[OSM-talk] [Talk-us] Karlruhe Scheme addressing ways from 2009 TIGER data

[Anthony]

On Sat, Nov 14, 2009 at 11:49 AM, Mike N. <niceman at att.net> wrote:
>
>> TIGER is fine if you don't have anything else.  But that's not the
>> case where I live.  It's fine to fall back on if your actual address
>> search fails.  But adding it in a way that encourages people to
>> *update* the potential addresses with actual addresses defeats that
>> possibility.
>
>   Perhaps I'm missing something here, but if TIGER potential addresses are
> added by Karlsruhe Schema, shouldn't people actually update the potential
> addresses with actual addresses?

I'd say no.  Preferably the potential address information and the
actual address information would both be available.  They are
different pieces of information, useful for different purposes.

> Mere mortals can do that quickly.  They
> can even reposition over the approximate house line if  the houses are
> visible from a Aerial photo.   And us nerds with GPS's, aerial imagery, and
> editor tools will replace them with house number nodes in the exact
> position.

I don't see the savings.  99% of the work would be in surveying.
Adding a new way is trivial, especially using good editor tools.

But if you want to do that, in locations where there isn't better
information already available, be my guest.  Just don't do it where
there's already better information available.  Especially not in my
county.

> To address comments made by others - having something present prompts me
> to update and doesn't remove motivation.   I'm the exact opposite of the
> "TIGER kills crowdsourcing" point of view; I probably wouldn't have even
> gotten interested if there were no TIGER roads present.

I very much agree with that.  But TIGER roads are half-decent
approximations of what we want in OSM.  TIGER addresses aren't, at
least not in all areas.  If you want to add them in areas where the
data is good, and there's no better data available, fine.  But please
don't add it to Hillsborough County, Florida.  And preferably don't
add it to any county in Florida without first checking with the county
to see if there's better data available - because chances are there
are.

See http://wiki.openstreetmap.org/wiki/Image:Qgis.png for an example
of something I whipped up for my neighborhood in a few hours (which is
about how long it took me to figure out the query "select
ST_Difference(ST_Boundary(a.the_geom),
ST_Boundary(ST_Union(b.the_geom))) as the_geom into quick_test from
neighborhood_parcel as a join neighborhood_parcel as b on a.gid!=b.gid
where ST_Intersects(a.the_geom, b.the_geom) group by a.the_geom;").
I've got address information for every one of those parcels.  This is
much better than the TIGER data, if you want actual addresses and not
potential address ranges.

I haven't uploaded it yet, mainly because I want to do it right.