...
We will want resources identified and linked with PIDS to be indexed in popular search engines (Google, Yahoo!, msn, etc.). The URL that the resource is indexed under should be the PID and not the final server 
(true?). What requirements for redirection etc. should be fulfilled? I think this means that 302 redirect would be appropriate.
Play nicely with the Semantic Web (should we consider this?)
A client makes a request for a resource identified with a PID indicating that it wants data (RDF/XML) in the Accept header. Should this be supported at the resolver level (return via 303 some description of the resource, its ids and any public metadata in the resolver?) or should we consider this not a function of the resolver and instead leave metadata query to some other interface?