FWIW, in the session on Wednesday, I did try to plug the interest in “linked data” and other practical applications of the “mid-lower levels” of the SemWeb layer-cake – the sort of things I think Mikael is referring to as “class 2″ in his post here

http://mikaelnilsson.blogspot.com/2007/11/position-paper-on-semantic-technologies.html

Also I wholeheartedly agree with your comment that “Almost no one is really engaging in a conversation that explains and demonstrates the ways in which semantic technologies might improve current applications and make new applications better.”

Indeed, someone asked the question, “OK, why should I, as the developer of an application that supports SRU, OAI-PMH, XML etc (i.e. a class 1 app in Mikael’s classification), be interested in exposing RDF data?” (or words to that effect).

And I couldn’t help feeling that what we were missing was a couple of nice little demonstrator applications that showed – not told, showed – why that step would indeed be a valuable one to take. Even just to enable the simple – but hugely important – things like merging data from distributed sources.

And, yes, as you say, those demonstrators need to have a presentation layer that doesn’t terrify the non-graph-reading majority!