Just to toss into the mix a notion of mine that has been deprecated on a number of occasions, but which I stubbornly maintain makes some sort of sense:

CSIRO maintains a classification: "Codes for Australian Aquatic Biota" - CAAB codes. CAAB codes actually form a taxonomy: Agassiz's Glassfish (37-310009) falls within group 37 and I think subgroup 37-31. Now, what kind of thing is 'CAAB code 37-310009'? Well:

It's possible that the taxon that it identifies - the "taxon concept" - may change a little over time, but hopefully over that time it still means pretty much the same thing. It's stable, but has a history.

In short, "CAAB code 37-310009" behaves in many (if not most) respects like a *name*. Not a scientific name, or a cultivar or common name, but a name nevertheless. A persistent identifier for a series of taxon concepts which are intended by the namer to be "the same" - even if they wobble a bit over time.

The main reason I like this notion is that one of our databases keeps CAAB and CAVS codes in a special column, and it always seemed a little arbitrary to me. Why these two particular classifications? What if we wanted to incorporate others - would we just have to keep adding columns? Treating these things as another kind of name brings into play all the existing frameworks with which synonymy is managed.

To fully recognise these sorts of things as names, there would need to be the idea of a "name space" - after all, if people are assigning numeric codes to things then they are going to collide. "Scientific Names" is a very important namespace, of course, with a number of particular rules that DwC is very interested in. It might be reasonable to treat namespaces as hierarchical, handling the problem that the meaning of the common name "cow" depends on what language you are speaking. The idea of namespaces being hierarchical can also accommodate the notion that some scientific names are managed by ICBN, some by ICZN.