"In the short term OPML data which is clearly constrained, like that provided by Bloglines, is machine interpretable and potentially integratable when mapped to a logical model like that of RDF. But there’s still so much manual work needed (to say e.g. Bloglines OPML means this, Top Ten Sources OPML means that, Alex’s reading list means the other…) that such an approach is limited in its scalability across the web space.
With RDF we have the specifications and tools for the kind of system Dave’s suggesting, it’s just a simple matter of programming (I’ve gone a fair way there myself). I’m sure it could be done using OPML-oriented systems, but some kind of integration model will still be needed, not to mention standardisation of the format. Lots more work."