[Tagging] Proposal for a study about tagging "suitability"

[Pieren]

In recent times, I've noticed few tags that are massively incorrectly
used in OSM with the help of taginfo.
Two examples :
"designation" : 31000 values for 111000 objects ([1])
"location" : 7300 values for 31000 objects ([2])
Of course, some keys are per se multivaluable, like "source", "fixme",
"comment" or "note". Most of the other keys are wikied with a list of
values which can be very long but cannot be closed by definition (e.g.
"amenity", "landuse", etc). But we have a category of keys where the
list of values is limited by definition like "oneway" or "tunnel".
Some QA tools even reports "unknown" values as "errors" or at least
"warnings" for them.
I think it would be interesting to study the ratio values/instances
for the last category. When it is exceeding a ratio 'x', it is
indicating that the tag failed on its job either because one of the
OSM editor is using it incorrectly (P2 presets for "designation") or
simply because it is not enough self-explanatory ("location").
Any thought, comments ?

Pieren

[1] http://taginfo.openstreetmap.org/keys/designation
[2] http://taginfo.openstreetmap.org/keys/location