Cybersyn and Allende’s Semi-Automated Luxury Socialism
One of the interesting “what ifs” of history is “what if the 70s-80s commies used computers to do their planned economy.” Men like Kantorovich, Nikolay Fedorenko and Victor Glushkov helped develop some of the mathematical tools and computer systems which would have made this possible.There were abortive attempts to build this in East Germany (pdf link), the Soviet Union and Allende’s Chile. As far as I can tell, the Soviet and German efforts were crushed by old guard party rednecks who feared losing control to technocrats. Oddly, Allende’s attempt at this, which would have been constructed of bone knives and bearskins, seemed to come closest to being deployed.
The visionary behind this was Fernando Flores, who is still alive despite being Allende’s minister of finance and later “General Secretary” back in the early 1970s. Flores was inspired by a sort of futurist “cyberneticist” operations research proponent named Stafford Beer. Operations research is generally now thought of as the field of applied work involving optimization; linear programming and all that. In those days it was something more general: mathematics applied to the problem of management. Guys like Beer with this sort of training ended up running large parts of the war economy.
It’s difficult for me to characterize what “cybernetics” is, probably because it doesn’t really mean anything. Norbert Wiener, who I respect, coined the phrase, and more or less defined as a hand wavey general study of systems of feedback and control mechanisms. As far as I can tell, “cybernetics” was a complete bullshit field, and what it really meant was “I know futuristic looking words and have Wiener’s book on my shelf; pay me more.” Stafford Beer was a proponent of “management cybernetics” which, as far as I can tell, meant “using data to make business decisions.” The books are hysterical; you can go look at them on filesharing sites. They appear to be total horse shit. FWIIW the Soviets more or less agreed with me, at least in 1947; the field of cybernetics was condemned as “a science of obscurantists, a pseudoscience wedded to obscurantist epistomology.”
This looks like it pertains to something real; nope
Flores hired Beer. Beer cut his rates to $500 a day (about $2500 in today’s money) along with unlimited cigars, chocolate and wine; items the Chilean government had a surplus of. Mind you the Chilean government was being starved of dollars at the time as a form of colonial pressure, just as the Venezuelan government is now in 2019. The results were hilarious.
The thing Beer built for the Chilean government is most famous for its control room, so we’ll start there. It had a bunch of cool chairs where powerful human intelligences would examine data on the walls and vote on the cybernetically optimal next steps. The chairs are TOTALLY not based on Captain Kirk’s control chair from Star Trek; every account of the thing makes certain to mention this. I assume Beer got a lot of shit over it, and rightly so, because he totally copied this from Star Trek. Or, if he didn’t, his designer did.
The chairs come equipped with ash trays (men of power always smoked in those days) and a place for a whiskey glass, presumably in case Castro or Beer wanted to relax while making important decisions. There are 7 of the chairs to make sure they can always achieve consensus when they vote. They use ridiculous glowing geometric buttons (like on Star Trek) instead of keyboards. The main reason is because they’re all hardwired to physical mechanisms; the images and graphs on the wall are not computer generated or directly interfaced with a computer at all. They’re slides and viewgraphs that are manually put in there by human helpers behind the walls. The lack of keyboard has also attributed to keyboards being “feminine” or confusing to men, which may be true in some way, but which doesn’t make any sense, as male factory workers used the teletypes to communicate factory data with home base.
The important pieces in the background, besides the workers behind the curtain, were the computer, an antique that they had on hand, and dozens of teletype machines they distributed to factories and control points to relay important metrics and data back to home base. These metrics would be entered into the computer (by hand), which would presumably generate reports (it’s not clear this ever happened) which would be manually turned into viewgraphs and slides by artists. The display of these slides would then be controlled by heavy drinking/smoking men in Captain Kirk Star Trek chairs, just like when they show you their vacation photos on a carousel slide projector. Except it was 7 people controlling the projector and arguing about what it all means.
People go on about how futuristic this was, because … I dunno, muh internet and muh powerpoint is used to make business decisions now. The reality is, Beer built the Allende government the economic equivalent of one of those WW-2 era RAF air defence sector station operation rooms with a little futuristic woo slapped onto it. It has some trappings of high technology with the Star Trek like buttons, but those buttons really didn’t do much of anything. The data allegedly eventually feeding back to the computer could have been more effectively gathered by people on the telephone with yellow legal pads rather than the teletype, just like it was in a WW-2 era RAF air defence command center. In fact, factory managers ended up routing around the central command center and teletype process by calling up other factory managers and working out potential supply issues.
The one time it was allegedly useful was during a CIA organized truckers strike. You’d expect any central data/controll station to be useful in a crisis situation, more for putting decision makers in a room where all the data is available than anything else. Pretty sure the slides, assuming any were made for the event, and the computer, assuming any data was entered into it, were not helpful here.
It may have eventually been a helpful tool if Pinochet hadn’t taken over, thrown many of those involved out a helicopter, and destroyed the room of socialist power. The fancy appearance of it was almost entirely Potemkin village though. It relied on people entering stuff into the system at the collection end. It relied on “data science” people writing code. And it relied on artists marshalling the data into useful and insightful visual reports. It also relied on people who managed the factories to obey, and in a timely fashion; presumably once you finished your whiskey and cigars, you summoned a telephone to tell the factory workers what to do. All of these things are guaranteed to more or less fail.
Even small design decisions were foolish. The controls; why should everyone in the room be able to control the view? How could everyone in the room control the view? We don’t give everyone a remote when we do powerpoint now. Beyond that, how does the cigar smoking drunk in the chair know what slides are where when he’s pressing his buttons? It changes every meeting, and some boob in the background could screw up the order of the things. Who sets the agenda? I strongly suspect the guys in the chairs would be shouting through the walls to get the secretaries to pull up different slides. Also the voting buttons. Seems very scientific and futuristic to have the seven cigar puffing drunks in the Captain Kirk chairs voting by some kind of electrical secret ballot to make decisions. Why wouldn’t they just say “dude this sucks?” How would it make a difference? And of course, there was no way to actually transmit decisions using all this fancy electronics, besides the telephones they should have used in the first place.
The SAGE system was a real world embodiment of this sort of thing, built 15 years earlier. Unlike Cybersym, it was a real time system, with real time data flowing into a central processing unit, and real time commands being sent out to remote bases. SAGE also had human operators who helped the computer make the right decisions. SAGE worked (we think) because it actually was entirely networked, and the data flowed quickly, and split second decisions were absolutely necessary. SAGE also didn’t have any goofy fake control panels with a place to put your liquor, or LARPY telex stuff where secretaries had to do data entry to put the data into a computer. But, at the time, it was probably fairly difficult to see this, and people who watched a lot of Star Trek were probably impressed.
I’m not sure how the detailed history of this played out. It’s entirely possible American spooks ramped up their efforts to depose Allende in part because of fear of this. US economists and analysts almost always overrated Soviet and communist efficiencies. In part this happened because the Soviets would occasionally surprise people with things like Sputnik. In part it was rice bowl politics; you get a bigger budget for the scary threat rather than the third world threats made of coconuts and rubber band slingshots. But mostly it was because the technocrats and economists who purported to study Soviet economics thought their own farts smelled of roses. Western economists in those days were not the mere bean counters and toadies for capital they are today: they had vast powers to regulate the economy, set prices and so on. They just assumed more regulation of the economy would cause it to run more efficiently. They believed their own bullshit.
The threat of a more efficient civil service is taken very seriously by governments. For examples from history, the late Imperial government of Russia was seen as a great threat by Germany as it had developed an efficient and productive civil service. One which was rapidly industrializing the country and improving its logistics with a fraction of the per capita civil service manpower of Germany. WW-1 may have been partially a result of this fear. I can’t prove any of this without access to spook internal documents and reports on Cybersyn. But I do have the example of an article in New Scientist who described Cybersyn as potentially “one of the most powerful weapons in history.” I mean, it very obviously wasn’t and probably couldn’t have been with the approach they took. But the CIA had no real way of knowing this.
As a sort of weird coda to all this, apparently Brian Eno and later David Bowie and David Byrne befriended Beer. I kind of wonder what they talked about, or if they just partied. As mountebanks go Beer seemed like a fun guy. Any proto data scientist who puts ashtrays and whiskey cup holders in a Captain Kirk control chair can’t be all bad, even if he was full of shit.
Bones send whiskey and smokes, stat
https://www.newstatesman.com/world/2018/08/project-cybersyn-afterlife-chile-s-socialist-internet
https://www.newyorker.com/magazine/2014/10/13/planning-machine
https://mitpress.mit.edu/books/cybernetic-revolutionaries
https://www.jacobinmag.com/2015/04/allende-chile-beer-medina-cybersyn/
http://www.guardian.co.uk/technology/2003/sep/08/sciencenews.chile
Locklin on science 
I wanted to ask you to write about this….you read my mind. Awesome work
And regarding teh definition of cybernetics. To me it is just a superset of control systems engineering.
In Germany it is a respected degree, pretty much like EE, CS or Aerospace Engineering. IMHO their curriculum is quite interesting:
two semsetrs probability theory (which is more than a typical german engineer is exposed to)
basic phyiscs (1 semster statistical physics,1 semester mechanics).
Some engineering (3 Semester mechanical engineering, one semester EE and one semester signal proccessing).
Plus two introductionary courses in CS and one course in simulations and one in numerics..
I think you could hire worse people.
I know several doing interesting and practical relevant stuff in their companies.
I had no idea, though I figured you would educate me that the term wasn’t that weird.
I suppose it looks like an American robotics degree … with more probability theory.
Yeah, you are right, quite a lot like robotics. A little more “generalist” maybe, but not like the voodoo stuff in “complexity science”. Studying “technische Kybernetik” at University of Stuttgart IMHO is a quite solid education.
Have you ever read Red Plenty, by Francis Spufford? It is a historical novel that follows various characters, fictional and real (such as the aforementioned Kantorovich) trying to make the Soviet economy of the 1950’s work and beat the USA (spoiler warning: they fail). Unfortunately, I haven’t read it, so I cannot say if it is good or not. The left-wing academic blog Crooked Timber had a “seminar” on the book, including a piece by statistician Cosma Shalizi on the plausibility of a computer-optimized centrally planned economy.
Also, the economic history blogger wrote some interesting posts on the Allende and the blockade and Chile’s economic performance as a whole.
I started reading it, more or less after reading about it on Cosma Shalizi’s blog somewhere (or maybe it was his collitch roomie Philip Apps). It was pretty dry and shitty, so I couldn’t do much more than skim. Cool idea, but fairly disrespectful to readers: plot, characterization and a story arc are important in any novel. At least that was my impression of it. Seemed more like filler in a complex board game manual; like warhammer40k for commie nerds. I was thinking of thumbing through it again before posting this, but didn’t.
Thanks for history links!
Using the word “commies” is not helpful. And coupling smoking and drinking as if they would be equally distracting makes for fun copy but is totally mistaken. Did love the reference to the CIA sponsored trucker strike – exactly the kind of thing they have been doing in Venezuela for 20 years – and then screaming about the “failure of socialism” as if the CIA and Sweden did not exist!
I love it when people come onto my blog and attempt to police my choice of words. When, exactly, has this ever worked for you?
I’m actually fairly sympathetic to figures like Allende, but modern day fans of commienism are generally bugmen.
Commies were not particularly helpful, either. As for Allende, he made the capital mistake of believing his own BS, according to his KGB handlers. A decent but basically stupid guy, in way, way over his head. Had Pinochet’s coup failed, the Soviets would almost certainly replaced him with a harder, smarter guy, as they did in Afghanistan.
What a nice story! It reminded me about some official n Russia who at the beginning of nineties got a computer on his desk. He looked at it and then said: keep the TV, remove the rest. Typically this is presented to show how a stupid person can rich a high position. But for me that guy was not stupid at all. At the very least he was very realistic.
The Glushkov-style cybernetic economy management cannot work in principle as was proven in 1920 by Ludwig von Mises (the so-called economic calculation problem). The reason for that is very simple: the individual utility functions are hidden, and the only way to obtain reliable information about them is to observe un-coerced economic exchanges (i.e. voluntary trading) – the very act of regulating these trades destroys the information. The outcome is that in a planned economy all input information regarding what people value and want is destroyed, and thus no rational economic calculation is possible. Having a computer in the loop doesn’t change this fundamental fact. (In a free market economy this information is aggregated by the process of formation of prices: prices end up tracking the balance between utilities of consumers and producers, which makes economic projection based on prices of inputs and outputs possible). While this objection to central planning was ignored central planning-leaning economists (which would make about all of them – they can’t just admit their field is junk science, could they?) the Soviets ended up planning the economy based on prices they obtained by reading American catalogs (I worked with folks from CEMI), which allowed them to produce plans which weren’t completely insane. In the end it only helped to delay the inevitable, and Soviet economics self-destructed in exactly the manner predicted by Mises: increasingly getting out of whack and producing tons of stuff nobody wanted while creating shortages of basic life necessities (such as TP), until the country got to the brink of starvation by 1991.
ironic that you say “SAGE also didn’t have any goofy fake control panels with a place to put your liquor”, because SAGE famously had consoles with a built-in ashtray. and I don’t know if it counts as “LARPy” but they had light gun input and that’s pretty futuristic if nothing else
The SAGE system had real control panels, and no place to put your liquor, because Air Force officers who drank while on duty could probably be court martialed. The fact that people smoked indoors in the 50s is apparently forgotten knowledge.
Sorta reminds me of Asimov’s “Psychohistory,” which, while I generally enjoyed the books, I thought was a stupid, socialist, Utopian vehicle used for the world creation.
I think it could have been an interesting experiment. One thing you have forgotten: the system involved the knowledge, imagination and skills of every employee of the factory and only certain problems, which couldn’t get solved by the factory itself was intended to be directed to the controlroom. And only a few important data figures were intended to be shared, all based on an open, improving system.
That means, that the factories were not to be intended to be under strict government control. The government, represented in the controlroom, would interfere only in case of very big problems concerning the whole or bigger parts of the nation or to help connecting the right parts of it to solve the problems. I think that is pretty smart, isn`t it?
These functions today are highly used in hightech firms like Audi, Porsche and Mercedes, which are on the top of technology, management, social achievements for workers and profits as well, so the idea behind it has proven not to be that bad.
The Star Trek layout ist pretty funny, but is only show to sell the idea. They used the same technique as commercials: create nice and interesting, forward thinking pictures (as the Star Trek pictures) to keep high the interest of people and as a projection of a possible better future.
It is a pity that it couldn’t evolve autonomously in Chile, so we cannot say what would have happened.
I’ve definitely considered the fact that ops research (mostly originating in Soviet Union) and data science techniques applied to business are pretty close to what a lot of communism had in mind for scientific management of the economy. My first boss in the business was a commie who got me to look at the book “Red Plenty.”