I frame the analytic discussions about markets versus socialism as a knowledge problem. The key to the success in any “economic system” is to figure out a way to have valuable knowledge and information produced, transmitted and acted upon – even by actors who do not know that they are part of the solution. The standard discussion then revolves around how challenging it is to produce and transmit such knowledge, especially when it is tacitly held and dispersed.
Some may counter that the reason socialist planning does not work is simply a matter of computational power. Once AI and computers are good enough to figure out what everyone wants, when they want it, where they want it, and to figure out how to produce things, we can put the entire economy on autopilot and the government can plan away. This is mistaken for a few reasons, not the least of which is that the process does not allow for entrepreneurial dynamics.
But my point today is simpler. If in fact we get to a world where AI and computers and machine learning are powerful enough to do this, then of course we would also be living in a world where all of the “textbook” assumptions about the Coase Theorem hold. In other words, all transactions costs would drop to zero, potential negotiators would know who and where and how to negotiate, and therefore we truly can live in a laissez-faire world – just make sure someone has an initial property right, and we can seamlessly negotiate our way out of any social problem.
So, dear socialists, the coming computer revolution does not mean what you think it means – there is just as much reason to suspect that it makes a more transactions driven world easier to sustain. And certainly, if you ask people whether they cherish autonomy or some computer system optimizing on their behalf … well, OK, strike that.
