Doc Searls covers a lot of territory in this Linux Journal commentary on identity.
One concern I have is the need for simplicity, for the Principle of Good Enough that accounts not only for the success of the Net, the Web, e-mail and their founding protocols but for infrastructural building materials like Linux as well. To me, Liberty looks too complicated for that. Yet Craig Burton recently told me the vision I outlined at the top of this piece is actually too complicated for the Liberty Protocol to handle alone. But then, if you take away all the surrounding BS, perhaps the Liberty Spec is really quite simple. Again, I don’t know.
My interpretation of what he’s talking about is personal identity management — where I’d, for example, operate my own personal identity manager (a service on my computer) which would federate with the various identity-based services out there in the real world. Not exactly sure how trust would be establishing in this kind of a structure — seems to me that’s the hard problem.