… the [image] representation derived via [content negotiation from a generic resource] is equivalent to the RDF [served from it]
I asked for a “a note, a specification, etc. that normatively defines what equivalency really is”.
So, after some back and forth between the TAG and the IETF HTTPbis Working Group I happened to receive an answer. Thanks to all involved, I guess it was worth waiting. Seems like the upcoming HTTPbis standard will address this issue, essentially stating that
… in all cases, the supplier of representations has the responsibility for determining which representations might be considered to be the “same information”.
As an aside: I guess I’ll have to be patient again – this time I asked the above mentioned HTTPbis WG why the caching heuristics exclude the 303 header (see the current draft of HTTP/1.1, part 6: Caching, section 18.104.22.168.). But it’s not even two weeks into the question, so I don’t recon I’ll get mail from the chaps before 01/2011 😉