Don't need no hierarchy, so sit in a hierarchy 
David Weinberger isn't clear about language. It's not about two polar opposites (folksonomy and it's arch-rival, taxonomy) that try and talk to one another. It's about the fact that when folksonomies get to a certain point, they don't scale either. My delicious tags have a "from:" namespace that helps me separate an article from the New York Times and an article about the New York Times. That kind of namespacing can be better expressed semantically. Hierarchies don't always work because we don't always think in hierarchies. Tagging doesn't work great because when I say something it can mean something different from when you say something. If I say "Sellotape", am I talking about adhesive tape or am I talking about the particular brand of that tape called Sellotape®? It's about semantics. Semantics, semantics, semantics. That's why I am tracking my attention metadata and buying a notebook to experiment with attention and semantics.
Words have many meanings. Words do not act the same. If words did act the same, then we would have no need for poets.
If we use a tag, we are still sitting within a namespace. It's called the tag namespace and it's controlled by the vendor. That vendor may be Flickr or del.icio.us or Technorati. That vendor has control over how that tag is used because if you tag something inappropriately, they or their variably defined community will change it for you. The "tagosphere" is an oligopoly.
That's why semantics matters. Because you, the "user in charge" can actually be in charge. You can say "this is how I organise my data". Limits of my language; limits of my world.
If you don't like the way that your data is organised, you should be able to throw off all the chains and say "I'm organising it differently". The trick is (a) making it easy for the end user to grok this and (b) making it easy for computers to understand that when I say "trousers" and my American friends say "pants", we both mean the same thing.
Of course, here you get everyone saying "well, you need top-down organisation - you need a thesaurus written by some so-called authority to assert that Americans say pants and Brits say trousers - and authorities don't work!"
No, you don't. I can say, in a bit of Notation 3:
britishenglish:trousers owl:isEquivalentOf americanenglish:pants
And you can too. And so can someone else. Authority, I lack. Communication is indirect at best.
We all put our equivalences out there as much as we need them, and we engage in a free market process of assertion making and negotiation. The SemWeb becomes equivalent to how we use language in reality. Perhaps we need SocialOwl, a simplified subset of OWL to allow people to make assertions about equivalence and disequivalence as 'votes' in a global or local game of he-says-she-says.
'When I use a word', Humpty Dumpty said in rather a scornful tone, 'it means just what I choose it to mean - neither more nor less'.
'The question is,' said Alice, 'whether you can make words mean so many different things.'
'The question is', said Humpty Dumpty, 'which is to be master - that's all.'
See here and all this will become clear. My training in the philosophy of hermeneutics becomes suddenly somewhat useful.
And, remember, everything is not miscellaneous. Everything is language. Language is everything. Get that concept in to your head, and then we can finally progress. Language can be embodied in speech, writing. Why not on the Web also? Just because there's a lot of language, doesn't mean it's not meaningful.
Hierarchies don't disappear if we stop talking about them. They just become hidden. The fun starts when you can take hierarchies and remix them. OPML gave us this with the 'include' type. Let's run with it.
Finally, if you didn't understand the above, it's not because I'm trying to be obscuritanist. If you want that, e-mail me and I'll point you to people who are obscuritanist. Lack of understanding on the part of the reader is due to lack of understanding on the part of the writer. Over and out. 
Further reading: Upper ontology and Semantic interoperability
