Assuming GP is talking about Von-Mises style criticism of command economies, it's not computational tractability, it's information tractability. The problem is

Assuming GP is talking about Von-Mises style criticism of command economies, it's not computational tractability, it's information tractability. The problem isn't that you don't know whether action A results in N units of X, it's that you don't know if the aggregate preferences are such that it's optimal to produce N units of X or M units of Y.