When Paul Krugman stated that he thought the stimulus was possibly too small perhaps he thought just that. But perhaps he also thought he was rather transparently laying the ground work for an immunizing strategy to counter another immunizing strategy. The end of his post seemed to indicate as much. The strategy was this: If the economy did not improve, he (Krugman) could claim that was because the stimulus was too small not because stimulus does not work. But then surely the door swings both ways. If the economy does improve, as in something that approximates pre-boondoggle (recession) rates of growth, then those who argue against the effectiveness of stimulus can simply claim that even according to proponents of stimulus spending its magnitude was too small. Hence it was not the stimulus that “caused” the recovery.
In any case, what the debate between reform and neo-classical liberals (freshwater and salt water, inversely respectively) strains of scientific liberalism makes clear is that each side is well immunized and there is not any definitive test.
Some would conclude that this marks-off a degenerative research programme (i.e. neoclassical economics writ large). I would argue it is the difference between open and closed systems. Open systems do not admit to definitive tests. This was perhaps always the crucial weakness with the popular Popper. The Social system of which the economy is a non-trivial subsystem is notoriously open ended which means that it is difficult if not impossible to validate or invalidate big propositions. This is not to say truth does not exist in the social world it is just that we do not get to access it in any direct and irrefutable way.
And this existential proposition is the most robust and well tested finding of the social sciences and humanities contrary to Cochrane’s belief that such a distinction belongs to the Efficient Markets Hypothesis.