While I believe that this is in general a good idea, I don't agree with
tagging those place objects with admin_level tags. It would lead to more
confusion and I don't see why we would need it, or what it would solve
(rather the opposite: it would contradict the place concept of a place
being something orthogonal to administrative structures).
Clearly (and I think you wanted to say this but have just omitted it by
accident), place=* should be added to these place entities (to indicate the
kind of place, like city, suburb, etc.).
I don't think (boundary) relations are strictly needed (the only advantage
is explicit coupling of the node with the area, which otherwise would have
to be done by name comparison and spatial analysis), but if it makes
mapping those easier I won't oppose. Naturally multipolygon relations can
make sense if the area is big / has complex boundaries.
I very much agree on separating landuses from places / quarters / etc.
Btw.: never quite understood why we (still) have place values like
"county", "municipality", "state", "borough", "district", "subdivision",
"country", "province" and some others, which clearly refer to
administration. As you're already tackling places, IMHO you could try to
deprecate these in the same go.
Tagging mailing list