I think, that the relation #7633959 should not be a boundary=administrative, because this is a union of communities headed by a mayor from a including community.
This administration can not be directly elected by public, because it is only appointed by all municipal committees what are member of this union.
Maybe this relation should something like a master-relation,(dont know the exact tagging for that) that includes all the boundarys of the municipalities from this union.
Edit: there can existing, or popping up in the future, more relations like this in Italy, because it existing more unions like this all over Italy.
To summarize this discussion somewhat for wambacher:
What the Italians are essentially saying is:
The mapped overlapping relations of grouped municipalities represent essentially two different types of entities:
A) First, describing municipalities collaborating on topics of social and economic discourse, as happens in a lot of other countries as well, and sometimes or often ultimately resulting in a full merger of the communities into one administrative entity (at least that is what happens a lot here in the Netherlands, the situation in Italy may be slightly different in terms of actual mergers).
B) Second, describing municipalities collaborating on a more specific topic: e.g. communities in mountainous areas working together on topics of forestry, tourism and maintenance of e.g. cross-communal hiking routes.
(* C) Possibly even more types of relations between communities collaborating on a specific topic: “cominità montane, aggregazioni funzionali di comuni, città metropolitane//o altro ancora”)
While relations of type A and B may well overlap, two relations of type A or B only, should NOT overlap (and if I understood it well, they don’t in Italy). The examples shown, are of type A and B overlapping, and considered legit in Italy.
The reason why both A and B carry admin_level=7 is that both represent an intermediate level of governance between municipalities and provinces. So from a hierarchical point of view, they both belong at level 7 (there was some suggestion in that list discussion to use level 9 for type B relations, but this was dismissed as being incorrect in terms of the hierarchy of admin levels).
Martin also suggested a possible “type=boundary, boundary=region” relation as a possible option for type B relations, but this did not get traction either in the discussion.
Overall it seems to be considered a legit situation to have both of these A and B relations at admin_level=7 and overlapping, and not a problem, “reality is more complex than we can describe…” is what Sergio wrote.