Doc Searls: "Customers don't just need a 'user' interface (provided, of course, by some company). They need a business interface, including their own APIs and their own datasets and schemas, that they alone control. They need to be fully in charge of who they are and what they can do in a business relationshihp. This cannot be conferred by an outside organization. It has to be native to the individual.
Identity services need to be part of the Net's infrastructure — its operating system. Just like Web services and mail services. When identity services get specified, and the protocols (which nobody will own, everybody can use and anybody can improve) virally spread across the Net, the whole demand side of networked market relationships will finally have the power to equal supply. Whole new businesses, new industries, will be built on full-powered customer identity services."
This would be great as it promotes small businesses. However based on my experience, I doubt that every "customer" is eager to start his/her own business. Thus the net perhaps effect wouldn't be that earth shattering.
Arnold Kling further notes that this would in fact slower necessary overhaul of current infrastructure: "Doc insists that the identity infrastructure has to be peer-to-peer. I am more agnostic on that. I trust my mutual fund company, Vanguard, with my savings, and I also would trust them with storing my personal data on my behalf. I worry that insisting on peer-to-peer architecture imposes a requirement that could complicate and unnecessarily delay the development of infrastructure that is long overdue."
The point worth mentioning here is that technically both architectures are not that antagonistic. For instance, decentralised Groove was developed as peer-to-peer application and later enhanced to connect to centrally controlled LDAP directories or corporate portals. To satisfy both end of the spectrum, the identity infrastructur should be build in a way so that it can be used with current centrally controlled client-server systems, but that would be open enough to work with peer-to-peer later on. And as far as I can see, this is the direction identity standards are heading to.
4:32:43 PM