[jessriedel]

I wasn't disagreeing that it was testable.

I think most people who are consistent with their Bayesian reasoning would grant that SUSY as a solution to the Hierarchy problem is ruled out at the 90% level. More precisely, if you had asked these theorist 25 years ago where they would expect to see SUSY conditional on it existing and solving the Hierarchy problem, they would have put well over 90% of their confidence weight in regions that have already been ruled out. (But obviously, if your prior for SUSY is sufficiently high, you can still claim its likely no matter how much parameter space is ruled out.) I am eager to take 10:1 bets that SUSY will not be found at the LHC.

It does seem telling to say "Well, yes, most of these guys haven't produced theories which are testable, but these two guys have! Oh, and yes, of two places where they have produced testable predictions, one (SUSY for hierarchy) is very likely ruled out and the other (AdS/CFT for QGP calculations) is a mathematical technique (rather than a new theory of physics) used to calculate something that few people care about other than because it is mathematically related to quantum gravity speculation.

(I mean, don't get me wrong, these guys are very smart and some, especially Witten, have made very important contributions to understanding the math of QFT better. And Nima has made all sorts of testable models that were eminently worth developing even if they turn out to be wrong. These are awesome physicists. It's just that $3M each is a lot of money...)