The fundamental problem with mainstream macro economics seems to be the assumption that markets tend toward "equilibrium". This doesn't get questioned because the method of model first, observe second, is strongly supported as the only proper methodology. My working theory is that all the "victories" claimed by this theory/method are illusory because they are simply triumphs over other economic theories that create confusion. We only need Krugman, DeLong, and Simon Wren-Lewis because of Fama, Mankiw, Taylor and Hubbard.
But I must say that I have no way of knowing if the sample seen in the econoblogosphere is representative. There are strong hints that econ as a whole maybe more empirically based than I give it credit for:
I'd say the rest of economics is doing much better than macro. A huge amount of current research is solidly empirical based on experiments or natural experiments. Research doesn't need a general theory to progress. We can learn what sort of policies work without theory. I am thinking of, say, does preschool cause better life outcomes, does means tested cash assistance to the poor create dependency which lasts generations, and can people without medical degrees who harass (da noia a) other people about their weight, blood pressure, diabetes and blood cholesteral save lives ?