[29athrowaway]

Do you truly believe that some bayesian economic simulator on a IBM System/360 Model 50 was going to be better than the market regulating itself via supply and demand and free competition?

And that 7 people were going to be able to effectively manage the production of every sector of the economy?

[sudosysgen]

Cybersyn was a pilot project. The idea wasn't to manage the economy, but to have a first proof of concept that could manage some parts of the economy, with the help of people directly involved.

That being said, the free market "regulating itself" is pretty bad at it. First of all, it only has one cost function to optimize, which is never going to be what humans want to optimize. Then, it forces production information largely through low fidelity price signals. Finally, these signals are very much delayed, and by the time you answer the price signal, it may have changed wildly, which means that in some cases straightforward organisation that anyone with the relevant information could rapidly approximate can take years to come because the signal changes more rapidly and with considerable delay leading to dysfunction as other prices signals attempt to respond to it. There are many other issues with it.

Of course, an IBM System/360 and 7 people couldn't do it. It's just an attempt to see where the system had to improve. If it did work out it would have ended up with a staff of thousands of people, direct connections to much of the machinery and to people in the production lines, specially built supercomputers with linear and non-linear optimisations accelerators, etc..., while still allowing many markets where it made sense and use them as information.

As it was, a bunch of old crusty bureaucrats in the Soviet Union managed to get within the order of magnitude of the market regulating itself via supply and demand and free competition with data manually collated and corrupted every step of the way and refresh cycles on the order of months if not years, with the computation itself being in large part done manually. So it was an interesting and worthwhile experiment.

[cardanome]

Yes.

To speak metaphorically, decentralized market regulation by supply and demand is the Bitcoin style of solving the problem. That is extremely inefficient, wasting lots of resources, being very slow but avoiding a single point of failure.

If you are surprised by the comparison, consider that you are avoiding the calculation problem with "the market regulating itself" phrase which falsely implies that this regulation would happen for free. It is equivalent to saying "Bitcoin approves transactions itself". That is true but there is still huge processing power employed to do so, as the huge mining farms show. In a free market calculations are performed by every individual actor in the market. In fact companies spent huge amounts of resources on market research to find out what and how much they should produce.

Central planning does naturally require far less processing power and can be more efficient (it has a much better best case but, to be fair, also a much worse worst case).

Even primitive, paper based, Soviet style planning has been proven to be quite adequate for early 20th century economies (don't be tempted to move the goalpost here, yes there were inefficiencies and it sometimes did perform worse that free markets, the point is it was already good enough to keep a country running for decades)

Now cybersyn was the right idea to go a step further. Make the central planning faster, more flexible and more democratic. To answer your question a little bit less simplistic, it depends. Obviously cybersyn was an early experiment and would surely not sufficed but would have Chile been better off with it than with the bloody CIA backed dictatorship that was put in power instead? Yeah, I am quite sure. We have now strong empirical evidence that neoliberalism does not work as promised.

[qsort]

No.

Central planning works horribly and there's wide evidence of that.

> decentralized market regulation by supply and demand is the Bitcoin style of solving the problem.

This is word salad. The criticism of bitcoin - that I would absolutely agree with - is that it wastes resources to solve the same problem that a SQL database solves, just more inefficiently.

In order to solve the same problem a market solves you would need to take into account the preferences of millions of independent actors and mash them together into a Frankenstein solution, which is clearly beyond the capabilities of any government, with or without computers.

> Even primitive, paper based, Soviet style planning has been proven to be quite adequate for early 20th century economies

This claim is complete insanity. At no point in history the USSR has performed better than the first world in any sector, with the exception of a few propaganda stunts (like the space race) that came at costs so enormous that could only be borne by a brutal dictatorship, and that were promptly and easily matched and surpassed by the US.

Frankly, if your argument for central planning is to point at how well it worked out for the Soviets, I don't need to argue any further.

> it was already good enough to keep a country running for decades

Anything is good enough to keep a country running for decades, that's not a reasonable indicator of any kind of success whatsoever. The alternative to central planning is to stop doing it, not to go back to the 19th century.

> would have Chile been better off with it than with the bloody CIA backed dictatorship

A comment that contains a literal invitation to not move the goalposts ends with an irrelevant tangent. How ironic.

Yes, the OG 9/11 was bad, you get a gold star.

Cybersyn is interesting as an information system for the technology it employed, and is for sure a great display of ingenuity.

But come on, let's stop beating this dead horse. Central planning is a pipe dream.

[cardanome]

> In order to solve the same problem a market solves you would need to take into account the preferences of millions of independent actors and mash them together into a Frankenstein solution, which is clearly beyond the capabilities of any government, with or without computers.

That is like trying to model the exponential growth of laternfish[1] by emulating the behavior of each individual fish and then claiming it impossible to compute on consumer hardware. (The reference might sound like "word salad" if you have not participated in the last advent of code, it was just the first example that came to my mind. The point is, just because some naive approach at computing something is not realistic, does not prove the problem to be impossible to solve.)

For example here is a modern piece of software that can create a five years plan in reasonable time:

https://github.com/wc22m/NewHarmonywithJulia

But anyway, you missed the point of my "word salad". People have to obviously plan in a free market. I will not be successful if I just produce things blindly, or at least it will be pure luck. Would you say that success in the free market is purely based on luck? Maybe part of it, for sure, but obviously it is possible to plan ahead, to determine and model demand and so on. While overproduction is impossible to avoid in a free market economy, most companies at least hilt the general ballpark, they will produce a bit too much but not hundreds times too much or they wont stay in business. So obviously planning does happen by the individual actors, they do calculation. They do the work that individual miners in a bitcoin network do, hence the metaphor. The free market itself is a giant computation engine, that solves the problem of economic planning through the interactions of the individual actors of the economy. A very crude machine but it solves it. From this it follows that economic planning is obviously possible and can be solved.

> This claim is complete insanity. At no point in history the USSR has performed better than the first world in any sector, with the exception of a few propaganda stunts (like the space race) that came at costs so enormous that could only be borne by a brutal dictatorship, and that were promptly and easily matched and surpassed by the US.

The comparison is meaningless, as the Soviet Union and the US had vastly different starting conditions. If one country is economically stronger than another, it does not follow that the economic system is better or worse as many factors influence performance. If we run a 10k race against each other and you start from the beginning while I have a 5k lead before the run starts, it does not follow that I am the better runner.

The Soviet Union started as one of the least developed areas of this planet, literally still using the wooden plow and was devastated by two world wars and one civil war in a short period of time. It even being considered a competitor to the US is more that a miracle.

Could it be that your views on the efficiency of central planing are more based on prejudices that looking at actual empirical data? And again, we are looking at the most naive implementation of central planning. This is like judging the speed of garbage collected languages on some weekend project.

[1] https://adventofcode.com/2021/day/6

[qsort]

> The free market itself is a giant computation engine that solves the problem of economic planning

No. This is the issue. Markets solve a different problem. If I like A better than B and you like B better than A, this has to be taken into account for the eventual solution to be Pareto-optimal.

You would need to set prices for all of the hundreds of thousands of goods and services a modern economy produces, which requires access to an amount of information that no entity can possibly handle. It's not just computational power, it's a fundamentally intractable problem that is only solvable if you make wild assumptions.

All of that only to get a solution at most as good as the market equilibrium.

> The comparison is meaningless, as the Soviet Union and the US had vastly different starting conditions

Come on, this is a total charade. Don't want to compare it to the US? How about Japan? Or any other first-world country of your choice? Or how about West and East Berlin?

> Could it be that your views on the efficiency of central planing are more based on prejudices that looking at actual empirical data?

Yes, I have priors, as we all do. I believe my priors are correctly adjusted given the available data.

[sudosysgen]

>No. This is the issue. Markets solve a different problem. If I like A better than B and you like B better than A, this has to be taken into account for the eventual solution to be Pareto-optimal.

>You would need to set prices for all of the hundreds of thousands of goods and services a modern economy produces, which requires access to an amount of information that no entity can possibly handle. It's not just computational power, it's a fundamentally intractable problem that is only solvable if you make wild assumptions.

That just isn't how cybernetic planning was ever designed to work, so I agree with who you responded to that it's a case of only considering a naïve approach.

Almost all proposals have a consumer goods market - others have a functional equivalent. You let consumers compete to buy products, and in doing so you don't have to set the prices and predict what people like more - you just let them choose.

Markets under a capitalist system attempt to use prices on the other hand, by having the capitalist decide where to invest his capital for maximum return, to decide how things should be produced. This is a terribly inefficient solution computationally because you are going through hundreds of layers of indirection and noise, and actively hiding a lot of the relevant information. In doing so, most of the necessary information for production is constrained into low information price signals, which causes endless issues.

> Come on, this is a total charade. Don't want to compare it to the US? How about Japan? Or any other first-world country of your choice? Or how about West and East Berlin?

Firstly, you have to realize that the Soviets did not do cybernetic planning. They didn't continuously integrate realtime information or production and demand. They did their central planning in yearly increments using pen and paper. The USSR started off much worse than any first world country, even Japan

The fact that even though it was planned so poorly using such inefficient techniques and with such low responsitivity, and still managed to perform better than most of the world and compete with the best capitalist economies, is in my opinion a good signal that the advantages of central planning are considerable, even if with the horrible implementation of the Soviets, they were ultimately outbalanced by their issues.

[29athrowaway]

They wanted to achieve supply chain optimization at a scale similar to what Amazon does, but with IBM computers from the 70s, and instead of the Internet they wanted to use Telex machines.

Good luck with that! I am telling you, it was guaranteed to fail. We never saw it fail because it got dismantled.

People still fantasize about sitting on that control room and those unergonomic chairs, controlling the economy using 8 buttons at a time.

[nahuel0x]

It seems it worked in some way:

The programme was initially viewed with traditionalist scepticism by Allende’s Socialist Party. But Project Cybersyn’s hour arrived in October 1972 during a strike of 40,000 truck drivers led by the hard-right Confederación Nacional del Transporte. As Allende’s opponents sought to wreck the economy by preventing the transport of food and raw materials, Cybersyn was deployed to underwrite the resistance. Through the electronic network, the government was able to coordinate deliveries by active trucks and to evade blockades. “We felt that we were in the centre of the universe,” Espejo remarked.

After 24 days, the strike was defeated. Ministers then became “much more interested” in Cybersyn, Espejo told me. Allende even proposed transferring the operations room to La Moneda, the presidential palace.

On 10 September 1973, the government finally prepared to install an upgraded version of the system. But the following day, with the connivance of the CIA, Pinochet’s forces stormed the palace and bombed it from the air

From: https://www.newstatesman.com/culture/observations/2018/08/pr...

[29athrowaway]

There is merit in Cybersyn. There's serious theory, analysis and work put into it by respectable individuals.

But if you could compare the performance of Cybersyn against a self-regulating free market economy, my instinct tellsm me that Cybersyn would underperform in comparison.

[maze-le]

What it was is essentially a supply chain management system... Something every big coorporation does today -- some of them on a scale comparable to a nation state.

[_vbnz]

Yeah, Amazon has a system to determine the most cost-effective shipping, including where to ship from to begin with, and even shipping the same item even if it belongs to another seller, if it is easier to ship.

It's not just a greedy algorithm like a market system to get the lowest costs for the first orders only.

[pietrovismara]

Then why shouldn't states do the same?

[maze-le]

I am not arguing for that, they should when the economic situation demands it. And they should be equally eager to drop it when the situation changes. I am all for more pragmatism and less ideology in terms of economic policy decisions, its not a one-way street.