Hacker News new | past | comments | ask | show | jobs | submit login
Project Cybersyn (1971) (wikipedia.org)
70 points by pavlov on March 14, 2014 | hide | past | favorite | 36 comments

This keeps coming up. I read pretty much everything Stafford Beer put up about the functioning of the system, but we never really saw a specification of the system dynamics-based model of the chilean economy it was supposed to be based on.

Now, if you've ever played with how a sys-dyn model is built... timing your delays is crucial. "Real-time control" is a super cool idea, but there are stochastic deferrals between actions by policy-makers or decentralized agents and their systemic consequences. Mis-time these and you'll just be inducing cycles.

That's a reason why modern economic planning is all based on watching response signals. We watch inflation and unemployment not because the relatively low thresholds that trigger action are catastrophic, but because inflation is a symptom of an overheated economy. We can't track demand and supply at the level of every commodity, and if we did, we wouldn't know how capacity expansion relates to effective availability to the consumer (this is a mistake countries that try to deal with hyperinflation with heavy intervention keep making, with the inevitable shortages in supply).

TL;DR: Zomg tech

This is cool but... given the moden economy is built on automated statistical models, couldn't we look at advanced capitalism as a sort of highly supervised, decentralized version of this?

The problem for me with the way we do things currently, is that the criteria the model is optimising (arguably economic growth) for don't seem to completely describe what most people would want the economy to do (improve living standards, individual freedom, 'happiness')...

In a way that's what some of the early 20th century Marxists (in the group of "Orthodox Marxists") were arguing. In the extreme form of that view they had a view on management that aligned nearly 100% with the view of the big capitalist "barons". A common view among industrialists of the era (starting in the late 19th century) was that modern scientific management, statistics, supply-chain-management, etc., was making bazaar-style capitalism obsolete, and ushering in an era of efficient, centralized, vertically and horizontally integrated, scientifically optimized production. That was one of the arguments used against anti-trust legislation: industrialists argued that anti-trust legislation was just purposely shoring up obsolete, inefficient models based on the medieval bazaar, which were out of touch with a modern scientific, centralized industrial sector.

A certain variety of early Marxism said more or less: yes, we agree with all that. Except for one thing: once industrial capitalism has centralized the economy, with a few huge trusts running things as a smoothly oiled production machine, the shareholders are now more or less superfluous. So, we'll just knock 'em off and distribute the proceeds of this machine among all the workers equally. The industrial capitalists only disagreed with the idea of socialist central planning in that they didn't want to distribute the proceeds of central planning, not in believing that decentralized planning was better. Many did agree that some kind of redistribution was necessary so that all society benefitted, but they preferred redistribution managed by the monopoly capitalists themselves in the form of philanthropy, which they felt they would be able to thoughtfully manage (Andrew Carnegie's The Gospel of Wealth is representative of this view).

The main difference I'm seeing here is that the current model is siloed, e.g Military-Industrial, Media, Wall Street. The interesting thing here is that this model aspired to optimize the entire economy and all of it's pieces. What we have now is really expansion over optimization, and it will continue to do while we follow the culture of ego. Things like improving living standards and happiness really involve going beyond this solipsist mess we seem to be in.

For anyone interested in this the Eden Medina book provides some historical background. I think eventually that Cybersyn will be viewed as an early attempt at post-capitalist economics in which resources are efficiently managed in real time and there is feedback at every level.

From the Marxist perspective, isn't communism post-socialism?

IMO, as an idea, this isn't post-anything. It fits very nicely (almost to the point of cliche) into the "modernist" family of ideas. Socialism was the most prominent political and economic philosophy of modernism. Objectivism is another example. Anti-socialists of the modernist period (who often objected to modernism as a whole) kind of parodied modernism in a way that is similar to this. Super efficient computers that calculate truth and can run everything optimally.

Fredrick Hayek (an early postmodernist) parodied social "scientism" and socialism in just this way. He argued that this kind of planning is doomed no matter how big your computer is because the requisite information is unknowable, not just unknown.

I see this as something like a 1950s science fiction work. But I'm biased. I don't think this will ever work and I hope it won't either.

That argument, commonly associated with Hayek, quite trivially does not make sense.

If the information is unknowable, then it cannot play any role economically. If it is unknowable, the market itself cannot act upon it either. The idea lurking here is of inarticulable knowledge or inexpressible information -- and it is a contradiction in terms.

Hayek's big point -- in 'The Use of Knowledge in Society' -- is really this: an economic system is a cooperative system, a cooperative system requires networking/sharing information, and money and markets are a way to convey and share information.

Hayek does not prove that centralisation is impossible, nor does he prove that money and markets are the only means to run an economy. An argument based on information is not going to get you there -- rather the opposite in fact.

Later in that article, Hayek says -- about gathering dispersed knowledge etc. -- (not) "likely" rather than "never", and also says "it is just conceptually possible". That is the sensible view: if the market is performing a computation, then any sufficiently powerful general computer could emulate it and any sufficiently good network could transmit the relevant information around. Indeed, with enough power you could process more information than the money-market system.

And that is actually where we are now. When there is a pocketable networked computer for every human on the planet, that rather blows away the idea that we need money to solve the 'economic calculation' problem. The money-market system is already obsolete, it will just take some time to evolve away from it.

The information is expressed through demand. People spend money according to their values. If people don't have money, then it's very difficult to determine their actual preferences. That is how I always understood the argument anyways. Likewise producers need to express their values for the materials they need, and those producers have values for the materials they need, etc.

I second this. For anybody who cant be fucked sear hing up the link on wikipedia the Eden Medina book is called 'cybernetic revolutionaries'. It's a great piece of research (although, in my opinion, lacking somewhat on both political and technological analysis. It seems to be the best book out there on cybersyn).

Also, for anybody interested in the general ideas check out the novel 'Red Plenty' by Francis Spufford. Its about attempts at a similar scheme in the USSR and the problems it faced (more ideological and organisational than technological)

What makes Cybersyn interesting for me, particularly in relation to the Soviet schemes, was the decentralised nature of it, including the (unfulfilled) ideas of bringing workers into the fold.

Also, the Adam Curtis series 'Pandoras box' (youtubeable I think) discusses both cybersyn specifically and the ideas around it.

From my perspective it is even post-socialist, not to say post-capitalist. Since socialism supersede capitalism. What was attempted in Chile was to jump even further within a time of single human generation. It will never be forgotten.

the term 'post-socialist' is generally termed 'communist'.

Allendes Chile was actually relatively moderate in its socialism, in the sense that he was a radical reformist rather than a revolutionary.

I agree, however, that there was a great potential for something much more dramatic (ie, a true decentralised communist planning) within the cybersyn project.

It reminds me a lot of the present day "Zeitgeist Movement".

I love this kind of modernist, socialist utopian ideal. It's gratifying to think that whilst the idealogical battles of the 20th century were still playing out, the idea of marshalling effort and resources toward the common good without capitalism was viewed positively.

And the chairs, I love the chairs.

It reminds me very much of the Dharma Initiative from the TV series Lost.

And now think that quite the opposite idea, the post Pinochet free market economy, was what made Chile the richest country in South America.

As a rule of thumb, good intentions bring very bad results if you need government violence to implement them.

Whereas Pinochet implemented his "good intentions" without violence? Reality check, please.

He obviously did use violence. However, are you suggesting that you need government violence to minimize goverment's involvement in economy?

You must be aware your second paragraph negates the first. Pinochet's "good intentions" (which is subject to debate) were implemented through the very liberal use of violence. They should have, therefore, yielded very bad results (which they did, in the short term).

One is left to conclude Allende's reforms could have yielded equally good long term results without the illegal resort to violence.

Huh? Allende was no saint either, he certainly used violence directly. One of the biggest complaints about Pinochet was that he did everything that Allende was accused of (including torture) but worse, of course Pinochet was in power for a longer period of time.

What accusations are you talking about? Allende did a lot of things some may consider questionable (land redistribution, statization) but torture is usually not among them.

There was some preparation for guerilla operations, from both sides, in case a coup happened (as it did) and some armed resistance to the expropriations (which were legal according to the then valid laws).

you're right, torture is usually not among them these days, but it's in the historical record, and chilenos of the time were aware it was going on.


"g) It has made frequent politically motivated and illegal arrests, in addition to those already mentioned of journalists, and it has tolerated the whipping and torture of the victims;"

So, you are bringing up a document used to justify the coup as a credible source of accusations that were never proven in the years following its original publication, despite the interest of a government that used violence routinely to prove them and justify its own brutality as a lesser of two evils.

Excuse me if I find it difficult to believe in it. I don't think you should take it as a representation of the truth either.

Sorry, but your criticism is equally valid for basically any sort of legal propaganda, for example the Declaration of Independence. Moreover, I don't know how hard you've tried to actually compile evidence that "accusations... were never proven in the years following its original publication" - since you clearly weren't aware of those accusations, say, 24 hours ago.

Also your history is wrong; the document was used to justify 'impeachment' (because such a concept did not exist in the chilean constitution); the coup itself came afterwards, when Pinochet wrested his leadership of the military into overall civilian leadership, and I would not be so sure that the authors of the document intended the course of action that happened to be an indirect consequence of their petition.

The document, btw, passed the chilean legislature 81-47, which while not the two thirds required to remove the president, is still quite a supermajority.

Anyways, I think it should be the default assumption that governments torture people. There are scant few that haven't.

You are being very naïve if you think Chilean legislature had no idea of the impending coup. As for the lack of evidence, I tried to find some and came back empty handed. If there were anything even distantly resembling the systematic use of violence Pinochet employed made by the Allende regime, I'm quite sure it would be amply documented by now, if, by no independent historians, by the Chilean military. Since I can't prove it didn't happen, do you have anything more solid than this to demonstrate it did?

Oh yeah, I can picture the scientist guy walking out to talk about the rules of some experiment.

Don't we all miss the days computer programmers used lab coats?

Millions wasted on expensive systems and software to manage and negate the damage caused by an inherently inefficient method of running an economy. One size does not fit all, no matter how much state software you create to make it so.

If you combine this level of technology at a micro level, combined with the incomparable efficiency of the market, THAT is when you start to see incredible things starting to happen.

From small businesses monitoring their sales and orders in real-time, sharing their data with their marketing agency via api's. To construction companies using internet of things to monitor structural wear to alert them of replacements needed. It's at this market oriented macro-level that you will see the greatest advantages.

Creating and making the technology to power that is most important. And that's where we come in!

Stafford Beer's Chilean Cybersyn project was one of the inspirations for the Community Memory (http://en.wikipedia.org/wiki/Community_Memory) project in the Bay Area in the early seventies, which, being the first public computerized bulletin board system (and flea market), helped lead to today's World Wide Web and related.

...but it was not even let to to either work OR fail on its own, so we'll never know if some of the ideas could have worked.

...like it happens with all attempts of any kind of technocratic government: sooner or later some groups of power totally freak out contemplating god knows what imaginary consequences of its future development and pull the plug on it and paint it as another "failed experiment".

When will we actually become mature enough to be able to truly run social-engineering experiments in a scientific way and asses their results? (repeatedly and periodically, because with time people's mentalities change, so what didn't work in the past might work in the future and viceversa). When will we accept that "political 'science'" can only be an experimental 'science', so we have to at least try and run unbiased experiments from time to time and see what works better, instead of just letting the blind watchmaker of evolution run its inefficient search of the problem space?

As an analogy [EDIT+ to clarify] of how wrong I think we do politics and social engineering now [/EDIT]: Imagine that in medicine you'd run 10% of a drug trial, then stop/pause it and make an "educated guess" based on this 10%, and based on this published guess have people vote whether to put the drug on the market or not. Sounds reasonable to anyone? :)

To run with your analogy. Let's take the case of vioxx, which was voluntarily withdrawn by merck after post-approval experiments showed statistically significant increases in heart attacks - never mind that this is exactly biasing the 'experiment' against vioxx (http://www.evanmiller.org/how-not-to-run-an-ab-test.html).

The people vote to withdraw vioxx from the market. Because, heart attacks! Well, what of the plenty of people who decide that they'd rather have a slightly increased risk of heart attack instead of debilitating, quality-of-life-shattering daily pain?

To take it further, what about conditions that specifically only address a minority of people, let's say that we made a drug that cures sickle-cell anemia, but gives a 100% chance of getting cancer by age 70. Who should make these decisions?

Also: Who gets to be the person deciding on the 'educated guess'? And how can I be this person so that I can pick my pharmaceutical stocks ahead of time?

Generally speaking:

Whether something should be "on the market" and "how it should be used" are two different things. As political examples, natzism should definitely be "off the market"! But different forms of socialism can work well for some groups of people. Same as unregulated anarcho-capitalism might work well for other that share the values and accept the tradeoffs.

But about the whole family of technocratic governments or of forms of governance base on cybernetic models, or of social orders with a "minimum guaranteed income" or maybe others even more exotic we have no usable experimental data!

Addressing the specifics:

Withdrawing vioxx from the market was, rationally speaking, a bad idea for the patients at the time, the rational thing to do would have been to modify the prescription guidelines and availability. But considering the reality of lots of MDs being not always up to date or just stupid (especially when it comes to 'kind of subtle to some' pharmacology issues, speaking from a 'past life' experience in the healthcare area), lives might have been saved by removing it from the market, especially in the USA where it was overprescribed. Nowadays there are other selective COX-2 inhibitors available, so the fact that vioxx is no longer on the market doesn't matter - there are mostly equivalent or better alternatives.

About a hypothetical "drug that cures sickle-cell anemia, but gives a 100% chance of getting cancer by age 70": it should obviously be on the market. There are countless people dying from sickle-cell anemia, and, if regulated as a "last resort" treatment it would be a true life saver. I would be used just like aggressive chemotherapy or heavy dose radiation therapy is used (is can cure your current cancer or put it into remission, but it will almost surely give you cancer if you live long enough after it).

Also: maybe the drug approval example was a bad analogy and it just pushed things off topic...

Also: kudos for the link, it really is insightful!

EDIT+: I think you misunderstood me, the example with the drug trial was about what should not happen, but happens now in politics! I phrased it like this to make it obvious how wrong it would be if, for example, we did medical research the way we do politics.

The era of evidence-based medicine will be remembered for trying to apply population-level analyses to individual cases, which as I've understood it is not really the point of population level analyses.

well, in medicine we'll soon have personalized/individualized-medicine.

but you can't really have personalized/individualized societies, unless we all isolate ourselves from each other and leave as ascets in caves :)

You can do population analysis, but thats valuable in terms of making population level decisions. Not decisions about individual care. An example being; where do we build a hospital. Not; which medication should we give this person. Right now we have universalized standards of care for things like hypertension that clearly do not work for the entire population yet we have accepted that x% improvement overall is the best. I think the future will involve identifying the subtleties of genetic and microbiome impact on all medications, and tailoring regimens based on those indicators rather than the EBM-based standards of care.

If you want a 5 minute video digestion of this, including who Stafford Beer is, I gave a lightning talk at Liverpool Ignite.


There is a software implementation of Beer's Viable System Model in Smalltalk. The VSM was part of the basis for Cybersyn. See:


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact