I grabbed this book because it was making a minor stir in the blog-o-sphere.
It was kind of a let down. Rodrik is a bit of a rebel when it comes to economics discourse (once wrote a book doubting fully the benefits of globalization *fainting couch). But even then the options that he looks at seem narrowly circumscribed - mostly macro from new Keynesian to new classical! To his credit, Marx raises his head, if only to be dismissed.
What I really like is the framing of the book as looking at models, and how they are useful and how they are not as necessary simplifications of the real world - he even cites one of my favorite Jorge Borges stories to tell about the necessity of simplification. He also looks at the fox / hedgehog divide, in that true knowledge is knowing what model to use when looking at the explanation of a past event, and not over reliance on a strong theory that will too often lead you astray.
The big problem for me was that the model using is almost all backwards looking. If you know what happened and you can pick the right model you’re golden, but he discounts being able to know the future. Maybe I missed something, but isn’t a large part of trying to understand the past so that you can more accurately predict the future? If there is no ultimately one correct model and they do grow horizontally (a multiplicity) instead of replacing dead models as he calls it vertically, are we doomed to be having the same arguments generation after generation only with more powerful computers and fancier math?