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

[Felix Hartmann]

On 03.05.2010 21:47, Ben Laenen wrote:
> Felix Hartmann wrote:
>    
>> If in OSM we really want to get in more mountainbikers, we have to start
>> with unofficial routes. I will think about it for the night, and put up
>> a wiki page tomorrow, put some notices on this on the big forums
>> (hopefully they will get ~5000 pageviews, put them in my feedburner
>> newsletter (1200 recipients) and as of next Friday render unofficial
>>   routes.
>>
>> Once we have more than 500 unofficial routes (I'ld say this takes no
>> more than 14 days), I will take out official routes form my maps, except
>> if they are labelled with additional information to make sure they are
>> not a trekking bike route labelled as mtb route.
>>      
> Here's the thing: we just do not map unofficial routes. Only the ones that are
> signposted. There are enough sites where you can submit your route
> suggestions, and there's no reason why this should be in the OSM database.
>
> Greetings
> Ben
>    
Well there is a big reason to do so. You cannot import such routes and 
map then onto existing streets. Also no such sites feature a tagging 
interface that is accessible with the tracks. They will not help 
autorouting at all. The data therefore has to be in OSM to be used (a 
parallel database would be stupid, because due to the small amount of 
data would IMHO cause more traffic and data growth than doing it 
directly in OSM).

We did not yet do so. But we also map other unofficial unphysical things 
like boundaries (which are in no way public domain in Austria or 
Germany, you are allowed to have information where you are, but not 
where the boundary is running). We also map skiroutes, and they are 
usually not signposted, and only randomly officialy noted. We have keys 
for grooming status of skipistes, and if you look in OSM there is loads 
of other info that is not physical or where there is good reason to not 
include it. However the big strength of OSM is that we do have all this 
data, and with time and crosschecking data, also other user classes can 
make good use of it. One nice example is that we don't map whether a 
street is inside a city or not. We do however map 
source:maxspeed:CountryCode=local/urban. With this information we can 
indirectly find out if a street is inside or outside of city boundaries. 
Over time with some smartness you can make up for many missing keys, but 
this is not enough to exclude others.
In future there will be a strong request for traffic information (oh 
yes, TMC is nothing better at all than unofficial routes, it is run by 
private companies and adding TMC serves no open data request at all - 
though I am sure people could argues pros here for pages too). Therefore 
I simply don't accept the point that we "don't" do something (as long as 
implementing it hurts noone). So having unofficial routes would have 
enough reason, and the only contra you bring is we don't do it because 
others do. Come on, at least try to be creative and give valid reasons 
why it should not be inside the OSM database. We don't do is is none. 
And that there are other websites for routes is even more lame. There 
are also other mapping data providers, but still we decided to go out 
and map.