She bounces around a bit between characterizing it as a not-a-theory versus a failed theory. It's a failed theory in the sense that the framework hasn't yielded any predictive theories nor has it provided a simpler way to understand existing theories.
It provides simpler calculations in some cases. That's why physicists today elect to use the standard model, string theory, or brane theory depending upon whichever yields the easiest calculation for the problem they're trying to solve. One thing of note is that there's never been a calculation that hasn't agreed with observation. We simply have three different models that all work equally well. I think it's dishonest to not point that out.
