In a conversation with a facebook friend (who is largely on the same page on this), I had this to say and thought it was worth sharing here:
"The Walrasian auctioneer picture of the economy is OK, but it's a very incomplete view of how humans interact when it comes to allocation and production decisions. I think Coase does a great job filling in the blanks left by Walras. The problem comes in, I think, when Coase is treated not as a way of filling in Walras, but instead treated as a counter-argument to Pigou."
I would add that just because Coase is right, that doesn't mean that Walras is wrong. Often when we see an incomplete explanation economists have a tendency to turn their nose up at the simple version.
That's dumb, I think.
Walras is great. Neoclassicism is great. It's just not the whole story. It's wrong to think about some kind of institutionalist story as displacing them.
You see this with information theory too. People look at work by Stigler, Stiglitz, and people like that and say "Ha! They're only dealing with risk! I plumb the depths of fundamental uncertainty, etc. etc.". If you want to talk about fundamental uncertainty that's all well and good. Good luck (you'll need it!). But let's please not act like risk is some trivial thing to look into! Those risk-based investigations were still really hard to work out and major advances in our thinking, and they are very relevant to a lot of problems we deal with!