John Horgan has a new (or should I say NEW) interview with Peter Woit here. Sample:
Woit: No one thinks that the subtle "demarcation problem" of deciding what is science and what isn't can simply be dealt with by invoking falsifiability. Carroll's critique of naive ideas about falsifiability should be seen in context: he's trying to justify multiverse research programs whose models fail naive criteria of direct testability (since you can't see other universes). This is however a straw man argument: the problem with such research programs isn't that of direct testability, but that there is no indirect evidence for them, nor any plausible way of getting any. Carroll and others with similar interests have a serious problem on their hands: they appear to be making empty claims and engaging in pseudo-science, with "the multiverse did it" no more of a testable explanation than "the Jolly Green Giant did it". To convince people this is science they need to start showing that such claims have non-empty testable consequences, and I don't see that happening.
Clear-eyed and to the point.