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

[Richard Mann]

On Sat, May 8, 2010 at 9:48 PM, Jens Müller <blog at tessarakt.de> wrote:
> Am 03.05.2010 19:29, schrieb Richard Mann:
>> I think routers would be better served if we identify good through
>> routes (ie the equivalent of highway=primary for motorists), and
>> record them as relations,
>
> I thought a router is there to identify exactly that. Why do it manually?
>

Because routing algorithms can't really deal with lots of multiple
factors efficiently. Sometimes it's simpler just to give them a big
hint: if you're heading in this general direction, this route works
better than the others. It also matches what humans do - we learn
certain pathways then tend to stick to them.

Richard