fernando.trebien at gmail.com
Mon Mar 17 16:35:29 UTC 2014
Two subjective opinions that agree create "consensus", this is I
believe what we seek in OSM when defining tags. Replacing 'stiffness'
with something else is absolutely fine with me. I think the word we
replace it with will essentially be the definition of tracktype. I'm
sure I'm not the best person to do it since, not being a native, I'm
subject to the language barrier and to some "translation noise" that I
haven't been able to filter out yet (despite speaking English
When I thought of 'stiffness', I was coming from the apparent focus of
grade descriptions on "compaction"/"hardness" (could these be better
words?). See these excerpts:
grade1: "heavily compacted hardcore"
grade2: "unpaved (...) surface of gravel [a hard material] mixed with
varying amount of [soft materials] sand, silt and clay"
grade3: "even mixture of hard and soft materials"
grade4: "prominently with soil/sand/grass [soft materials], but with
some hard materials"
grade5: "lacking hard materials"
So tracktype seems to be describing the mixture that makes up the
surface according to how hard or soft these materials are. We can then
guess how hard or soft the entire mixture is, and therefore how much
resistance and risks it would impose on the vehicle or even the
To me, the idea a firm/soft mixture seems closely related to "how well
maintained" the track/road is, as mixtures that are not so
durable/steady/firm quickly wear down and look "poorly maintained".
They are so closely related that they both could describe the
essentially same thing (where "maintenance" is observable and
"hardness" is the cause). Where they differ is in some exceptional
cases such as when we have surface=rocky/stone: suppose it is nearly
flat (smoothness=good for instance), could it ever be
tracktype=grade5, despite never being maintained? The idea of a
firm/soft mixture also seems closely related to "how smooth" (how flat
versus how irregular) one would "expect" the surface to be, even
though we can be surprised (as in the case of surface=rocky/stone),
for better or for worse, and this makes "smoothness" a necessary
additional attribute. If we know how irregular the surface is and how
easily it deforms when run over by a vehicle, we can more confidently
come up with a routing strategy that uses both characteristics to
limit speed to a "safe level". And the same logic can be used for
rendering (to alert drivers on situations that require careful
These ideas led me to several articles in Wikipedia in search for
better descriptions (where I believed there could be better
descriptions for what we want to grasp with tracktype):
And also to propose this to the OSRM team:
For another topic: there is another characteristic that nobody has
pointed out yet that also influences the maximum safe travel speed,
and I think it could be mapped easily:
This could be meaningful for some surfaces such as "grass", "mud" and "ice".
What I mean is that the various surface characteristics (current and
future) could be assigned "limiting safe speeds" for each value and
then routing software would simply have to decide what the combined
safe speed is by choosing the minimum, that is, by being restricted by
the characteristic that is most limiting in some particular
combination of surface characteristics.
Let me give you several examples using this interpretation I've just
described (not necessarily correct): if we have tracktype=grade1 (hard
material) + smoothness=bad, we can't travel fast safely because the
road is likely very bumpy. So smoothness is the limiting factor on
safe speed. If we have tracktype=grade5 + smoothness=excellent (a
perfectly flat road with no small rocks, only earth/soil), we still
can't safely travel fast because the soft material would slow us down,
specially if it has just rained (or, in some places, because snow is
melting). Now if we have tracktype=grade1+smoothness=excellent, it
doesn't matter much if the actual material is asphalt or concrete or
tarmac, we know it's hard and flat, so we can expect to be able to
safely travel fast. Conversely, if it's
tracktype=grade5+smoothness=horrible, we know it's loose and bumpy, so
we can't travel fast safely at all. Could you travel significantly
faster (and safely) if we had tracktype=grade1+smoothness=bad or if we
At the same time, many people seem ok with the "surface" tag to
describe the surface. Since I believe they map with the hope of using
this information in apps (rendering or routing, most of the time), I
believe they have expectations regarding surface flatness and material
rigidity for several values of the "surface" tag, and that's why they
don't think tracktype and smoothness are necessary. So we may try to
agree on what is expected by each value of "surface" (that's
essentially what I tried to do in my proposal to the OSRM team), and
also try to agree on which values present too much variation as to
require additional descriptions (tracktype and/or smoothness) or
better values in the surface tag.
As a result, mappers could end up mapping:
- all 3 characteristics, allowing the application to decide which one
is the most important for its purpose
- surface alone, when they believe it is sufficiently specific (the
application then would be based on what's usually expected from that
- tracktype + smoothness alone, if they don't think the actual surface
- only tracktype or only smoothness, allowing the application to still
be able to impose some safe speed limits based on partial information
about the surface
- surface + tracktype OR surface + smoothness: the application could
then consider typical expectations of the surface and partially modify
them, which I think is a somewhat clever way to handle the partial
Following this same logic (safe speed is limited by independent
surface characteristics), we can use these characteristics
independently to easily render warnings (the need for this was
mentioned at some point in this other discussion:
On Mon, Mar 17, 2014 at 8:40 AM, Dave Swarthout <daveswarthout at gmail.com> wrote:
> I knew I would be opening Pandora's box when I made those statements. As for
> tracks, I should have prefaced my remarks with In My Opinion -- I am well
> aware that it's too late to change the current situation.
> I would still argue that smoothness is a valuable parameter. Ignoring speed
> limits and such, it determines how fast you can comfortably travel on a
> particular highway, among other (more subjective) things. And I'm familiar
> with the long thread about trafficability, to the extent that I could follow
> all the various opinions and problems it exposed. This whole thing is a
> tough nut to crack.
> Which is one reason I suggested those other terms to describe "stiffness"
> which I just can't get my head around. Stiffness just is not right for that
> use. I dunno what I would rather do. Maybe as you suggest, it should simply
> be removed.
> On Mon, Mar 17, 2014 at 1:59 PM, David Bannon <dbannon at internode.on.net>
>> Good on you Dave, I do like a good rant !
>> On Mon, 2014-03-17 at 10:47 +0700, Dave Swarthout wrote:
>> > <Begin rant>....IMO tracktype should describe the physical
>> > characteristics of a track, not a highway, and it should have nothing
>> > to do with "how well maintained" it is.
>> Great in an ideal world Dave. However, there are many highways in the
>> world that are also 'tracks'.
>> Recognising this, the OSM Lords have given us highway= tags to describe
>> the purpose of a road. And tracktype= to describe the condition. Many
>> cases, the tracktype= is not needed as its condition matches its
>> purpose. But in situations where that is not the case, life threatening
>> situations arise when a map user is not appropriately warned.
>> I agree tracktype may not be ideal but its better that the rest and I
>> think its too late to dream up a new one.
>> > ...... those surfaces have an additional important characteristic
>> > called smoothness.
>> 'smoothness=' is not really appropriate as there are many, many roads
>> that have issues beyond smoothness. I have seen tracks that appeared
>> beautifully smooth but were beyond my ability !
>> > How a "highway" ever got a tracktype tag is beyond me and seems a big
>> > mistake.
>> That was part of the original definition of tracktype= when it was
>> > .... As far as smoothness is concerned, many have derided it as being
>> > too subjective.
>> Look, lets be honest, just about anything in this world except the
>> integer series has some subjective aspect. Lets get over it !
>> > say, "particularly regarding surface stiffness", IMO the word
>> Yes, Dave, I agree, Fernando's use of the word 'stiffness' is a bit
>> dodgy. But thats a 'subjective' opinion.
>> > Perhaps soundness, permanence, or better yet, durability.
>> No, I really think this is about how usable a road is given a set of
>> vehicle and driver experience. We, on the AU mailing list discussed
>> words like 'trafficability' and, from memory, some even worse ones !
>> But I do want a good solution and I'll agree to an OK one if its all I
>> can get. I want to badger the renderers to take note of the state of a
>> road before someone gets killed using an OSM map. Its only a matter of
> Dave Swarthout
> Homer, Alaska
> Chiang Mai, Thailand
> Travel Blog at http://dswarthout.blogspot.com
> Tagging mailing list
> Tagging at openstreetmap.org
+55 (51) 9962-5409
"The speed of computer chips doubles every 18 months." (Moore's law)
"The speed of software halves every 18 months." (Gates' law)
More information about the Tagging