Wednesday, August 28, 2019

Some Confusion About Intellectuals and Socialism

Over at Target Liberty, at the post, Friedrich Hayek: Why Intellectuals Drift Towards Socialism, where Hayek discusses the improper thinking that occurs amongst many intellectuals about what can be planned, a commenter wrote:
"People who are smart tend to believe they can engineer others' actions to a better outcome. However, since they don't have the sensibilities of the others -- i.e. most of economics is about WANT, not NEED -- they fail miserably.
"The free market is a neural network, and intellectuals try to apply sequential programming to it. It's like trying to replace your brain with an IBM chip running Microsoft. It's the wrong architecture."
There is a mix of some truth and a lot of confusion in this comment.

It may be true that many intellectuals believe they can engineer others' actions to a better outcome but the second part of the comment "they don't have the sensibilities of the others -- i.e. most of economics is about WANT, not NEED -- they fail miserably," is not the problem.

The problem is there is too much information, changing all the time, that makes central planning unsuccessful, even if individuals "sensibilities" matched up with the general public. In short, it is not the type of person doing the planning that is the problem but that no person, group, or computer could successfully do central planning.

Finally, calling free markets a "neural network" confuses things even more. It suggests that free markets can be modeled, at least on a theoretical basis, to reach some sort of equilibrium but this is not the case at all. The free market is about adjusting to circumstances that are always changing and no neural network or sequential programming can, so to speak, "solve the problem" because the data is always changing.

-RW

2 comments:

  1. "The problem is there is too much information, changing all the time, that makes central planning unsuccessful, even if individuals "sensibilities" matched up with the general public. In short, it is not the type of person doing the planning that is the problem but that no person, group, or computer could successfully do central planning."

    That was my point, both in the sensibilities comment and the neural net comment.

    Individuals may not even completely understand all the sensibilities they bring to the table. There are billions of individuals. No one can successfully model this.

    The economy is a self-organizing neural net. I don't think there's a lot of doubt about that. Each individual acts as a neuron with many more connections than a human brain. The interactions between the neurons fire when the weighted sensibilities of the individuals trigger an interaction. You could potentially model it if you could collect enough information to do so. Maybe there will be a time in the future when that is possible, but it's not possible now.

    ReplyDelete
  2. The whole idea that collecting detailed economic information could help (with the aid of whatever mighty computers, models, and algorithms) to manage the economy is totally idiotic. The very process of "managing" (i.e. regulating and manipulating to prevent some trades and artificially force and encourage other trades which wouldn't happen in a free market) is destroying the information about subjective utilities of market participants. The more you "manage" the less actual economic information you get - so the computers and whatever basically get garbage as inputs, so they produce garbage as outputs. The purely managed economy is flying blind. (Of course, this is just the good ole' economic calculation problem restated in modern terms).

    The "neural net" analogy is just ridiculous. If you want AI analogies, the market is a distributed evolutionary optimization algorithm using prices and division of labor to split the impossibly complex global optimization problem into manageable chunks, along the lines of the Hayek machine by AI researcher Eric Baum who demonstrated experimentally how simulated market economy of really stupid evolving programs can solve problems which are intractable to the simple evolution (such as evolving recursive solver to Tower of Hanoi problem).

    Even more immediate cybernetic analogy to the market and prices is the concept of homeostasis based on negative-feedback loop (i.e. the way the prices adjust to remedy shortages or excess of production and consumption), though this analogy does not address the information problem: people like Soviet cybernetist Victor Glushkov argued for national computed-mediated economy management (project OGAS) using negative feedback loops to stabilize the socialist economy. Needless to say this had no chances of working. I guess he never heard of von Mises.

    ReplyDelete