Ludwig von Mises’s essay “Economic Calculation in the Socialist Commonwealth,” references Aristophanes’ play The Birds and the medieval fantasy of the idyllic and work-free Land of Cockaigne when Mises notes of socialist planners that, “Economics as such figures all too sparsely in the glamorous pictures painted by the Utopians. They invariably explain how, in the cloud-cuckoo lands of their fancy, roast pigeons will in some way fly into the mouths of the comrades, but they omit to show how this miracle is to take place.”[1] Don Lavoie similarly points to the science fictional/fantastical aspect of socialist planning discussions when he comments in Rivalry and Central Planning that, “Details of future social life are not the province of economic science but of speculative literature.”[2]
Certainly one feels the fantasy novel being written in Michael Albert and Robert Hahnel’s article “Participatory Planning,” as they outline the precisely thought-out details of how their version of socialist planning will function:
Every individual, family, or living unit would belong to a neighborhood consumption council. Each neighborhood council would belong to a federation of neighborhood councils the size of a city ward or rural county. Each ward would belong to a city consumption council, each city and country council would belong to a state council, and each state council would belong to the national consumption councils. One reason for the nesting of the consumers’ councils is to allow for the fact that different kinds of consumption affect different numbers of people. The color of my underwear concerns only me and my most intimate acquaintances. The shrubbery on my block concerns all who live on the block. …Real national security affects all citizens in a country and protection of the ozone layers affects all humanity––which means that my choice of deodorant, unlike my choice of underwear, concerns more than me and my intimates! [3]
That is, for those who have lost count, at least five levels of councils, each of which will presumably have regular meetings to manage the work produced for them by the other councils. The socialists are certainly going to be busy. And they would have to be, since something as personal as deodorant will be a matter, apparently, for public discussion by the council. (One wonders, incidentally, how Albert and Hahnel can so callously dismiss the immense importance of choosing unbleached, organic, fair-trade cotton for one’s delicates. But their neighborhood council will presumably deal with this lapse in social responsibility.)
The fantasy here is the “Cloudcuckooland” dream that these meetings will be frictionless, will avoid endlessly-offered opportunities for rent-seeking and log-rolling, will not devolve into a rule by those who have the most time to attend meetings, or the greatest facility with language, or simply the loudest and most intimidating voices. The fantasy is that enough meetings will allow us to overcome human nature. That is why plans like Albert and Hahnel’s are easy to mock. Seeing their many flaws requires no more experience or common sense than one can gain by attending one or two local planning meetings, or PTA gatherings.
Perhaps less transparently problematic is the increasing focus on a different kind of fantasy of how we can make socialist planning really work this time. That is the fantasy of the omniscient computer that automates planning and removes human flaws from the process. It is easy, and I confess, tempting, to mock this as a piece of Star Trek fan fiction gone wild. However, the argument deserves to be taken seriously on its own terms––not just because of its potential technical challenge to classical liberal objections to socialist planning, but also because of its moral challenge.
Allin Cottrell and W. Paul Cockshott pointed out in “Calculation, Complexity, and Planning” that those opposed to central planning have tended to dismiss the argument that in order to work, socialist planning just requires increased ability to do enough calculations fast enough and correctly enough. Now, however, “The ‘computational argument’ is relevant, and…recent advances in computer technology do make possible an effective socialist planning system.”[4] Advances in brute computation are increasing at a mind-boggling rate. As Christopher Lee noted in the Washington Post:
The first “petascale” supercomputer will be capable of 1,000 trillion calculations per second. That’s about twice as powerful as today’s dominant model, a basketball-court-size beast known as BlueGene/L at the Energy Department’s Lawrence Livermore National Laboratory in California that performs a peak of 596 trillion calculations per second.
The computing muscle of the new petascale machines will be akin to that of more than 100,000 desktop computers combined, experts say. A computation that would take a lifetime for a home PC and that can be completed in about five hours on today’s supercomputers will be doable in as little as two hours.
So we can do more math, and we can do it faster. But it is particularly the advances made in neural networks where twenty-first century planners seem to put their faith. These advances have begun to allow computers to better approximate biological brains. They are, very roughly, an attempt to abstract the complexities of biological neural networks and create an artificial version that will allow the computer’s processors to focus on the most important aspects of the given information and to make decisions without needing to “do the math” all the way down.
Cottrell and Cockshott point to the way a butterfly presumably assesses costs and benefits when hunting nectar. “It appears that neural networks are capable of producing optimal (or at least highly efficient) behavior, even when faced with exceedingly complex constraints, without reducing the problem to the maximization (or minimization) of a scalar.” That such work can be successfully done by a butterfly for its own purposes, they argue, suggests that it can be done on a large scale by a large computer for general economic purposes. That is to say, if one “wishes to perform global optimizations on the whole economy, other computation techniques, having much in common with the way nervous systems are thought to work, may be more appropriate, and these can in principle be performed without resort to arithmetic.”
Good enough technology, in other words, eliminates the need for math, and eliminates the argument that socialist planning is a technical impossibility.
Arguments like Cottrell and Cockshott’s, it seems to me, present a more important challenge for twenty-first-century classical liberals and libertarians than the easily mocked arguments of old-style planners. As we are often fans of innovation and technological progress, it can be a bit tempting to place a lot of hope in technical solutions. So the question then becomes a dual one. The first is the same as it has always been: Is the technology, as yet, up to the task? A random sampling of the scientists I know indicates that so far, it is not. But they were not inclined to dismiss that possibility for the future, given how rapidly our technical capabilities are expanding. That raises the second and much more important question: Even if it were possible for a computer to work as Cottrell and Cockshott’s argument would require it to work, should we use it that way? What about that choice would be moral or immoral? Liberating or enslaving? That should be the new focus of the old socialist calculation debate.
[1] Mises, Ludwig von, “Economic Calculation in the Socialist Commonwealth” In Collectivist Economic Planning: Critical Studies on the Possibilities of Socialism, edited by F. A. Hayek, 87-116. London: Routledge, 1947
[2] Lavoie, Don. Rivalry and Central Planning: The Socialist Calculation Debate Reconsidered. New York: Cambridge University Press, 1985.
[3] Albert, Michael and Robin Hahnel. “Participatory Planning.” Science and Society 56, no. 1 (Spring 1992): 39-59.
[4] Cottrell, Allin and W. P. Cockshott. “Calculation, Complexity, and Planning: The Socialist Calculation Debate Once Again.” Review of Political Economy 5, no. 1 (July 1993): 73-112.
Sarah Skwire is a Senior Fellow at Liberty Fund, Inc., and she regularly publishes at Econlib and AdamSmithWorks.
READER COMMENTS
Chris
May 24 2019 at 2:08pm
‘That is to say, if one “wishes to perform global optimizations on the whole economy, other computation techniques, having much in common with the way nervous systems are thought to work, may be more appropriate, and these can in principle be performed without resort to arithmetic.”’
Optimize for what?
RPLong
May 24 2019 at 3:10pm
It is interesting to compare the socialists’ computing fantasies to the future world Michio Kaku talked about in his keynote speech at this year’s SAS Global Forum. There, he suggested that data, AI, and increased computing power would rather make capitalism frictionless. He called it “perfect information,” which would lead us to “perfect capitalism.” Prices that can adjust instantly to any new information, and none of that information asymmetric.
I have to say, I rather prefer Kaku’s telling of it.
Matthias Görgens
May 25 2019 at 9:11am
Matt Levine has an interesting and somewhat ironic account how these developments and increasing shift of investments into passive index funds might lead to a better, frictionless capitalism that might look almost like a socialist Utopia.
(He’s context was index funds that use eg factor models to model some traditional investment strategies. And an investment strategy also doubles as an opinion on capital allocation.)
John Alcorn
May 24 2019 at 3:30pm
A society consists of individuals, each with his or her own motivations. Inferior incentives, rather than putative computational limits, are the Achilles heel of socialism. Compare Bryan Caplan, “Is socialism really ‘impossible’?,” Critical Review 16 (2004) 33-52; available gratis here.
Two intuitions:
1) The real action, within prosperous societies, in changes in the nature and scope of planning will occur at the shifting boundaries of the firm. See Ronald Coase, “The nature of the firm,” Economica 4:16 (1937) 386-405; available gratis here.
2) Insofar as socialists are motivated by altruism, let’s patiently explain that market prices are crucial to effective altruism:
robc
May 24 2019 at 4:22pm
I would like to know how today’s computers, had they existed in the 70s, would have gone about predicting the craft beer market.
Don Boudreaux
May 24 2019 at 6:07pm
Sarah,
Nice post.
But I disagree that arguments such as Cottrell’s and Cockshott’s present a greater challenge to libertarians and classical liberals than was presented by planning enthusiasts of the past. The ultimate problem isn’t inadequate computational speed; it’s failure of any system other than one that involves voluntary exchanges of personal property (usually in the form of money) to reveal and to convey the relative values of alternative uses of resources and of alternative means of producing outputs.
Thomas Sewell
May 24 2019 at 7:45pm
Even if the technology reached the point where it could faithfully simulate all the decisions an individual human made about their own preferences in order to make them for them (and it’s probably 50-100 years at least away from that), “planners” are still left with at least two issues:
Why create a simulation which constantly updates from the original to express preferences when we have the real original available to express their preferences just as effectively? I could see some minor time/effort savings involved, but not much more.
If something other than the individual’s preferences are to be considered and “optimized”, then what’s the basis for deciding that and why wouldn’t it be an immoral imposition on them?
Sarah Skwire
May 30 2019 at 9:42am
Don-
I agree. I’m not persuaded by the challenge. But I think the people who pose it think it’s a great argument. Therefore it’s probably a good idea for us to treat it as if it is, and explain (again) why it won’t work.
Benjamin Cole
May 24 2019 at 7:07pm
Interesting post. I think I prefer free markets but it might be worthwhile to ponder how a Red China, with a centrally planned economy, could deploy computing power to obtain good results ( in combination with Han culture).
robc
May 24 2019 at 10:31pm
Easy. They can’t.
Matthias Görgens
May 25 2019 at 9:12am
China is no longer centrally planned.
(And never fully was to the level of detail the fans of the computer have in mind here.)
Matthias Görgens
May 25 2019 at 9:18am
Even assuming we had a computer and a program powerful enough to solve these problems: a big question is still how to design the mechanism/plan to be incentive compatible.
Ie how do you set up things such that telling the computer your true preferences is your optimal strategy? And how do you set things up so that working to the full extent of your abilities for society (instead of slacking off), is also your optimal strategy?
Economists have a lot to say about these kind of incentive compatible systems. I suspect they might look like the kinds of auctions Google and Facebook etc run to eg allocate ad-space and reward the providers (after taking a cut).
There’s certainly an element of central coordination and even planning there. But it’s nothing like what people normally envision when they talk about central planning.
Phil H
May 26 2019 at 4:40am
This is well-written, but not very convincing. It’s amusing enough to imagine committees deciding the colour of underwear, but one can quite easily write similarly absurd-sounding account of how underwear colour is decided in the real world: “Lacking any way to communicate their preferences to the underwear manufacturers, shoppers drive half an hour to large markets simply hoping that the colour they want will be available; keen to retain their custom, market owners will stock dizzying ranges of colours and styles, leading to significant waste, but often failing to address simple issues like variation in skin colour, because market procurement specialists fail to engage with shoppers.”
In reality, the multiple levels of committees is how many things get decided, including the law, and the internal working of many large companies. The multinational I worked in set big-picture strategy at head office, had some regional policies, then strong national teams that decided implementation in their own country. This system was clearly working for them.
In reality, both the committee and the free trade models fail when they assume that preferences are fixed. A superior way of looking at it is: for public goods, and those goods where preferences change slowly, a committee model will be necessary; for private goods, and sectors where variety and innovation are desirable, open markets and competition are better. The arguments are over which sector falls into which type.
For example, in healthcare, variety is not desirable: everyone wants the best available medicine. But innovation is desirable, which is why the status of this sector is so bitterly contested.
robc
May 26 2019 at 10:25pm
The concept of personalized medicine is that variety is not only desirable, but necessary, so the health care industry shouldnt be contested at all.
Mark Z
May 26 2019 at 11:40pm
First, there’s what robc said. But even beyond that, we should in fact want quality of medical care available.
If a billionaire wants to pay thousands of dollars per hour to get a checkup from a Nobel prize winning physician (i.e., the best medical care) that’s fine; if I’d prefer to get my checkup at a much lower price from a nurse practitioner, I should be able to do that as well.
More realistically: there are tests (e.g., biopsies for polyps; or rather, varying degrees of liberality on the part of physicians in ordering biopsies) that are very expensive and may improve the ability to detect cancer by very little. I don’t see why either 1) people should be compelled to purchase such tests if they don’t think it’s worth it to them or 2) be prevented from purchasing it if they have the money and the slight increase in detection probability is worth it.
Healthcare isn’t that exceptional here (though it’s made to seem like it by all the restrictions and institutions that have largely eradicated both consumer choice and incentive to care about costs of specific treatments). People differ in how much they value and are willing to pay to improve disease detection, diagnosis and treatment by some marginal amount, just as they vary in how much they’re willing to pay for a safer car, healthier food, a more reliable home alarm system, etc.
Phil H
May 27 2019 at 5:06am
Robc, I’m afraid I think you’re just being dogmatic in an unhelpful way. Healthcare is highly regulated and/or government-run in every developed country. That’s objective fact. Now, you claim that your theory is better than the evolved practice of every advanced healthcare system on earth. Why should anyone believe you? Your claim isn’t impossible – sometimes things have to change. But it’s fantastically unlikely, particularly given that the one supporting claim you made is obviously wrong. The greatest health intervention of the modern age, without a shadow of a doubt, is vaccination, and the one key property of vaccination is that it’s not personalized. The next health revolution may be personalized medicine; but it’s equally likely that it won’t be.
Mark – you said healthcare isn’t exceptional, but it is in one important way: there are objective criteria for success. Getting a checkup from a Nobel prize winner is objectively not better for you than getting one from a local nurse, and there’s evidence to prove it. No similar objective standards exist in fashion or food or cars. But doctors are committed to this objectivity, because it has brought real benefits. Restaurateurs couldn’t give a monkeys about objective standards, I mean, Michelin stars are fun, but they’re nothing compared to bums on seats.
So there is a real difference to be negotiated. Denying that difference doesn’t make for a sophisticated argument!
Tom DeMeo
May 27 2019 at 10:48am
Seems like you’ve shopped for an easy argument to win.
Explain China and get back to me.
Comments are closed.