K-k-k. There is in science and engineering a distinction between pure science, which is descriptive in character, and applied science and engineering, which use the descriptions of pure science to some goal. In economics, this distinction is often lost. There is a descriptive social science which might be called pure economics and an applied practice of economic policy making which doesn’t have a separate name. I think, as a matter of scientific practice and ethics, it is important to clarify the distinction.
Equally important, and widely overlooked, are the scientific basics of explicitly stating the assumptions of theory and checking theory against experimental data. In the area of stating assumptions, economics fares poorly: economic behavior occurs in a social context, and it is part of the assumptions of any economic theory (= model.) Without stating the social context of an economic result, one can get remarkably boneheaded results: as Galbraith points out, one can end up assuming that people negotiate for jobs in the same way they haggle over fish.
Which brings us back to Krugman. Points in Krugman’s favor which I don’t think you’ve covered: he is aware of the distinction between pure and applied economics, he is willing, after a struggle to be sure, but willing, to check his theories against experimental data, and he accepts that people behave differently in different markets.
On the other hand, he himself has acknowledged a theoretician’s weakness for elegant general theories and he is not a strong applied economist. Since his time with the Reagan administration, he has refused policy-making positions, and I think this may be why.