Hi all,
First I wanted to say that I second what Jason, Kevin, and Rich have said earlier in this thread. The issues we are discussing here are mainly about metadata modeling and they are relatively independent of our choice of identifying scheme. If we comply to the LSID HTTP proxy recommendations we proposed, then LSID and HTTP URLs would be almost equivalent. I say almost equivalent because of the arguments Donald put forward in his last message.
I don't believe that the Handle system would be a suitable identifying scheme for our community, however. That is simply because the Handle System has its own proprietary system for performing content negotiation that is completely incompatible with the standard WWW and Semantic Web applications.
In the Handle system, each handle has a set of values assigned to it. These values may be different representations of a digital resource, such as an HTML page or an RDF metadata record, or even a PDF file. Or can be any other types that a community agree upon. Each value is accessed by an index and a type. In our case types may be mapped to mime-types, but this mapping is only recognized by clients that are aware of the handle protocol. See section 3.1 of the informational RFC 3651 (http://www.handle.net/rfc/rfc3651.html) for more information about "content negotiation" in the Handle System.
The trouble is that this proprietary scheme is a brick wall for standard WWW and Semantic Web applications. These applications have no way of getting to the right object representation using standard content negotiation and thus the scheme can't be used to represent digital or real objects effectively. In fact, these applications won't know how to ask for different object representations and will then get whatever is defined as the default representation, which can be different for distinct handles. In some cases it can be a PDF, in other cases it can be an HTML page asking for payment to see the full article, and so on.
For that reason alone the Handle System is not a feasible identifying scheme for our community. Again, only LSID (with the HTTP Proxy recommendations) and HTTP URLs present content negotiation mechanisms that are interoperable with standard WWW and SW applications.
Regards,
Ricardo