You also know some people who desperately need a course in computational complexity. Markets aren’t perfect, of course, but good luck trying to centrally compute distribution of resources.
The various hardness results for economic problems (e.g. computing Nash equilibriums is PPAD-complete) cuts equally much against free markets as against central planning. If a central agent can’t solve a given problem within cosmological time scales, then neither can a few billion distributed agents.
The free market doesn’t solve the central planning problem. It reliably climbs local hills in the solution space by putting more decision-making ability (money) in the hands of those who make better decisions (make more money).
Free market, in a typical situation, has the advantage of having more raw computing power, simply because every person uses their own brain to optimize for themselves. And (some people believe that) the benefits of this additional computing power overweight the costs of not having this computing power coordinated, thus wasting a part of it. (It also has the advantage of having local information, having access to this information before it was filtered by political processing, etc.)
But technically, we are speaking about a linear increase in the computing power here. Like, a few million people, instead of a few dozens of government experts. Computational complexity typically does not speak about linear factors. -- Thus you have sinned against the narrow meaning of “computational complexity”. I believe the downvotes reflect this.
That doesn’t sound like a terribly good argument—the fact that it would take O(something big) time to compute something exactly isn’t terribly important if you can compute a decent approximation in O(something small) time.
My model of how to approximate the optimum solution—specifically, break it up into tiny pieces and keep track of price-analogues by means of a real number valued function of the industrial output being managed—looks an awful lot like a free market with really weird labels for everything. It goes up to and includes closing sub-units that detract from overall optimization (read: unprofitable firms).
You also know some people who desperately need a course in computational complexity. Markets aren’t perfect, of course, but good luck trying to centrally compute distribution of resources.
The various hardness results for economic problems (e.g. computing Nash equilibriums is PPAD-complete) cuts equally much against free markets as against central planning. If a central agent can’t solve a given problem within cosmological time scales, then neither can a few billion distributed agents.
The free market doesn’t solve the central planning problem. It reliably climbs local hills in the solution space by putting more decision-making ability (money) in the hands of those who make better decisions (make more money).
Free market, in a typical situation, has the advantage of having more raw computing power, simply because every person uses their own brain to optimize for themselves. And (some people believe that) the benefits of this additional computing power overweight the costs of not having this computing power coordinated, thus wasting a part of it. (It also has the advantage of having local information, having access to this information before it was filtered by political processing, etc.)
But technically, we are speaking about a linear increase in the computing power here. Like, a few million people, instead of a few dozens of government experts. Computational complexity typically does not speak about linear factors. -- Thus you have sinned against the narrow meaning of “computational complexity”. I believe the downvotes reflect this.
That doesn’t sound like a terribly good argument—the fact that it would take O(something big) time to compute something exactly isn’t terribly important if you can compute a decent approximation in O(something small) time.
(I’m not a Communist, FWIW.)
My model of how to approximate the optimum solution—specifically, break it up into tiny pieces and keep track of price-analogues by means of a real number valued function of the industrial output being managed—looks an awful lot like a free market with really weird labels for everything. It goes up to and includes closing sub-units that detract from overall optimization (read: unprofitable firms).