Johnson has a chapter out [PDF] in a new online book about "Extreme Democracy," which includes contributing luminaries such as Clay Shirky and Joi Ito. In his chapter, Johnson clarifies how his thoughts about emergence in democracy have evolved since his book came out (and since the Dean campaign crashed and burned):
Watching the Dean campaign's meteoric rise and fall helped me to see some of my original ideas about emergence with a new clarity, particularly as they related to collective behavior among humans. I think now that I was really forcing two kinds of emergence to coexist under a single umbrella term. Imagine it as the difference between clustering and coping. Some simpler emergent systems are good at forming crowds; other, more complex ones, are good at regulating the overall state of the system, adapting to new challenges, evolving in response to opportunities. Sometimes, I suspect, it's helpful to blur the distinctions between clustering and coping for simplicity's sake. But when you subject them to the intense scrutiny and pressure of a national political campaign, the fault lines inevitably appear. Right now, emergent politics is brilliant at clustering, but clustering is not enough to get a national candidate elected. In fact, without the right coping mechanisms in place, clustering can sometimes work against your interests. You need crowds to get elected to public office, but without more complex forms of self-regulation, crowds can quickly turn into riots. And riots don't win elections.
He goes on to suggest that the Dean campaign (and, by extension, current forms of emergence in politics in general) was excellent at bringing groups of people together, i.e., "clustering," but lousy at responding to changing circumstances, i.e., "self-regulation." The Dean campaign may have been a complex system, but it was not a complex adaptive system.
As a side note, Johnson cites al Qaida as a good example of emergent organization, just as I had suggested in an earlier post (though I think we know little enough about the top-down vs. bottom-up organization of the group that this remains to be proven definitively).
This is a good read.