[Talk-us] Address improvement through imports?

[Mike N]

On 11/2/2011 6:59 PM, Richard Welty wrote:
> for what it's worth, i think Martijn is on precisely the right path here.
> imports go wrong when they're done carelessly and without any quality
> control. it doesn't have to be that way.

   I'm not against importing high quality address data, but can see 
Frederick's point: Nominatim could be made to search the US in order: 
OSM addresses - Public addressing data - TIGER addressing data.  OSM 
would hold only the most current and accurate addresses.   That makes 
updating the public and TIGER data sources drop dead simple, and could 
be done in 5 minutes.

I have done a number of complete neighborhood address entries in the 
past and can confirm Ian's math - complete address survey is not a 
realistic goal.

I now bother to create OSM addresses only for popular tourist 
destinations or where the roads have been reconfigured and a classic GPS 
sends you to the wrong location.   Combining these manual entries with 
public sources would be a good fit to the OSM model.