[OSM-talk] Philosophy about Autorouting for Cyclists and new key class:bicycle

[Felix Hartmann]

Even though there is a huge userbase in OSM that are avic cyclists, most 
of the information is still car centric, even though there are good 
intentions to change this. The problem is, we are living in a motorcar 
centric society, hence our whole road network is based on the idea to 
enable motorists to get quick from A to B. There are plenty of street 
types and access rights, only to enable motorists to go quicker. On 
motorroads, there are even laws to exclude any slow vehicles from even 
accessing them. There are laws and street signs to make slow motorrists 
(trucks / hgv) to stay on the right side, so that other motorrists who 
have nothing else in mind to go as quick as possible, even though this 
is not sustainable at all, overtake without being hindered.


Do I think there is a problem with that? No, it's just reality. What I 
want however is that maybe in a more environmentally friendly future, 
such rules and infrastructure is also setup for cyclists - because noone 
likes going slow just because there is a lot of traffic. Be it as a 
pedestrian, as a cyclist, or as a horserider. There is no such 
regulation or infrastructure for any other mean of transport but 
motorists. There are no roads where slow cyclists, are forced to stay on 
the right side. There are probably only a handfull of cases where a 
pedestrian was ever fined for walking on a cycleway for slowing down 
bikers, and it would be unfair to do so, because the infrastructure is 
not built to have cycleways to cycle fast.


      Current Situation

The overall situation is however not so bad. Every biker can in most 
places of the world find their preffered way with good local knowledge. 
This is where OSM should help. If inside OSM there is information about 
how welll a way is for what kind of cycling, then we could not change 
the system, but at least use those ways, that are already more than 
acceptable. the  Sadly though many people in OSM are not able to leave 
their small focussed mind and cannot espace their caged mind and try to 
use a motorist perspective to do bicycle autorouting (e.g. CycleStreets 
<http://wiki.openstreetmap.org/wiki/CycleStreets> or 
Cycle_routes/cyclability 
<http://wiki.openstreetmap.org/wiki/Cycle_routes/cyclability>). They 
want to use objective tags based on hard and accountable facts only, to 
describe how well a way is suited for cycling (e.g. Radverkehrsanlagen 
<http://wiki.openstreetmap.org/wiki/DE:Bicycle/Radverkehrsanlagen_kartieren> 
). They argue that we should have only verifyable keys and tags. They 
miss the big point, that for motorists the whole system is based not on 
objective but subjective thinking, and that therefore we can use 
objective keys to say, this is a primary road. If we only want to have 
objective tags, then I applaud anyone who goes out an puts a bot to 
delete the key "highway=*" and only use verfiable keys like, traffic, 
lanes, access rights and so on to describe it. Would this work? In my 
opinion never, because it is far too complicated and will not allow us 
to classify a way as "primary" or "secondary" or "track". The 
highway=path versus highway=footway discussion of what key one should 
use, shows us that people want to be able to classify a way from their 
subjective position, and not fiddle with access rights and measurable facts.


As well as we have already classed any way from a motorist driven 
perspective into a motorway or into a track, we will have to do so also 
from other perspectives. And Openstreetmap can be the tool to achieve 
this. Just a little example to make this point clear. In the wiki we 
have the guideline, to tag any non classified way - often without right 
of way for motorists - as highway=track if it is wide enough for a car 
to use it, or as path/footway if no car can use it. In an objective 
world we should instead describe this situation by surface, width, 
traffic, access_rights, and so on, and I support to do so. However the 
large majority wants to only say this is a footway, this is a track, 
this is a cycleway. What they really mean by this, and want to know 
however is. Would I like to cycle here. Would I like to drive my car 
here. Would I like to use this way to Walk from A to B.


I'm in no way esoteric, and a highly objective and agnostic person, but 
there are things that one simply cannot describe using the old patterns 
and objective measurable facts to describe ways suitable for cycling. As 
our world is not made for cycling, it is much more difficult to describe 
and classify than to do such a thing for motorists! Therefore one can 
argue, that a single key like "cyclability" will not work, because every 
cyclist is different. However we do the same for motorists and it works 
so well (though not perfect) that it is common to use it. Autorouting 
for motorists would never work without taking account of the "highway" 
tag (which is as described before, not objective but subjective, and 
maybe even this subjectivity has made it into laws and our legal 
system). Anyone who is able to write a program within one year that only 
based on objective facts inside OSM (wihout traffic patterns - which 
would work - but is not what OSM is about) autoroutes me from Hamburg to 
Munich inside a motorcar wihout breaking any law, and being able to 
route me as good as a program disrespecting all keys and only looking at 
the highway key will get me as a sclave for the rest of my life :-)


      Solutions

The big question therefore is not, do we need such a subjective key, but 
how to design it so that it is as objective and comparable as possible! 
Because if it gets too complicated it will not be used. If it is not 
specific enough on the other hand, it will be of no use... I must admit 
that theese two pages on the OSM wiki (Radverkehrsanlagen 
<http://wiki.openstreetmap.org/wiki/DE:Bicycle/Radverkehrsanlagen_kartieren> 
and Bicycle <http://wiki.openstreetmap.org/wiki/Bicycle> ), even though 
written with best effort and well done, are not at all making me want to 
use them. The wiki page about quick and simple cyclway mapping 
(DE:Bicycle/simple_mapping 
<http://wiki.openstreetmap.org/wiki/DE:Bicycle/simple_mapping>) does not 
make it any better. It is too complicated for me as Joe Average and it 
does not help me to make me find the best way on a bicycle from A to B. 
It is perfect for making a nice looking map, showing me a very objective 
description from a cyclist perspective (see here: velomaporg/map-legend/ 
<http://www.velomap.org/velomaporg/map-legend/> ) and to a point allows 
90% correctly guessed autorouting after setting up a very complicated 
system of rules (1 <http://svn.origo.ethz.ch/wsvn/openmtbmap/lines> and 
2 <http://svn.origo.ethz.ch/wsvn/openmtbmap/bikingmap/lines> ). This is 
good enough that a moderator, "Rik" of the biggest traffic cycle (mtb) 
forum writes ( post_by Rik 
<http://www.mtb-news.de/forum/showpost.php?p=7105747&postcount=6> ):

/Ich bin gerade mal etwas sprachlos!
Mein Weg von der Arbeit nach Hause (von Potsdam nach Berlin, ca. 46 km) 
entspricht in der Planung mit VeloMap in weiten Teilen exakt der Route, 
die ich mir über einen längeren Zeitraum nach und nach immer weiter 
optimiert habe. Das ist wirklich beeindruckend!/Translation: /Currently 
I am a bit speechess!; The (calculated by autorouting with the 
velomap.org sic!) way from my home to my work (from Potsdam to Berlin, ~ 
46km) is in large parts the exact route that I have optimized over a 
large time span. That is really impressive./


However that is just lucky guessing and routing with the same map may go 
wrong to as long as we miss a subjective key like "highway" for bikers. 
There has been some discussion about how such a key should be designed 
(Proposed_features/Cycleworth 
<http://wiki.openstreetmap.org/wiki/Proposed_features/Cycleworth> and 
Cheltenham_Standard 
<http://wiki.openstreetmap.org/wiki/Cheltenham_Standard> ) but there is 
no widespread use of the first, and no guidelines for the second. I 
personally gave up trying to argue against people who cannot get out of 
their "caged mind" (see above). What I want is a constructive talk about 
how we shall design such an subjective tag, that has the goal of making 
OSM the perfect database for bicycle autorouting. I have started a new 
wikipage here: ( Class:bicycle 
<http://wiki.openstreetmap.org/wiki/Class:bicycle> ) summarizing the 
most constructive schemes found until now to a new key "class:bicycle" 
and in my opinion we should also have such a key other types of 
transport like class:hiking or class:horseriding. I do like the 
Cheltenham_Standard but am of the opinion that we need to have a much 
broader approach and keep it simpler, as much of the information that 
the Cheltenham_Standard integrates, is already in OSM and there is no 
need to multiply it. The class:bicycle key IMHO should be relative to 
the "highway" key and take into account objective tags (e.g. lanes=2 & 
oneway=yes) that already describe a road well, with the aim of allowing 
good autorouting for bicycles by user type, however not the aim of being 
directly displayed on the map (though clever ideas to transport that 
information to map users is of course closed out, it is simply not the aim).

Cyclists are different, and therefore besides a broad class:bicycle key, 
we should also consider of making it possible to use this key for 
mountainbikers, roadbikers and differentiate between someone who want to 
find the quickest route versus someone who wants to find the quitest or 
most scenic route. The aim of the key is not to map every street by it, 
but be used when it is needed. So if your "autorouting program" in 
bicycle mode sends you over imperfect streets/ways, then use this key to 
tell the autorouter that it should search other streets instead. The 
other streets should then have a "positive information" class:bicycle" 
so that the autorouter can prefer them.


      Opinions / Discussion

I do know that the ML is quite an anarchistic place where everyone 
writes what he wants where he wants however most threads are endless and 
do not serve the purpose of really bringing work forward. I don't want 
to discuss a single phrase anymore with people who think we should not 
have such a key (but I'm sure you'll find many others that like to 
discuss this with you, so "go forward and have your fun"). I'm okay with 
a debate that is constructive on how we can best design such a new key, 
but will not engage in another endless discussion like the "footway vs 
cyclway vs path" Therefore I have started a wiki page here for you, 
where you can lay out why you are against this key 
Contra_subjective_keys 
<http://wiki.openstreetmap.org/wiki/Contra_subjective_keys>. If you want 
to work constructively on achieving a class:bicycle key Class:bicycle 
<http://wiki.openstreetmap.org/wiki/Class:bicycle>, please reply to the 
"class:cycleway - how should it be done" message instead of this one. 
Replying directly to this message however anyone can bring forward their 
arguments and discuss in an anarchic way without rules (but 
psychological pressure of others :-) about my message. But DO NOT write 
a single line about why you think it is a bad idea to have "a class:*" 
key into the "class:cycleway - how should it be done" thread.


There is a supporters / objectors page here to voice your opinion: 
http://wiki.openstreetmap.org/wiki/Class:bicycle:opinion

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20100503/8d1fb077/attachment.html>