[jedharris]

But has anyone proved that markets actually do approximate the (intractable) correct allocation?

They could be a very poor approximation or even be actually computing something we don't want.

Furthermore if the point is to approximate an optimal allocation we should be designing them with that in mind. Auction design and a few specialized solutions such as medical residency assignment do this. But given your argument we should be applying this kind of design throughout the economy.

[chrisdevereux]

You're right, I really meant to point out that there's an interesting relationship between markets and computational complexity (and that it's odd for the efficient markets hypothesis if it assumes that P=NP) rather than advocate for markets being an unqualified good (I'd probably describe myself as some kind of Marxist, fwiw).

> But has anyone proved that markets actually do approximate the (intractable) correct allocation? They could be a very poor approximation or even be actually computing something we don't want.

> They could be a very poor approximation or even be actually computing something we don't want.

Well, it would be wrong to take the market allocation to define the desired outcome, but assuming we have some other understanding of what our desired outcome, we can at least assess whether it performs better or worse at the computation than some other approach, right?

It's interesting also that markets seem to share properties with other sub-optimal approaches to computationally hard problems. Getting stuck in local maximas, for example.