Calculation and Socialism | Joseph T. Salerno
In 1920, Ludwig von Mises destroyed the intellectual foundations of the case for socialist central planning.
Download lecture slides at: https://mises.org/MU23_PPT_12
Recorded at the Mises Institute in Auburn, Alabama, on 25 July 2023.
https://odysee.com/@mises:1/calculation-and-socialism-joseph-t.-3:2
I watch this lecture every year... doesn't get old.
Do Big Data and AI Solve the Socialist Calculation Problem? | Tate Fegley
“Our civilization is inseparably linked with our methods of economic calculation. It would perish if we were to abandon this most precious intellectual tool of acting.”
Download lecture slides at: https://mises.org/MU23_PPT_28
Recorded at the Mises Institute in Auburn, Alabama, on 27 July 2023.
https://odysee.com/@mises:1/do-big-data-and-ai-solve-the-socialist:8
This speech was OK, but I felt the critique could have been a bit stronger.
Let's start by trying to make myself a competent central-planner. I have one very valuable asset, which is that I am human, so it is easy for me to intuitively understand the kinds of things you value, even though I'm not you. With only a modest amount of information about you, I can make very good guesses about your relative valuations across a wide variety of goods and services. If you are a fan of the outdoors, I can guess that you will value sports footwear more than office footwear. And so on. The reason I can make all these "pretty good guesses" is that I have an intuitive model of what it's like not only to be me, but also some ability to make accurate inferences about others based on my social experiences and my ability to empathize (to imagine what it would be like to be in your shoes).
Imagine that I have a time-machine that allows me to slow time, for everyone but myself, by a factor of, say, 1 billion. So, if I can examine the lives of 7 people per day then, after 1 billion days, I will have examined everyone on planet earth, for one day. Using this immense amount of investigation, I construct a "central plan" that truly factors everything into it... I thought about whether you prefer thin socks or thick socks, I thought about whether you have a melancholic or jovial disposition. I use my own onboard "model of a human" to intuitively "guess" what each person in the world values most, based on the available data (I have access to all records, information, etc.) I'm like some kind of time-traveling Santa Claus who is able to go around the whole world in a single day, taking all the countless subtleties of human valuation into account. To keep all that information organized, I have access to an AI even more powerful than ChatGPT which can call up all the world's information in an instant, and an enormous database where the results of all my calculations are saved.
After 1 billion days of doing all these calculations, I generate an enormous data-file that is sent to the Commissars of the Global Politburo. They rigidly implement all the results of these calculations and the right number of sports shoes and office shoes get produced, etc. based on the data (not prices, of which there are none). But notice something interesting -- the
best I can possibly do is as good as the unhampered free market would have done, if we had prices. The reason for this is that each of the 7.7 billion people on the planet are the "gold-standard" of their own valuations... the
best I can do in approximating those valuations, even one-by-one, is what the free-market would have done, with prices. By definition, I can't know what you want
better than you do, because what you want is the definition of what you want! If you prefer thin socks to thick socks, the
best I can do is correctly guess that you prefer thin socks to thick socks. If I get it wrong (and I will get it wrong some amount of the time, no matter how capable I am in guessing your true subjective preferences), I make things worse... but even if I get it right, I only do as good as the market would have done.
This shows two important points. (a) Socialism is inherently normative in respect to individual valuations. The whole
point of socialism is to tell you that you want the wrong things. "You prefer to drive an SUV to other types of vehicles...
but you shouldn't." That's the whole point! (b) Socialism is fundamentally incapable of improving conditions relative to the unhampered market because, even when we grant every conceivable advantage to the central-planner, he can only do
as good as the market would have done, never better.
In the thought-experiment above, I was slowing time and using my own human intuitive model of the values/preferences of other people in order to construct some kind of proxy table of valuations for every individual over the whole world. Note that substituting an AI in place of my human brain changes nothing in respect to the achievable limits of the thought-experiment. The AI,
at best, perfectly models every individual's subjective preferences the world over. It then crunches all the factors of production, armed with all information about every available resource in the whole world. It does this at blinding speed, billions of times faster than a human could, so it is able to complete all of that planning in 24 hours, and sends the data-file to the Commissars of the Global Politburo to be put into place for the next day. But again,
the best case is that it has arrived at exactly the same coordination of economic resources as the market would have, with prices. You're just doing the very same thing in the hardest possible way. And that's assuming the absurdity that you have access to practically infinite computing power, and so on, and so forth. The only change that the super-duper-AI-socialist-central-planner-brain can make, relative to the unhampered free market (as measured in terms of the subjective valuations of the global population) is to make things worse than they would have been otherwise. It
cannot improve them.
Another argument that might be raised is that, if the AI can build perfect models of consumers' subjective valuations, then it could outperform human entrepreneurs in the process of economic calculation. Armed with higher intelligence than humans, and better modeling capability, it would be able to outperform human entrepreneurs in the market. This is conceivable, but we don't need to "scrap capitalism" in order to find out whether this is the case, we can test it
in situ by simply having entrepreneurs compete with AIs. Whoever wins is, by definition, better. So, perhaps it turns out that AIs are hyper-competent at forecasting agricultural futures. Then the outcome of this is that humans would stop trying do agricultural futures by human reckoning, and would just use the AI for that. But the choice to use an AI, or to pay human forecasters,
is still an entrepreneurial one. So, once again, AI doesn't bring any kind of secret-sauce to the problem of economic calculation. It might do some economic calculations better than humans can, but it is then just part of the very same price system which we already showed above is the best achievable, in terms of satisfying human subjective valuations through price calculations. In the end, AI is nothing more than a tractor for the mind...
