[Tagging] Housenumber interpolation with regularlyskipped numbers

[Mike N.]

I'll have to admit that I haven't been surveying for *missing* numbers - I 
have been only collecting the beginning and end of each group.   A search by 
address number would yield quite an accurate location.   But the reverse 
case, using my data in the OSM database as a postal mailing address 
validator would fail.   In my mind, having a navigable address would be more 
useful than nothing,  (or in the US, an obfuscated Tiger address).

>Actually, I'd say the interpolation is *better* than mapping each
>individual store, because the store numbers are going to change, and
>14409 might be added, because 14407 splits into two, and if someone
>wants driving directions to 14409 they ought to be given an
>approximate answer, not told "address does not exist".

   I would add to this sites with new construction - I can generally tell 
the entire area of a new subdivision addressing scheme when it is partly 
complete.  I can create an address interpolation way that will not be 100% 
accurate because it includes empty, future houses, as well as lots that will 
never contain a house for a number of reasons, however it is useful for 
navigation long before the commercial companies include it.   The existence 
of an address interpolation way in my mind was a marker that a final survey 
is needed before creating individual house number nodes.

   There are plans in the US to import Tiger address interpolation 
information - which is intentionally obfuscated for privacy reasons by law. 
Tobias mentioned a possible tag  interpolation:complete=yes to represent 
fully accurate address interpolation, which sounds like a simple solution. 
If it is necessary to tag estimated address interpolation differently, it 
would be good to know before the Tiger address interpolation import begins.