[Tagging] paved=yes/no

[Alan Mintz]

At 2010-07-16 21:55, Steve Bennett wrote:
>On Fri, Jul 16, 2010 at 2:57 PM, Richard Welty <rwelty at averillpark.net> wrote:
> > from a data modeling perspective, though, it's redundant and thus creates
> > the opportunity for inconsistency and unresolvable error.
>
>Do "data modelling perspectives" normally deal with folksonomies
>though? By its very nature, the data entered by OSM editors is far
>more susceptible to inconsistency than, say, a corporate database.

I really worry about the long-term effects of this, though. Eventually, if 
renderers have to deal with all those duplicate (and not-quite-duplicate) 
cases, they will slow to a crawl. Shouldn't we at least try to stem this 
where we can?

The current planet tagwatch has 19243 different keys, and shop, leisure, 
and amenity have over 1000 values each. Many of these are mis-spellings, 
capitalization errors, import-source-specific tags, etc., <rant>but it also 
seems like a lot of people aren't bothering to search for existing tags for 
what they want to map - they just make something up, often without even 
consulting a dictionary - as if nobody else in the world had ever tagged a 
fast-food joint or shoe store. </rant>

This is not a great comparison, but it's all I have access to at the 
moment. If I use the US tagwatch from 20091206, and look only at keys 
starting with lower-case letters and that do not contain colons, there are 
just 1650 such keys. By comparison, the 20100707 planet tagwatch has 7223 
such keys. If someone has the last US tagwatch, we could do a better 
comparison, but there does seem to be a growing problem here.

--
Alan Mintz <Alan_Mintz+OSM at Earthlink.net>