| For what little it's worth, I've taken a moderately detailed look at handles, and I reached the same conclusion some others have: it moves the existing problems of URIs around to other parts of the architecture, but doesn't fundamentally eliminate any of them, and doesn't provide enough capability improvements or social improvements to outweigh the social and adoption costs. (It may be better than minted URIs -- it wasn't obviously better to me, but I'm not an expert -- but I thought not sufficiently better to abandon minted URIs.)
I'll be interested to hear others' reactions. I think it may have been discussed on the semantic-web list, so might be worth checking the archives before posting.
John
On Nov 5, 2008, at 4:01 PM, Steve Ray wrote: I think we should be taking a hard look at Bob Kahn’s ideas of the Handle System. Take a look at http://www.doregistry.org. His idea of handles bypasses URLs and DNS and provides a clean resolution system at the object level, which would appear to work nicely with ontologies, and pieces of ontologies. One would still have to ensure a clean configuration management system ran on top of it all.
|