Much as I enjoyed Unlearning Economics' riposte to Chris Auld, I fear such meta-critiques of mainstream economics miss a trick - the extent to which good "normal science" in economics can be the ally of challenges to neoclassical economics.
Let's take one recent example. Experimental work by Matteo Ploner shows how people's investment choices are shaped by peer pressure. Now, the standard criticism of experiments in economics is that they lack external validity. But this mightn't be the case here. Ploner's work corroborates detailed empirical work by Per Ostberg and Hans Hvide which has found that workers' share buying is influenced by their co-workers.
This strengthens the challenge to the rational markets hypothesis. Sure, Robert Shiller showed years ago that shares are more volatile (pdf) than the "fundamentals". But this, in itself, doesn't establish that markets are irrational. As Davidf Meenagh has shown (pdf), small and reasonable (but mistaken) changes in the probabilities attached to booms and busts can also generate what looks like excess volatility. What these papers do, though, is challenge Meenagh's reply. They give us a strong reason to believe that excess volatility is a sign of irrational markets. Bubbles can happen if some peoples' buying leads other to buy, and Hvide and Ploner have shown that such behaviour is widespread.
Sure, their work doesn't suffice to show that irrational bubbles can happen. To do this, we need to show that the "smart money" does not short-sell aggressively when (aggregate) stock markets seem over-priced. And Bernard Dumas and colleagues have done just this, by using the sort of general equilibrium modelling favoured by the mainstream.
What we have here, then, is an example of how a challenge to neoclassical economics is buttressed by good empirical work; Ploner and Hvide provide microfoundations for the idea that markets can be irrational. They do so not by big-think theorizing about the nature of rationality, but by ordinary empiricism - looking at the data and running experiments. And their work is, I think, reasonably typical of what many economists actually do.
In this sense - and I think I could produce other examples - leftist critics of neoclassical economics are under-estimating the extent to which economic science can support their case. There's nothing so radical as empiricism.