Thaler on Friedman on Assumptions

Richard Thaler examines the assumptions of economic models in the introduction of his book The Winner's Curse. He writes that "Friedman's position is that it doesn't matter if the assumptions are wrong if the theory still makes good predictions". I would say that this is the most common interpretation of Friedman's 1953 paper "The Methodology of Positive Economics".

However, I interpret the article somewhat differently. I read Friedman's argument to say that a model which is perfectly realistic would include everything in the world and would thus not be very useful. That is, a model, by definition, is less than perfectly realistic. As such, assumptions must lack realism to some extent. The extent to which assumptions conform (or should conform) to reality can be determined by the predictive power of the model. Friedman believes that realism has merit but that total realism is not possible. This is quite different than saying that the realism of assumptions is unimportant. I look forward to hearing Thaler's discussion, but I wonder if he will be attacking a straw man.

(Also, I believe the greater question to pose to Friedman's 1953 work is whether or not prediction is the same as explanation. Presumably, it is the economist's job to explain the world. )