Complexity science in the name of politics: a travel diary

This is a reading diary of a naïve complexity / computational social scientist’s first encounter with F. A. Hayek + Eastern Bloc tektology & cybernetics.

You might have heard about project Cybersyn? In 1970s Chile, Salvador Allende’s socialist regime was betting on a systems-theoretical approach to the complex decision-making that’s an inevitable consequence of a planned economy. Captain Kirk would feel at home as operators were rocking the latest cybernetic algorithms. That all ended with the US-backed military coup that installed Augusto Pinochet as a dictator. But was that also the end of complex-systems ideas in Chile’s government? Oh no! Pinochet called in economist F. A. Hayek—a vanguard of complexity economics and a firm believer in emergence and self-organization for the social good—to kick-start a Chilean free-market economy.

In its broadest sense, is complexity science spacious enough to support any ideology? Curious about that, I went on a weeklong tour to the border zone between complexity science and Austria, with a brief detour to Novosibirsk. Honestly, I’m still very much a novice at this topic—take this blog post as a travelogue. Hopefully, I will return to the theme with something more erudite.

From Stafford Beer’s 1975 Platform for Change. Illustrating all data a person leaves behind, that by cybernetic algorithms could be used to fine-tune society. Beer was one of the architects behind Project Cybersyn.

Starting point

Before this journey, I had seen Hayek mentioned as a complexity-science forerunner, one of the first to see emergence (“spontaneous order”) as a key to understanding society. Such sources also described him as a liberal ideologue and staunch advocate of the free market. Sometimes, I also read his Wikipedia article and remember his controversies: racist comments, defense of Pinochet’s dictatorship, and the condemnation of the international sanctions of South African apartheid. Oh, and that the apex of his life was when he met Queen Elizabeth II. Hmm, well, let’s go on with the science.

Tuesday: Hayek’s Nobel prize lecture “The pretence of knowledge”

Where else to start? This was Hayek in the limelight, touching on many of his ideas about complexity. This “pretence of knowledge,” Hayek argues, comes from approaching the social sciences like natural science. Hmm, this was my first surprise. (I later learned that rejecting formal models is a hallmark of the Austrian school—to which Hayek is usually counted.) Why, then, would he be a pioneer of complexity science (that wholly embraces the natural-scientific method)?

Hayek’s argument, same as published decades earlier in The Counter-Revolution of Science is that social processes are complex, with many hidden mechanisms that render them unpredictable from observables. Trusting data without understanding the mechanisms could lead to faulty conclusions. It is hard to disagree with that.

Hayek’s main target was orthodox, neo-classical economics, and his arguments are similar to Arthur-style complexity economics would use decades later. Would he be OK with the mechanistic modeling of the latter? Well, Hayek managed to lambast the (then two-year-old) Limits to Growth—unsurprisingly, since it calls for state interventions at many levels. Still, the systems dynamics model Limits to Growth relies on is not too far, in spirit, from modern complex systems modeling. On the other hand, Hayek uses “organized complexity,” referring to Warren Weaver’s 1948 paper where he predicts the aid of computers and interdisciplinary collaborations will be how to elucidate complex systems in the future. Hmm, how does this all come together?

Wednesday: Axtell’s “Hayek enriched by complexity enriched by Hayek”

Next stop, a paper by Robert Axtell, whose writing I’ve always admired. (The time I got to ride in Axtell’s vintage sports car, I probably felt as starstruck as Hayek meeting Queen Elizabeth.)

The essay’s premise is that complexity science could have helped Hayek and can still learn from Austrian economics. Axtell assumes that Hayek’s aversion to mathematical modeling does not extend to, e.g., agent-based models (ABMs) and proposes to rebuild Hayek’s theories within the framework of complexity economics. Thereby Austrian economics would have a sporting chance to break the orthodoxy because “it takes a model to beat a model.”

Axtell also claims that knowledge from Hayek would benefit today’s agent-based modelers. He lists some insights—e.g., “the conventional theory of general equilibrium is sufficiently incredible computationally that almost any market process story is preferable”—none of which seem unfamiliar to complexity scientists. On the other hand, it is reassuring that a prominent scholar of a different era and from different premises has reached the same insights.

Thursday: Hayek’s “The theory of complex phenomena”

From the title, this seems like a must-read for complexity scientists, but it turned out somewhat disappointing. It is not about how to theorize complex phenomena but how to not misunderstand them. Hayek points out that “statistics” of a few variables will not produce any laws like in the natural sciences since there would always be other processes lurking in the background. (Maybe Hayek would have liked deep learning more than regression.) Furthermore, statistical theories of complex phenomena “can only in the rarest of instances be turned into specific predictions of what will happen in a particular case, because we can hardly ever ascertain all the facts which will contribute to determine the outcome.”

As in the Nobel lecture (and The Counter-Revolution of Science), there is little constructive in terms of how to understand complex systems. Nearly all the discussion goes to show how incapable we are

Friday: Hayek’s “The use of knowledge in society”

This is one of his most influential works, an instant classic which allegedly turned the tide against planned economies in some circles of American economists. Hayek’s main argument is that it is futile trying to make society more efficient through central modeling, data analysis, and calculation. Since data is inevitably incomplete and models perfunctory, they would never be able to compete with market forces able to tap into the collective intelligence and integrate local knowledge, tacit or not.

After reading The use of knowledge, I wonder if Axtell would manage to recruit (a time-transported) Hayek to his ABM project. Parametrizing ABMs to optimally price all products, hmm, that sounds like a tall order. Of course, the usual questions for ABMs are sketchier, but would they interest Hayek? Aren’t conclusions from ABM scenario testing the general kind of statements Hayek denounced in The theory of complex phenomena?

Furthermore, Hayek did know that simulating social systems was an option but must have thought writing lengthy essays was a better methodology. If I meet him on the other side of the river, the first thing I would ask is that: if he has a theory in words but can’t turn it into a simulation model that reproduces reality, then why should I trust him? To that, he at least wouldn’t be able to retort that he can reach the same conclusions by deductive reasoning, because to that, he has given the perfect counterarguments.

Saturday: Is SFI a neoliberal think tank?

Looking for Hayek papers also led me to some intriguing 2022 papers from the field of science and technology studies:

Both point out that science and management at the SFI are coevolving and are at the service of neoliberal stakeholders. “[T]here exists a liaison between complexity science and the theories and practices of liberalism.” Hmm, I think most of the liberalism you can see in SFI’s management has exogenous reasons. Given that it is an American institute relying on private funding, what can one expect? And the science? Yeah, slime molds are pretty laissez-faire when you think about it. The game of life doesn’t evolve into a socialist utopia. (Or does it?) I.e., I’m skeptical that there is much politics in the bulk of SFI’s science, but please read these papers and make up your mind.

Sunday: Cybernetics in Soviet

Let’s go back to the other side of the political spectrum. Systems science, cybernetics, and those more engineering-minded fields related to, or overlapping with, complexity science have a stronger focus on controlling (rather than explaining) complex systems (cf. Project Cybersyn above). Clearly, they should be interesting for planning / commanding economies. Interestingly, such ideas never took off in the governance of the Soviet Union. Cybernetics was introduced with some resistance, and by the time it reached the actual administration, it was so watered down that it had little practical influence.

Scientifically, there were significant developments at the Novosibirsk State University, sufficiently distanced from Moscow, under the leadership of Leonid Kantorovich, who (closing the circle) was awarded the Nobel economics prize the year after Hayek. That is on my reading list, and this paper could be a good start.

Monday: Tektology

For the final day of this tour, we stay in Russia but return to the time around the Bolshevik revolution.

Who was first with the ideas of complexity and systems science? It’s an impossible question to answer precisely since the further back in time we go, the more generous we can be with the similarity to modern ideas. One of those very early contenders is Alexander Bogdanov. His “Essays in tektology: The universal organizational science” indeed has many commonalities with the complexity science of today. One point that impressed me is that Bogdanov sees the need for integrative, multidisciplinary science and that such science inevitably has to do with organization (which I read as, down the line, network science). It’s also fascinating just how general he imagined the applications of his tektology to be—from anthropology to interstellar space, from disordered magnets to comparative linguistics.

Bogdanov was a Bolshevik but still critical of Lenin. (Guess how Bogdanov died? . . . No. Actually, by poisoning himself in an experiment to stop aging.) So tektology also had applications to contemporary politics (the theme of this post), the organization of the proletariat, etc. Those passages are, however, hard to parse over a century later with little contextual knowledge.

After-trip photo album slide show

After this journey, I still think ideas and results of the complexity and systems sciences are so broad that anyone politically inclined can find content to appropriate. Evidently, for some (Hayek), political motives can also be a scientific driving force.

I have more questions than answers about this topic. The biggest is very general—how does the politics of the day shape science, and how does this change with the times and topics? Of course, this is nothing new, meaning there are plenty of books to read, including a new Hayek biography.

Bonus links

The videos below give some feeling for two of the key characters in this post. I’m tempted to call Beer the “Orson Welles of cybernetics,” but what would the corresponding epithet be for Hayek?

6 thoughts on “Complexity science in the name of politics: a travel diary

    1. Thanks. I wonder if Hayek was against computational methods (social simulations, cybernetics, systems dynamics, etc.) because he was old and stuck in his thinking, or for the same reasons he was against regression analysis. There must be some clue in the literature.

      Liked by 2 people

      1. Apologies – slightly rushed attempt to work out my own thinking here.

        I couldn’t respond to this immediately because I’m not sure – I suspect that (even based on the interpretation of the Nobel Prize speech I referred to), he did feel that way. And of course this is tied up with the ideological areas to which you refer. It’s intriguing to think about the coup against the Allende government as a very meaningful transition point.

        But your question causes me to ask whether cybernetics *is* computational… is it? I’m not convinced, let me just say that. I definitely feel that there’s some attempt in those who seek to separate ‘complexity’ from ‘cybernetics’ to say that it is – and I think that’s a mistake (or, at worst, a motivated mistake).

        At a deeper level, it’s a very big question: what is computational, and what is not?

        ‘The game of life’ is certainly computational, isn’t it? Agent-based modelling?

        But to come back to Hayek, there’s a commitment to emergence there which is admirable (are coups emergent? Sometimes. What if they’re driven by the CIA? Not if you’re looking for emergence from within the populous, yes if you consider exogenous factors to the system of interest still to be emergent… takes us back to the frame problem, essentially – which is what I’m going to do next).

        He clearly states the crux of his argument

        “Unlike the position that exists in the physical sciences, in economics and other disciplines that deal with essentially complex phenomena, the aspects of the events to be accounted for about which we can get quantitative data are necessarily limited and may not include the important ones.”

        Coupled with “This is sometimes carried to the point where it is demanded that our theories must be formulated in such terms that they refer only to measurable magnitudes.” And reference to (cybernetician!) Weaver’s distinction between ‘organised complexity’ and ‘disorganised complexity’ – then a comment on scientism (‘There is as much reason to be apprehensive about the long run dangers created in a much wider field by the uncritical acceptance of assertions which have the appearance of being scientific as there is with regard to the problems I have just discussed.’)

        I can’t help but believe that Beer would have heartily agreed with all of that; I’m not sure many people had a more nuanced understanding of amplification, attenuation and transduction (as in, not cellular transduction) than him. He might well be accused of presentational issues which might make things look more certain than that, I think – probably we all struggle to put things usefully into the field that don’t concretise and fetishise the adaptability and flow of practical wisdom (especially when you think of the extreme pressures and limitations which attended the birth of Cybersyn).

        Where I would criticise Hayek and why I cleave to the value of ‘cybernetics’ (and am critical in various ways of a lot that goes under ‘complexity’) is that the smuggled-in assumption seems to be something like ‘there are measurable facts’ – i.e. an assumption that the world is in some sense made up of facts, without reference to the fact that those facts all arise from our (to use a shortcut word) ‘frame’.

        Hayek, to be sure, gives space to ‘the unmeasurable’ – I’m not sure that is enough (I find Nicolescu the most effective in terms of showing clearly how our instruments, our probes, define and confine reality Beer, of course, was inspired in this by ‘what the frog’s eye tells the frog’s brain’ and the strong concept that there is, in a true sense, no sensing and that signal transduction is a perturbation in the ‘receiving’ organisation which is in regard to the configuration of the receiver.

        The challenge is that we must still act. When we act, it turns out that however free market we are, we (rationally, and also instinctually) operate in a ‘socialist’ way with regard, for example, to ‘family’, and a planned economy way with regard, for example to ‘the business’ (Theory of the Firm / transaction costs etc – see also ‘Walmart as socialist utopia’ – – General Intellect Unit, being actual socialists, are quite interesting on this).

        So the next step would be to ask why the boundaries of organisation cf complexity are so important to Hayek in where he places them. (There’s a sense in which this is always the important point). I have no time for that now – I just think that there’s a sense in which Hayek and Beer are natural allies, seeing something which most do not see. But Beer has, I think, more of an affinity to think of this as being ‘in the flow’ of complexity, surfing, engaged with it – engaged, to be sure, in a broader and potentially shaping way, but engaged with complexity nevertheless. Would that be broken by hubris of assuming too much control and causing disaster? Almost certainly, yes – but the counter-revolutionaries got that out of the way by taking the responsibility of violently destroying the system right up front, prior to ‘letting freedom rein’ 🙂

        And we always have to managed managed markets, whether in the business itself or, for example, in regulated markets or with government as a monopsonistic purchaser (a great scholar on government commissioning, Professor Gary Sturgess, was looking at dynamic Yugoslavian models of managing economies to assist with the latter).

        Other points:

        You said “The game of life doesn’t evolve into a socialist utopia. (Or does it?)” – well, it can evolve into a steady-state organised, ordered system over different timeframes.

        My worry is that the belief in complexity being ‘bottomed out’ through computational power simply takes us further into correlation.

        Since we have to have boundaries of intentional organisation, wherever we set these (even in setting intent), and I said above that ‘we must act’, one escape of course (let’s call it the Stewart Brand solution) is to flee into the wonder at the wonder of the emergent and the possibility and impossibility of it all – that’s something I find very attractive, but always transient.

        Liked by 1 person

      2. Thanks for your thoughts! Very insightful, and with several leads that I have to check out.

        Some random comments back: I was hoping the readers would think it too far-fetched to see any politics in the game of life (thus that example). Honestly, I believe there is not much to learn about life in it either; more than that simple rules can generate that complexity. In extension, I think a large chunk of the complexity literature overstates the ubiquity of computation (or rather, the usefulness of computation as an analogy to understand natural and social processes). (Pursuing that hypothesis more seriously is on my to-do list.)

        Hayek’s anti-scientism should apply to Beer et al., too, don’t you think? Would they ever agree on anything other than that controlling the economy is hard? (Hard but worth a shot, according to Beer. Hard and useless to even try, according to Hayek.)

        If cybernetics is inherently computational . . you are the one to answer! They are connected by being born at the same time and in the same intellectual environments, though.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s