[Tagging] Tags useful _SUMMARY_ for rendering of roads in poor conditions

[David Bannon]

OK, this discussion is huge and conducted in a great manner.

But being so huge, I feel lost !  So, here is an attempt to summarize
where we are and what the options seems to be. Maybe by identifying what
we already agree on, we can see the way into the rest ?

If people think its a good idea I could post a more evolved summary onto
my OSM wiki page where we could all have a hack at it, might be more
manageable than the mailing list ? If nothing else, we need to break
this very complicated problem into manageable hunks.

Think of this somewhat like a flow chart, I just have not drawn it up...

Do we all agree that its important that significant maps, such as the
one on the OSM website, shows some indication if the road may be in a
state where some drivers are uncomfortable (right through to
dangerous) ?

If Yes, proceed, if No, please explain why not. You may like to address
this -
http://www.abc.net.au/news/2012-11-07/25yo-man-dies-of-thirst-in-outback-queensland/4357380
There are lots more. Tourists from outside Australia are at particular
risk.

OK, assuming we agree we want 'something' ...

We need some tag (or tags) associated with a way that tells a rendering
engine this way is one that might need caution. We can try and use
existing tags or invent a new one. 

The "new one" option (such as BGNO's trafficability) could be tuned,
based on experience, to do exactly what we want. On the other hand there
are currently no ways in the database using that new tag. There are 3
million surface= and 2.5 million tracktype= tags in there. Mappers put
used those tags in there for a reason. 

If you want a new tag defined, please say so, maybe with a new subject ?

Continuing on, assuming we support using existing tags, which ones ? At
lease three 'approved' candidates, four if you include 4wd_only.

Surface= has about 3 million ways that are what we, in Oz, call
'unsealed', dirt, sand, gravel, unpaved and so on. This is not a bad fit
but neither is it perfect. http://wiki.openstreetmap.org/wiki/Surface
"To provide additional information about the physical surface of
roads/footpaths...." However, my experience is that a precise statement
about the surface does not necessarily relate to its
"trafficability" (thanks for the term BGNO!). I have driven sandy roads
that were so easy and somewhere else, spent a day with a shovel digging
through sand. Similarly, hard packed clay can sometimes be preferable to
a made gravel road that has developed severe corrugations. And a sealed,
tarmac road that is breaking up is a nightmare.

Tracktype= has about 2.5 million grade2 and beyond ways. "Tracktype is a
measure of how well-maintained a track or other minor road is."
http://wiki.openstreetmap.org/wiki/Tracktype
Thats a lot closer to what someone (or a router) might be wanting to
know. It can and should be applied to all sorts of highway= ways, not
just =track and that seems to be a major problem. In some people's view
(Malenki..), it should be used only when highway=track. I and several
other people (and the wiki) disagree. The values of Tracktype are not
intuitive. The values are linearly expandable, to cover more extreme
road conditions, grade6 is already widely used but not approved.

Smoothness= has about 25 thousand ways. Thats drawing the line at
very_bad. But there are another 40 thousand 'bads' so its hard to call.
http://wiki.openstreetmap.org/wiki/Key:smoothness
I am personally convinced this tag would be used heaps more if the
values did not seem to make some moral judgment ! As I said before, I
could never label the pretty road I live on as "horrible".
There is some support for making a new set of values and that would be
cool (Fernando, Martin). But has the horse already bolted ? Surface= and
tracktype each have more than 100 times more use.  Further, if we come
up with new values, why not a new name ? Truth is, 'smoothness' is only
a small aspect of trafficability (there, I used it again!).

4wd_only=yes.  Used 3 thousand times, more in Australia than elsewhere.
In hindsight, maybe it could have done with at least three values,
'recommended', 'yes', 'extreme'. It does cut in somewhat beyond the spot
we are talking about, I get the impression that people want a road
labeled differently long before we get to 4wd_only=recommended.

Any one of the above, or a combination ? Personally, I think a
combination would be over complicating it. Just my view.


Other Issues -

How to render it ?  That can come later on I guess.

Wolfgang, Peter, Janko, Gerald  warns about subjective tags. Truth is,
almost everything we record is subjective to some degree. I'd take the
legal approach where they talk about a "reason person's view". For a
normal road, thats someone driving a conventional car. For a mountain
bike track, its someone riding a mountain bike....
Fernando pointed out that to make a truly objective assessment, we'd
need many more tags and some elaborate technology to measure. Gerald
suggested a smartphone app to do the measuring but is he allowing for
variation of suspension in the vehicle in use ?
David S and Dominic don't seem to want more detailed measures either.

David S reminds us what highway= tag is about. Its to describe the
purpose of a road, not in any way its "trafficability" (David S said
"usability but I am starting to like trafficability...). he is right but
the anomaly is highway=track, its use opens up, at present, the
tracktype modifiers, wrong, wrong....

One important effect of the highway= tag is "more important" roads get
rendered at lower zoom numbers. Useful when you want to see how to get
from A to B. Sadly, we hear of people taging important roads as =track
so their usability can be described by tracktype. And then you cannot
see them at all at sensible zoom levels. Sigh....


Now, I have not mentioned everyone nor every view, impossible !  Thats
why I think its time to move to the wiki, perhaps show a series of
options and just see who really wants to vote for what.

But, I really must thank Fernando for driving this issue, its very, very
important and damn hard as well !


David

On Fri, 2014-01-03 at 21:10 -0200, Fernando Trebien wrote:
>  a massive contribution...