My reading of Gould found that he is quite propagandistic with his claims, and that the ground that he covers would be more suitably done by a more careful analysis, which I believe to not exist currently.
I was very impressed by the book when I read it, but from what I gather now, I'd probably be much more skeptical on a second reading (also given his other controversial viewpoints, eg spandrels, non-overlapping magisteria).
I cannot see how spandrels would be particularly controversial, though you need to have the right perspective of such. I believe a spandrel in theory to be the potential minus the solution at hand; in certain problem spaces a spandrel may be present in the optimal solution. This is not to say a spandrel is useless, in that removing a spandrel with reference to some solution from the potential alters the problem space, thus in the probabilistic theory of evolution you would encounter a different spectrum of outcomes.
I would see non-overlapping magisteria to be more controversial, however.