Just to make my position more exactly known.
My opposition is not specifically to this papers method in itself, as much as it is to the way it is used in this debate.
My interpretation of the paper is that this is sort of a "B -> C" paper. Eg. take some model, and some of IPCC opinions (B) and see what happens if you apply them (C , big difficulties ahead). This is a perfectly valid topic for a paper, and their method could be sound (though I did not check those details since this is a rather uninteresting paper at this stage of debate), as long as it is understood that this is a "B implies C" paper, and not a "C is proven" paper.
The problem is when this is introduced into a debate where B is not agreed upon or proven. One can give us a hundred "B->C" papers, and this will not convince us of C if you do not first prove B.
So, instead of giving us the papers that shows "Model B outputs scenario C", give us the paper that proves/validates something like this: "Model B:s predictions are shown to be skillful when it comes to predicted data X, has withstood specific falsification Y, and has made fullfilled, in absence of the model unexpected, predictions Z during the years after it was a proposed model for {x,c}". (Or some variant thereof. Just a suggestion).
Only after this are the B->C modelings becoming interesting (for me) to analyze.