Sound interesting? Then have a look at http://www.ietf.org/id/draft-wilper-semantic-content-pkgs-00.txt Is this the next logical step in semi-structured data management or is Chris just trying to stir up a hornet's nest? You decide.
Seriously though, I've been thinking in this area for a while and I know others have, too. Particularly in the preservation & archiving community with things like BagIt and ORE and various combinations. I think something a bit more generic that has RDF at its core and, critically, acknowledges that copies of the same content can be made available from multiple locations...could have quite a lot of potential.
Anyway, I thought it would be interesting to get down to business and actually specify something for people to bang on. So if this is an area you've got an interest in, and you've got a few minutes, I'd appreciate your giving it a read.
Public comments or email to me are fine for now. I can also set up a group if there's sufficient interest.