[OSM-talk] Zero tolerance on imports

Felix Hartmann extremecarver at gmail.com
Sun Feb 20 23:16:48 GMT 2011


Couldn't agree more to it. Imports kill community and scare novices away.

Austria two years ago was actually one of the best mapped countries for 
the density of population. Much better in general than Germany or other 
countries. Since we began importing plan.at community stopped to grow 
and many many got pissed of. Also value of imports is not good. 
Switzerland 2 years ago nowhere as well covered as Austria, on the other 
hand had no imports, but largely good aerial imagery. By now Northern 
Switzerland map data quality and attributes are absolutely stunning. In 
Austria as soon as you move 20km away from Vienna, average quality is no 
good....

I also think that in the Netherlands the imports destroyed what OSM can 
do. In the Netherlands street and landuse data are really complete, but 
quality is just as bad as normal maps. There are far fewer attributes 
than in Germany and if you take pedestrian or bicycle navigation as a 
goal, even though data is really complete, what you can get out of the 
data is crap compared to the bordering parts of west Germany 
(North-Westphalia).

Most important things for OSM are good aerial photos coupled with large 
community. Worst are imports. The United States are so bad, I don't 
think OSM will ever become important there. The biggest thing to 
remember is that "creating" something is much more fun than correcting 
it. Imports make OSM a chore and no fun.
On 20.02.2011 13:00, Jaak Laineste wrote:
>   Without knowing local situation I cannot comment this particular
> case, but I can tell about mass-imports what I have done myself, spent
> each time many man-days for conversions, improving scripts,
> discussions with community etc
>
>   1. second largest city of our country full data (streets,buildings
> etc) import. The city had a few streets before, and person who had
> added most of them, agreed to replace them. It created positive buzz
> about "good quality OSM database", but killed local community.
>   2. Corine Land Cover, nation-wide import. Today I would do it much
> more carefully, or not at all. It has made more mess and troubles, the
> only advantage is that medium-zoom rendering looks nicer with a lot of
> green forests; but high-zoom is terrible and it is very hard to fix it
> now.
>   3. National administrative borders (from state source), all levels.
> This was the only good import so far you could not get the data with
> field survey.
>   4. I have also prepared quite good water info - decided NOT import
> it, keep for manual copy
>   5. In the pipeline is national address database (should have all
> addresses/buildings as points).  I have permission from source,
> accept/comments from the community, reject from local post agency
> regarding including post codes (surprise-surprise).
>
>   So out of 5 imports maybe 1 or 2 were really "good" ones, if you look
> it this way. So imports are very-very dangerous and should be done
> only if there really is no other alternative. My own optimism of
> re-using existing souces (as anyone with strong background in
> GIS/geodata would have) to improve OSM has reduced dramatically.
>
>   But I like the idea that there should be separate shared import-data
> layer in OSM database (API and editing tools) itself, similar to GPS
> files. Right now I have several new datasets just as bunch of OSM
> files somewhere in our HTTP, and URLs are posted to local talk-list,
> but this is not a good solution, I actually dont recall myself where
> they were.
>
> Jaak from Estonia
>
> 2011/2/20 Frank Steggink<steggink at steggink.org>:
>> On 11-02-20 11:54 AM, SomeoneElse wrote:
>>
>



More information about the talk mailing list