smooth, bumpy, smooth, bumpy, rough, rough, bumpy, rough, rough - personal preference obviously. The reason for those values is that from the photos, that is the kind of ride in a car you would get from those roads. The surface (grass, dirt, stones) would have an additional impact, but the combination of the two would allow for reasonable determination by a user as to whether that road was suitable for whatever vehicle they were currently driving. -- gaffa 14:03, 27 November 2008 (AEST)

Lack of criteria

The value of this proposal was in that no objective criteria is needed for tagging and the that the value does not tell anything of the criteria used. It is sufficient for common navigating users, a traveler in any of the mentioned classes, that the mapper has thought that the way is usable by "class N and all below". Assessing the suitability for those mentioned in the proposal seems hardly a challenge. To me that means that detailed criteria is then not needed.

Deducing any (exact) physical qualities of the road is impossible from the smoothness=* and never was meant to be possible, as a track can be unsuitable for sports cars for any of many reasons: high median (undercarriage damage), sharp humps (front or rear skirt damage), potholes (wheel damage) or other reasons. If such deductions are wanted, that would have to be some another tag. Alv 15:45, 21 November 2008 (UTC)

How precisely do you define "passable" or "usable"? There are certainly roads near me which are "passable" and "usable" by car, but that I wouldn't advocate actually using. According to the scale given here, it would be anywhere between "intermediate" and "very horrible". You could certainly get a wheelchair down it, provided it has brakes, and it stuck to the pavement. Chriscf 16:39, 21 November 2008 (UTC)
I read it to mean something along the lines of: If you'd say to a person asking directions "you can drive that road" or "I'd drive that road if I were you". A skilled or upset rollerblader can get forward on a cobblestone street but it's not usable, a racing bike rider can slow to a crawl on a non-usable street. Any user might get through on a road of a worse smoothness than indicated, but it's not fast, likely, guaranteed nor fast.Alv 17:33, 21 November 2008 (UTC)

Trying to stop an Edit war 2nd round

Looks like me (and apprently now some other) and User:Chriscf do not agree about the smoothness=* page, after 3 successive undo, I think it's time for help from others third party. My arguments :

  • That key was voted on
  • That template is shown on the map feature and I don't see problems
  • Might the result be to remove it from the Map Features, there are no reasons to comment it out completly on this page

Waiting for his arguments : ( Edit , they are all up here )


Chriscf is in the wrong: smoothness=* applies only to wheeled vehicles. access=no means that no one at all is allowed to use it, even ped- or equestrians. --Hawke 16:22, 20 November 2008 (UTC)
Looks like, my dear Hawke, that this is what you and I think. But not him. We are now at a point where, after I have undone his commented out twice and that you seams to have commented out his stuff n times we are on a dead end.
  • Either we let him win and all that is gone, and we'll go to it later (say month) once the tag is much used, but I does not appear to me as democratic.
  • Either he stops, but it looks unlikely to happen
  • Or we let that commented out for now, ask support from others, let him ask support for others and cast for a new vote on wish we all agree that the issue will determin the final result we choose.
Sletuffe 17:28, 21 November 2008 (UTC)
IMHO Users not liking smoothness as a tag shall start a proposal for deletion of smoothness and depreciation against something else. If there is then a positive vote smoothness has to go. Until that point however smoothness has to stay in map:features and is the officially proposed way of tagging. If he can't accept that than we should ask for his account to be deleted and his IP blocked (that will temporarily help) as well as asking some moderators on the talk list to block tag|smoothness on the map:features page.--Extremecarver 17:45, 21 November 2008 (UTC)

The vote is irrelevant. Valid concerns were raised, and not addressed, so the vote is a rogue result. There is nothing anywhere which says we have to accept the result of a vote, especially when it contradicts the result of any discussion. This tag is broken, and nobody has volunteered any means of fixing it, so until it is, it cannot be shown on Map Features, otherwise people will use it, which would be a Bad Thing. Since there is nothing which defines whether a way is "passable", users have no way of figuring out how to tag something. Your definition of whether something is "passable" will not always match everyone else's. Of course, the easy way to end this is to stop adding broken tags to Map Features. Chriscf 09:06, 24 November 2008 (UTC)

Come on Chris! If you disagree, open another vote to have it removed instead of this stupid edit war. --JLS 16:24, 25 November 2008 (UTC)
Please feel free to describe these non-existant "valid concerns" that were raised. The concerns you raised were:
* Impassable = access=no -> This is not correct, as smoothness is only relevant to wheeled vehicles, and not the legal access restrictions.
* The term "4wd" is not good. This has been changed.
So, please describe how this tag is broken beyond "Chriscf doesn't like it". Or at least point to some actual valid concerns that weren't addressed. --Hawke 21:52, 26 November 2008 (UTC)
He posted some concerns at Next step : deprecation of tracktype ? --Phobie 04:22, 29 November 2008 (UTC)

Please try to find an agreement on how to find a solution on this edit war. Use the mailing list where I raised a thread about your conflict which happens since 7 days now (title is "Edit war on the wiki "map features"". -- Pieren 16:31, 25 November 2008 (UTC)

Be constructive! Try to find good alternatives to smoothness!

first alternative

After I mentioned the edit-war and read all comments, I came to the conclusion that we need measurable values for smoothness.
Smoothness combines bunch of attributes which can also be tagged separately.
From all those values a user can estimate the smoothness for his vehicle and personal capabilities.
If a mapper wants to give a general hint he would use smoothness and if he wants to be more specific we offer him the keys needed.

Now lets try to find all attributes used by smoothness.

  • width=max:* (in meters)
  • height=max:* (in meters)
  • weight=max:*;min:* (in kilogram)
  • surface:plasticity=*;wet:* (how to get useful values?)
  • surface:frictional_resistance=*;wet:* (how to get useful values?)
  • surface:particle_size=0,* (in meters, yes always stick to SI!)
  • surface:glyph_depth=0,* (in meters, results into a minimum ground clearance)
  • surface:centre_distance=max:*;min:* (in meters, means distance between axis)
  • surface:incline=* (in meter rise per 100 meter horizontal)
  • surface:defects=* (yes/no/refitted)

Please improve my proposal or make a better one! I only want to see constructive comments! --Phobie 05:53, 29 November 2008 (UTC)

Please remove mine if you don't find it constructive : This is a solution I have thought of but abandoned since it won't we used by anyone because too complicate.
You could add almost-infinite number of things to describe a track (see my mail on the list recently) :
  • Max depth of holes on the track/road (it will express the chance you have to hit the ground)
  • Min lenght between holes (it express the chance to hit ground on resonance speed.
  • Roots mean and max size
  • max steepness
  • Possible water stream across the way
  • Autumn foliage on the ground
  • Mud's max depth in water conditions Sletuffe 12:41, 29 November 2008 (UTC)
I read your mail. Some people might find those complex keys useful. We should just offer a default way on how to tag them. It is perfectly ok for me to see a way tagged with smoothness=3 surface=asphalt incline=20 weight=max:5000 surface:defects=yes but smoothness without other surface description keys will be unsatisfactory for many mappers. Mappers decide which keys to tag and coders decide which keys to implement. --Phobie 06:30, 30 November 2008 (UTC)
Doh ! I mis-read you then, I am in no way against overtagging, I think that if some people needs it, they can propose it and I'll vote yes on it. But I am not in favor of replacing smoothness by ~10 tags or so as it would lead to not tagging. I have no problem for example to use (rather long list) surface=* in conjonction with smoothness=* but I am against to say it has to be used with. Sletuffe 11:24, 30 November 2008 (UTC)

second alternative

Since it is hard to impossible to fit the different vehicles into the 7 (8) categories of this smoothness it would be better to set the usability by vehicle or by vehicle group.
I would name that key severity and let it have much less than 8 categories.
If there are many categories it would be hard to distinguish between them.

  • 0 effortless/easy
  • 1 intermediate
  • 2 hard
  • 3 unusable (by average users)


  • severity=default:1;four-wheeled:3;pedestrian:0

unusable is sometimes better than access=no because bicycle=no implies mtb=no and severity=bicycle:3 only conveys severity=mtb:2
See Proposed_features/Any_moving_thing_grouping_system for some ideas about how to group-objects. --Phobie 07:45, 30 November 2008 (UTC)

Option 23,754

OK, another variation on the theme for recording road surface/smoothness/tracktype. I'm trying to cut out the subjectivity issues as much as possible. This approach doesn't solve the variable vehicle type issues (although I'm leaning towards some sort of access/usable tag such as usable:skate_inline=yes/difficult/no for anything that isn;t the primary vehicle type for that road/path etc.

So here goes. Basic theory is to define the surface condition and surface type separately (given they are distinct things). Using the existing surface=* as a starting point, define the surface as paved/unpaved (no change from current usage).

When required/known, add a surface_condition=* and a surface_type=* to indicate what kind of surface it actually is. For surface=paved, one basically assumes surface_type=asphalt.

Possible values for surface_condition=*:

Key Applies To Description Notes
maintained surface=paved/unpaved surface is regularly maintained Maintianed by local authority. Minor defects in asphalt are regularly fixed, dirt road are graded, grass is cut etc.
uneven surface=paved/unpaved surface occasionally maintained potholes are roughly filled in, surface is "bumpy", or contains minor potholes and other general unevenness
degraded surface=paved/unpaved surface not maintained Numerous potholes, areas where the asphalt has washed away and other major problems with the road
corrugated surface=unpaved corrugations are present Generally found on dirt roads that haven't been maintained, corrugations are a distinct surface type that slows a vehicle down (and can shake all manner of things loose in a car)
rocky surface=unpaved road base is rock rocky roads often have substantial uneveness, and can be unsuitable in passenger cars due to clearance. Generally requires careful navigation to drive on
rutted surface=unpaved road contains ruts rutted roads often have distinct wheel or washout grooves, and can be unsuitable in passenger cars due to clearance. Generally requires careful navigation to drive on
deep_rutted surface=unpaved road contains deep ruts Deeply rutted roads are generally only suitable for high clearance 4WD vehicles. Requires careful navigation to drive on

Once you've nailed down the condition or quality of the road, you can move on to surface type.

Obviously (before anyone jumps up and down), surface_type=* could be absorbed into surface=* very easily, although what the surface_type=unpaved values get converted to is a matter of debate)

Possible values for surface_type=*:

Key Description Notes
dirt Dirt, small stones This would be the obvious candidate for transferring surface_type=unpaved to
sand Sand, or fine surface dirt
cobblestones Cobblestones and other similar surfaces
asphalt AKA tarmac, macadam Kind of irrelevant given surface_type=paved, but here for completeness
concrete Concrete based roads Would also allow ytou to tag reinforced_concrete, giving the tank drivers amongst us some useful routing information.
grass Assumes the grass length is still short enough to be navigable
bulldust very, very fine sand (think talcum powder) Very different from sand, bulldust can be difficult to drive through, as the vehicle tends to sink into the harder surface 6-12 inches below, without any idea what is below the surface. This is a fairly explicit Australian tag requirement

Lastly, one thing that no one has really discussed (that I've found) is the variability of the road surface over the length of the road. For short roads, it's not a huge issue as the surface type and state tend to be the same for the full length of the road, but for longer roads, it's possible, and in Australia, common, for the road to change surface types and states quite regularly. It's obviously possible to break the road up into sections with a different surface type and state in each section, but that might mean having multiple 100m sections on a road that is kilometres long.

So, I would add a best and worst qualifier for roads where the state changes: surface_state:best=uneven and surface_state:worst=rutted. -- Gaffa 10:36, 3 December 2008 (UTC)

Made some changes: Changed the surface_state to be surface_condition, which is a better tag (thanks Phobie). Also added concrete as a surface type (for the tank drivers out there). Lastly, I change the surface_condition from "potholed" to "degraded". Still not sold on "degraded" as a term though (but it's slightly better than potholed) -- Gaffa 12:17, 4 December 2008 (UTC)
I started a (yet unfinished) proposal draft some time ago which mostly looks like your idea!
See Proposed_features/surface_unification!
The keys looks pretty good but I would rename them:
  • surface:type to surface:material, because type can be easily misinterpreted.
  • surface:state to surface:quality or surface:condition
There seems to be no problem if you split all ways into 100m segments, so the best/worst is not needed.
Perhaps you could help me to bring my draft in a usable form...
--Phobie 13:09, 3 December 2008 (UTC)
That looks quite good. We definitely have to add pictures to it, otherwise non native English speakers will have problems understanding. Also I think there needs to be tags for tracks and ways that are left open and dissappering as nature fights back their terrain. I would add another category above maintained for asphalt, something like very fine granularity. There are roads that may be well maintained but for inlineskates or skateboards they may nevertheless be not optimal, because the granularity of the surface is not fine enough. On the other hand some potholes don't matter. This proposal should also be used for sidewalks (often sidewalks are differing in quality from the street).--Extremecarver 14:41, 3 December 2008 (UTC)
All this looks full of promises, sound very hard to find all tags needed, but still looks interesting. I strongly suggest you to start a new proposal about this, with a first set (like this one) list of tags envolved or proposed. Anyhow, I'm scared that overtagging will scare some people, but it's worth trying. In the end, I doublt it could replace smoothness, not because it is worst, (it allready look more objective and precise) but because mappers do need different schema of complexity to tag things, I think both can live together without problem, at a low cost of redondancy. It will probably just end like footway=* against path=*+foot=designated+bicycle=no (something like a shortcut)Sletuffe
Added most of your proposal into Proposed_features/surface_unification --Phobie 14:27, 5 December 2008 (UTC)
No worries -- Gaffa 00:09, 6 December 2008 (UTC)

What about opening our eyes on what other have done ?

Instead of just fighting in the dark, I propose we find how professional map makers have solved the case, year and years of experience might well give us a clue.

IGN, the french map maker, with dozen years of experience, which map mountains, roads, tracks, footpath, etc. has a legend wich is : Ign track legend.jpeg

  • Regularly maintained is either paved and well suited for passenger car.
  • Not regularly maintained is rarely paved, or quite bumpy mostly suited for passenger car at low speed. But might also not be used without car dammage
  • Cart track is definetly not usable by passenger car, for an off road vehicle, it might or might not be usable. But for agricultural use or forest management it is usable by tractors or special 4 wheeled equipement.

Sletuffe 13:16, 29 November 2008 (UTC)

I don't believe that IGN puts all values of their map-database on the legend. --Phobie 06:05, 30 November 2008 (UTC)


Planograph wiki.svg

see also

cc by

Maybe in two years OSM mappers will have laser planographs on their cars.

Breaking a system

This tag is actually breaking a system.

Usually the smoothness tag will follow the surface tag (I admit that it might exist exemptions) and therefor absolutely not necessary to tag. I have heard one reason for adding this tag was that routing software only took surface paved/unpaved into account when panning a route, but adding another tag doesn't solve that. There are more than enough values of surface to actually tag whichever smoothness level you like, you only need your planning software to calculate with more values than paved/unpaved. Besides, if your road have a lower quality than indicated by your surface tag, why not add a caution or hazard indication? Especially when you have a high quality road (trunk+asphalt) with loads of potholes and damages. As I see it, the purpose of the smoothness tag is only to get more data to put into the system, there is no real purpose of this tag. --Skippern 14:49, 3 December 2008 (UTC)

In my personnal case, I'm not using surface=* at all with smoothness. What I tag is :
  • highway=track + smoothness=horrible to say that it's not possible to go there with a passanger car, and strong engine is needed, so only off roads vehicles can use it. I don't want to bother with surface because it will need me something like 5 tags to say the same thing. With still the ambiguity :
  • highway=track surface=gravel surface2=roots potholes=huge steepness=40% -> does this mean a passenger car is not able to drive here ? I'm not sure, but that is the only information I want to give Sletuffe 15:06, 3 December 2008 (UTC)
So you are saying that you are even contributing to the problem rather than solving the problem. If lack of usage of surface=* makes routing software not give you the desired result, than how would adding more factors for the routing software solve the problem? If you are thinking about specific types of trails and tracks, why not use scales together with the existing surface? We have an approved sac_scale=* and I am sure you know of the mtb scale as I have seen your name in the discussion. Adding a secondary surface tag is also pointless as the dominant surface is the only needed/used by any renderer or routing software. --Skippern 15:13, 3 December 2008 (UTC)
I'm not trying to solve, neither I am trying to contribute to the problem, I'm just tagging what I think to be usefull. And surface isn't usefull for me. I don't say it is not usefull, but noone will ever force me to tag things I don't care about, or not willing to tag. I've been tagging smoothness for 6 month now, and I find it easy AND usefull and suiting my needs. "By tagging you become a tagger". I wonder who as made usage of surface here and got something usable out of it. If yes, I have no problem with that. If no, then I propose something to increase informations on tracks and path.
The "scales for everything" you are talking about is discussed somewhere else, also I am absolutely not against, I not in favor of it replacing smoothness for now. sac_scale is very usefull, because not covered by smoothness (and smoothness is not able to cover it) so I'm using it a lot on path and footway. mtb:scale (S1 and more) is not convered by smoothness, so I'm using it also. I use smoothness to split S0 in fact so that I can give information for trekking bike, off road vehicules and mountain motocycle in just one tag.
"One surface to rule them all, and in the useless, bind them" are you saying we should spit tracks in pieces of different surface ? are you saying that a track should be tagged with the "worst" surface found ? what's "worst" ? what about potholes ? mud ? a stream crossing the way ? tree branches on the way ? those are common to different vehicles, I won't tag a track with 8 differents scales to say the same thing. If somone wants, please do ! don't try to force people to do it, they probably won't. So give them a choice : "a rough guess"=smoothness and /or "20 differents tags/scales" Sletuffe 16:23, 3 December 2008 (UTC)
Much of what you are implying should rather be covered as hazard=* than giving a series of confusing tags. If there are dangerous potholes in a road than I would like to have a warning sign telling me about it, I do not need to grade the potholes to see if it is a danger to a lamburgini but not to my ford. BTW: hazard have been proposed. (and here --Skippern 16:41, 3 December 2008 (UTC)
And how to solve the case when there are potholes every 5 meters ? Sletuffe 17:56, 3 December 2008 (UTC)
Hazard can apply for the entire road. Do you need to have everything in with a spoon? --Skippern 18:04, 3 December 2008 (UTC)

Another way : reduce the list and think "wheel"

The main concerns about the original proposal is the names of values and their subjectivity. Altough I think personnaly that it will be almost impossible to find a good solution satisfying everybody, I'll try to make this one.

First, I think that the amount of values must be reduced, minimizing the questions when the tag is adopted. Second, if the aim is to identify which type of wheeled vehicules may use the road or not, I suggest that wheeled vehicules are categorized in three groups, each group represented by a smoothness value. Third, we have to be carefull that smoothness does not deprecated the well established tag surface=*. The risk is high that confusion is made between surface and smothness.

The table is ordered by usability from best to worst. It means that a vehicle listed in one category cannot use the roads tagged with values of lower quality. (please english speakers, improve my poor sentence)

Proposal usable by :
smoothness=skate for vehicules using tiny or thin wheels or solid tires like skateboards, roller blades
smoothness=pneumatic roads for normal cars, race bikes, motorcycles, motorbikes
smoothness=offroad only for vehicules with special off-road or all-road tires like SUV, mountain bike, off-road cars, agricultural
smoothness=impassable not designated for wheeled vehicles

-- Pieren 17:03, 3 December 2008 (UTC)

We are allready thinking "wheels" ! "physical usability of a way for wheeled vehicles " is the wording for smoothness.
your table looks really like the very first we proposed, on which we have later added values. I have been fighting long to merge "excellent" and "good" but roller blade users have raised a good remark about their ability to make distinction about a "usual asfalt" and "very smooth asfalt for roller/skate". Since I don't use rollers or skate, it was not my power to impose those tags to be removed.
As of "offroad, offhighways" things, I consider myself being quite aware of that, since I use every week end, in the alps, tracks to reach paragliding take off, hiking trail starts, and even moutain biking trails. And belive me, there are differences that are hard to express, but my saxo is really unable to reach points where some of my friend do well with their non-city-non-plastic 4wd. For management and taking care of our takeoff, we need some times to carry huge wood sticks and sand and cement on places we usuly use our foots to go. In those case, no usual vehicles can drive on where we go. We then rent some special equipement to carry all that Sletuffe
I like it. Yes, it is like the early smoothness draft (which has been made worse). Like written before I would call it surface:evenness instead of smoothness! The different interest-groups should invent there own scales. While scale:mtb is already usable for bicycle and motorbike, there is need for a scale for vehicle with three or more wheels i.e. scale:car. --Phobie 17:52, 3 December 2008 (UTC)
Another remark is that the word "smoothness" is, what I think, a very bad word for discribing what we try to, and maybe much concerns are about that while not clear. I proposed earlier "usability", but I don't remember why, but it was refused. Since words doesn't matter that much to me. I accepted like that.
  • usability=roller / usability=bike / usability=car / usability=high_clearance / usability=offroad / usability=offhighways would have been my choice. But some have raise rather good objections that "car" is very centric and might suppose that "moutain bike" can't, while this proposal wants too. You'r pneumatic word is rather a very good idea... but might be hard to "feel" Sletuffe 17:47, 3 December 2008 (UTC)

Subjectivity: How are these any less subjective than the existing smoothness=* values? smoothness=skate in your example is in principle the same as smoothness=excellent. There's no change at all in how subjective it is -- in either case, you're deciding, "is this usable by an inline skater, or not?"
Number of options: While seven passable options is perhaps too many, three is definitely too few. "pneumatic" is not really helpful: There's a big difference between roads that can be used by, say, a Lamborghini, and those which can be used by an ordinary sedan/saloon (the latter can travel on many roads which would severely damage the former). Perhaps five passable options would work:
  1. perfect: suitable for any vehicle (perfectly smooth) (equivalent to current smoothness=excellent and your proposed smoothness=skate).
  2. suitable for sports cars, racing bikes, wheelchair.
  3. suitable for ordinary cars, "comfort bikes", etc. Most surface=gravel roads would be here.
  4. suitable for "light" off-road vehicles: ordinary 4wd/awd cars with high clearance, mountain bikes, etc.
  5. suitable for "heavy duty" offroad vehicles incl. ATV, tank, tractor, etc.
  6. impassable: useless for any vehicle
This gives one value at each extreme end (perfect, impassable), and two "high-quality road" and two "low-quality road" options. I think this is exactly the right number. Naming them is another problem though, see below.
It seems to me with this scale that, in general, it works out that a vehicle can go one step below where it's intended, but with some risk (of getting stuck, losing parts of the undercarriage, etc.) -- "proceed with caution, skill, and luck, and you'll probably make it"
Terminology: Having tag values which refer to specific transport modes is a bad idea, IMO. On the other hand, numbers are not very helpful for communicating the scale of something, unless you know where both extremes are. a "grade 3" road means something completely different when the scale goes from zero to three, than when it goes from zero to 100. Unless someone else has a brilliant idea, this leaves words like "perfect, excellent, good, bad, horrible, impassable". This problem seems to by why smoothness=* uses the values that it does.
--Hawke 21:54, 3 December 2008 (UTC)
Forget the names I gave for the three categories, I'm sure that native english speakers can find better words. We almost agree : keep the two extremes and only one for the middle. This will significantly reduce the risk of subjectivity, not totally but far better than the 6 or 7 values. Ask yourself which applications needs so many levels. We need tags that are self-understandable. -- Pieren 22:26, 3 December 2008 (UTC)
Reducing it to only three values total completely defeats the purpose of the tag -- the values would be so broad as to be useless. You'd have "perfect" and "impassable" and "something inbetween", which doesn't tell you anything about how smooth the roads are. That's as useful as counting things with "one, two, a bunch". --Hawke 22:40, 3 December 2008 (UTC)
Just a note that on a forest track there could be sections where a ordinary 4wd/awd (Say a Honda CR-V, Audi Q7, luxury Jeeps) can go further but some older high clearance 2wd vehicles need to stop only because of the lack of both ends being powered. And the latter can still go much further than some modern "normal" cars as they might have even more ground clearance than the modern day SUV's. But such distinction might be better left for some minimum_qc=*... And if it's lack of engine power, incline=* might suffice. Alv 08:17, 4 December 2008 (UTC)
I can't help but think that most of the problem is that we are effectively trying to combine two separate concepts into one tag, or alternatively only have one tag when two are required. To my mind (as I've sort of said above), without getting down to the nitty gritty, a road has a surface type and a surface condition. While they are inter-related concepts, they are describe completely different attributes of the road.
The problem IMO with using something like smoothness=pneumatic is that you aren't describing the surface type or condition, but the requirements to navigate that surface. That's fine, but I think in tagging just that, we've jumped a step and missed out on tagging what the road actually is (which seems to me like a critical failing of the process). I reckon that we should start with some sort of objective road surface and condition tags, and from that tags like smoothness (with regard to various vehicle types) can be added or partially inferred. smoothness=skate says absolutely nothing about the actual surface (it can be assumed, but it can't be inferred). smoothness=pneumatic doesn't indicate if the surface is asphalt, grass or sand, nor does it tell me what kind of state that surface is in. smoothness=skate allows me to assume that the surface is probably asphalt, but could just as easily be concrete, or planks (as in those that sometimes are used in small bridges). -- Gaffa 12:08, 4 December 2008 (UTC)
Here I see that I forgot to mention something important : in my point of view, the tag smoothness=* has to be combined with surface=*. I add it in the proposal. -- Pieren 12:26, 4 December 2008 (UTC)
To my point of view smoothness could be combined with surface=* and mtb:scale=* and (potholes, hazard, incline). I don't think we should try to force anyone to tag the way we'd like him to tag. We should stay with words like could be might be or would have advantage if. Many (implied) proposal we have here are related to smoothness and should really be proposed to increase the knowlege of ways. But I don't think it should replace it or we might well end with people tagging nothing. As an exemple, it has been suggested that once the mtb:scale=* is finished, we should give a word on the smoothness page such as "For mountain bikers, you should also be more specific by giving the difficulty of the trail with mtb:scale=* Sletuffe 13:05, 4 December 2008 (UTC)

New simplified version of smoothness

Based on some last comments (pieren and hawke) I'm trying to simplify a bit the numbers of values, try to make it much like a "vehicle scale" and return smoothness2 to it's true meaning : "not really be about smoothness" : Proposed features/usability . Comments have been explicitly said that it could be used in conjunction with other tags to increase knowledge of the road/track/trail. But it still keep the same idea in mind What vehicle do I need to go there Sletuffe 01:25, 6 December 2008 (UTC)

  • I've just read through this page and it took a long time and resulted in a long responice to it, so I've stuck it elsewhere so as not to consume the page. I've stuck it here. I map tracks frequently, so the standadisation of this tag will make a huge difference so I've dicided to contribute my opinions...hope it is in someway helpful. Ben 21:46, 7 December 2008 (UTC)


Why was this moved to the rejected features? Just because the second round of voting didn't go on as scheduled? --Eimai 18:12, 14 January 2009 (UTC)

  • Because it's widely agreed outside the small world of a handful of users here that the tag as proposed was irreparably broken. I suspect what usage there is at the moment comes from either these users, or from new users that have misunderstood it. I'd happily put money on the tag usage being inconsistent if the odds weren't so short. Chriscf 18:31, 14 January 2009 (UTC)
  • Case-in-point: use of "smoothness=bad": [11]. Apparently this is for tracktype values of 2, 3, 4 and 5. Clearly, one of the two is broken, and the tracktype=* values are based on objective, factual observations, and therefore unlikely to be the culprit. Chriscf 18:36, 14 January 2009 (UTC)
  • So it is for tracktype 2 to 5, what does that have to do with it? Smoothness is independent from tracktype. And note I'm not arguing in favour or against having a smoothness tag. All I see is that the tag is being used so there is some need for it, so this basically has to be kept as de facto approved until a better proposal comes along where all the ones tagged with this can be moved to. --Eimai 19:09, 14 January 2009 (UTC)
  • Chriscf, Smoothness=x doesn't = tracktype=y. There not related. +1 on Eimai's point. A grass track may be unusable to the degree that it's classed as 'bad' and a gravel track may be unusable to the degree that its classed as 'bad'. I can't see much with this criticism, although in general I see sense in what you say. The criticism I would have with 'bad' is not knowing where in the order of good-worst it stands. For tracktype:grade3 for example it's clear that it's between 1 and 5. The lack of selfdescriptivness to the word 'bad' isn't such an issue though. It's a side effect of making tags that are practical to use. "Because it's widely agreed outside the small world of a handful of users here" Would these people like to add there opinions? firstly it's not a democracy, so I couldn't care if it's 99 against 1, if the 99 all chant the same weak point. Secondly, I haven't seen this wide agreement, thirdly, the majority of users do 1% of the work, the minority do 99%, and fourthly, I wouldn't define it as a small world, since I have seen multiple names on this subject who are far from ignorant on mapping routes where this applies, and discussing it in general. I agree with your critism's on voting, but you seem to now be against adapting it in relation to points made within that vote, and these points, pro or not, are what need to be thought about, and used to make an adaption of the previous proposal. If usability is the spin off, this may as well just be rejected, but why it needs a whole new start rather than an adaptation I don't know. Ben 20:01, 20 January 2009 (UTC)
  • Tag is broken - fact. End of story. Chriscf 13:19, 15 January 2009 (UTC)
  • Where "fact" means "opinion of Chriscf", sure. --Hawke 16:28, 15 January 2009 (UTC)

User point of view

So much debate on some tags with so few declared users. I'm contributing for one of these users, the "motorcycle map of Romania" (see description in wiki), getting reports from riders, estimating the parameters, playing with them almost daily. For me is obvious that a change must be done, this is how I see a working combination with minimal change in fact. Rootcause I think that smoothness was defined with users in mind, and this is against the OSM policy of "don't map for renderer".

highway=track, very useful, as a distinction should exist between agricultural/forest roads and normal roads(tertiary and so on)

tracktype=grade1,2,3,4,5 - It is well defined in wiki as the visual persistence of the road given by the frequency of use, lack of vegetation destroyed by wheels. Tracktype does not deal with the surface, but this should be more clarified in wiki as it seems many users think so. We might think that grade5 is in bad shape, but it's not like that it can be usead easily or might be bumpy but it means you must search for it, almost cannot see it in the land. From my point of view, this grades are ok, not to be changed.

surface(=paved,asphalt...) it's the cover material, can be used for all type of highways, tracks included. Not much to change also on this parameter. Some might argue that ground/earth turns to mud in rainy weather, but as wheater changes are not recorded in OSM, this should be computed on the spot by the user-software that has the weather parameter active.

smoothness. Here I support fully the idea in "Option 23,754" with the surface_condition instead of smoothness, because an usable smoothness or travel feeling then can be easily computed at the moment of use from the combination of(surface,surface_condition, type of wheel/vehicle/user). We already generate different map for each type of motorbike(user parameter) using (surface, smoothness) combination.

A general argument like "highway=trunk" means automatically something smooth because in Germany it's so, it's not a general rule. Let's not forget that the terms highway= trunk/primary/secondary depicts the administrative category of the road, and not the quality of the work done or the smoothness of surface. If in Germany it might be so, but believe me, in Romania it's not the case, and in the rest of the world the situation is specific to each country. With today's parameters, it's very complicated to depict the difference between the following two roads - (highway=track, tracktype=grade1, surface=gravel, smoothness="best that can be obtained using gravel") and (highway=track, tracktype=grade1, surface=gravel, smoothness="bumpy, hard to use" )

For the first one, cannot use "excellent" or "good", I should use "bad" bud it's not bad at all for a gravel, quite excellent in fact.

Different type of wheel sizes react differently on the same road, for example blade-roller on sand(digs in) and truck wheels still ok. Therefore the smoothness of the road should not be defined starting from wheel or user types. This is now, and it's ok only for few types of surfaces.

Compared to the actual "smoothness", "surface_condition=well maintained/maintained/degraded..." is also easier to transmit verbally, or as a report. Reporting is important to consider, because we will get the most of information from verbal reports, not from accelerometers. Let's think globally, every contributor has eyes, can speak/write, but planographs are non-existent even compared to the number of GPS units. As we give to volunteers the ability to contribute with POIs, descriptions, corrections even if they don't have a GPS, let's keep this philosophy also for state of the road. If we go to a remote place and ask local people about the state of the road, we get mostly useful description of the way even if it's also subjective. I use to get reports like "It is a track, so well maintained, gravel, it was excellent and could ride with 40km/h" or "broken asphalt, speed 70km/h, so shaky I lost some screws" So I'm supporting this way of using the parameters since they do not depend on rendering capability, vehicle type or wheel type and size. --owene 20:52, 3 August 2011 (BST)

Renaming current values

From this long discussion I've felt that many people find the naming of smoothness=* values inadequate. I think that replacing its values with the suggestions in the text (as I've copied below) would make this tag much more popular and very useful.

Current name New name
excellent thin_rollers
good thin_wheels
intermediate wheels
bad robust_wheels
very_bad high_clearance
horrible off_road_wheels
very_horrible specialized_off_road_wheels
impassable no_wheeled_vehicle

--Fernando Trebien (talk) 18:56, 20 March 2014 (UTC)

Please move your suggestion to Talk:Key:smoothness. The proposal is only kept around for historical value and should not be changed anymore. --Tordanik 16:37, 22 March 2014 (UTC)