Roughly speaking McCloskey wants economics to make interesting, new, and true statements about the real world, and argues that proving the hypothetical possibility of an effect within an analytical framework is not a constructive way of doing this.
It has led to debates with such figures as Kenneth Arrow, who vigorously support the "Samuelson" approach, and argue that the quantity of analytical mathematical models in modern economics is a critical requirement for progress.
McCloskey says that most economists when they write are "tendentious", assuming that they know already, and concentrating on a high standard of mathematical proof rather than a "scholarly" accumulation of relevant, documented facts about the real world.
The advice she offers colleagues here is to spend more time in the archives, and write more heavily researched papers from specific observations in the real world (she argues that this is the norm in the natural sciences on which economics believes it is modelling itself, but that most economics practitioners actually base their methodology more closely on pure mathematics).
Calibrated simulation relies on measurement and numerical techniques (such as Monte Carlo methods) to test the robustness of its predictions, without requiring a closed-form solution proving that the postulated relationship will always hold (or will be reached in "equilibrium", or be impossible).