Brad Fitzpatrick, one of the creators of OpenID, has a paper on the social graph problem that posits an open social graph that records public data about relationships as the basis for other social networks. It's worth reading.
As Dave Winer points out, even though Brad is right, many people are tired of entering this information over and over again into the social network de jour, the economic incentives are aligned against such an open system arising from an existing, for-profit company. Says Dave:
There are enormous economic incentives for companies that run social networks to not let users of other networks access their services. Shareholder value is a function of how many users they have, how they are "monetized" and how hard it is to switch. The harder it is to switch, the more money each user is worth. Any exec that did anything to decrease the number of users they control would probably be fired. So anything that depends on this isn't very likely to happen, in existing networks.
It's important, however, to not conflate wide-area identity systems, like Open ID, and social networks--open or not. I believe wide-area identity systems are a pre-requisite to such a network, but an open social network could exist on top of any or many such identity systems. In fact, correlating identifiers (or not, based on user preferences) would be one of it's primary features, I would think.
Dave mentions that he thinks its politically incorrect to say OpenID isn't going to be the network of choice. I certainly won't throw stones at him for saying so. Maybe there are people thinking that OpenID will rule the identity world, but OpenID's real contribution has been to provide a platform for experimentation and exposing the possible.
OpenID has some serious usability issues. It remains to be seen whether they can be addressed within OpenID or whether a reboot is necessary. But speaking from my own experience, OpenID has opened new avenues of exploration in reputation, delegation, and user-attribute certification for my research group. Coming back to Brad's paper: the existence of OpenID and similar systems allows questions about open social networks to be contemplated--whether or not OpenID becomes the basis for those systems.
One of our goals for the Internet Identity Workshop in December is to support communities of practice that can explore the usability, legal, and other aspects of open identity systems. If you're interested in these issues, come participate.