The efforts to sabotage and disband central planning started as early as 1954.
In 1954 an executive order of 14 Oct 1954 reduced the amount of administrative personnel by 450 thousands.
The amount of metrics went down from 9 940 to 6 308 in 1954, to 3 081 in 1955, and to 1 780 in 1958.
Khrushchev moved most of the planning power from central planning institution to the regions and down to the factories and enterprises. What previously was strict targets from the center now became soft suggestions.
Article misses those basic things completely and instead goes into tangential theoretic topics and even goes to state things like that:
central planners preferred to use the CIA’s analyses of certain Russian commodities rather than reports from local Party bosses
Source: the comment #162 in some blog. Talking about quality of the data...
Fine you know more History but I don't think that the point of the article is to be historically accurate but rather to provide a theoretical explanation of why centrally planned economy is in fact quite difficult because of combinatorial explosion
It is difficult, this is why it required a whole institution. And not just an institution, but with proper mandate as well.
It also can be necessary.
In fact, Fred Brooks in The Mythical Man-Month, specifically in chapters 3 "The Surgical Team" and 4 "Aristocracy, Democracy, and System Design" explains why strong (even, maybe, authoritarian) central planning is imperative for large [software] projects.
Without proper mandate the work would be undermined constantly. Again, with regard to software projects (which with all accustomed with) a "Broken Ownership" article [0] [1] illustrates (among other things) the necessity of the mandate.
All in all, if you've been stripped of the mandate, lacking resources and your project is sabotaged in other ways, you don't need to look too far in the search of why it doesn't go as ...planned.
V. Gluskov designed OGAS to help with planned economy but for whatever reason it got denied funding. Decades later we're trying to make AI do just that on global scale.
From WIKI:
In 2016, a book dedicated to OGAS was published in the US, entitled How Not to Network a Nation: The Uneasy History of the Soviet Internet, by Benjamin Peters, professor at University of Tulsa. Harvard University professor Jonathan Zittrain commented that the book "fills an important gap in the Internet's history, highlighting the ways in which generativity and openness have been essential to networked innovation".[15] A reviewer of the book at MIT wrote:[16] "Soviet attempts to build a national computer network were undone by socialists who seemed to behave like capitalists".
I understand why it's appealing for techies, but I feel like talks about importance of OGAS are very much blown out of proportion.
Not that the project didn't have merit on itself (we're talking about proto-internet, basically), but its assumed role in planning is exaggerated. If anything, it's a pitch before MVP.
The issues were very much political (denied funding speaks in favor of that), not computational. You can't solve those with math and transistors.
I wonder if one could make a fun and entertaining game that's inspired by Soviet central planning and Factorio alike - so much of Factorio is balancing raw inputs, intermediate products, and the need for "final" products (e.g., ammunition, rocket fuel, etc.).
I think the steel tube example already shows that such a “game” would already be unplayable for even the 99.9th percentile HN reader.
For example, it would likely take dozens to hundreds of hours of study focused on steel tubes alone due to the very high dimensionality of real world things. And assuming a minimal forgetting rate of 1 fact for every 1000 facts after every 100 hours awake, they would pretty much be stuck forever on just the relatively simple goods.
Maybe a game with hundreds of times fewer dimensions then the real world, but then it might as well be pure fantasy anyways.
Of course you can. E.g. the steel production in the US has been government-planned (well, DoD-planned) since the WW2, and it's not even that big of a secret, just not something that's mentioned often.
And with modern analytics it’s harder to make up the numbers like they did. Also cheaper to transmit and centralize the data, especially with some IOT type stuff. Doesn’t make it immune to human nature, but modern tech reduces the friction and increases the transparency.
There is a book "The People's Republic of Walmart: How the World's Biggest Corporations are Laying the Foundation for Socialism" describing how e.g. Walmart (there is also Amazon now) does its own large-scale planning.
It's kinda funny that the "socialist calculation debate" has largely settled on "okay, it's probably possible but requires way too much of computational resources and accurate data gathering" — and now we spend insane amount of resources on AI data centers and tracking/surveillance on literally everything. And, of course, all of those resources won't end up being used for solving (or even accurately figuring out) the actual economical problems we face today.
Markets are optimal but capitalism is not. If you let supply and demand set prices but don't allow extracting value you get an economy that is both fair and efficient. The rule is all profits must be reinvested or paid out to workers and the state can impose price controls if margins are too high.
Yugoslavia tried it and they had the best economy from the eastern bloc growing faster than a lot of the west. China and Vietnam went further in their reforms and allowed private investment but they still kept state owned corporations instead of privatizing everything, and they don't worship markets and property like America does.
I don't think it's an accident that all these mixed economies got ahead of pure socialist or capitalist systems.
> The rule is all profits must be reinvested or paid out
We're already there today with struggles to properly allocate capital etc. If you pay (workers or investors/capital owners) too much, the facilities break down etc. and eventually you can't meet overall demand cashing prices to go up, encouraging higher investment (right when the workers want more payments because revenue's increasing, and maybe their personal costs) but maybe investment everywhere goes up, increasing capacity too much and tanking revenues (the business cycle) etc.
> Yugoslavia
For a lot of the time, East Germany by many metrics (GDP/cap etc. ignoring the (stronger) surveillance state etc.) was ahead of Yugoslavia (but because of a lot of debt, leading to East German stagnation in the 80s). But Slovenia (the wealthiest part of Yugoslavia) even surpassed the UK and Italy's standard of living in the 80s! (Northern Italy was still better off, though.)
I assume the parent comment meant extract as in vampiric. Collecting rents merely for owning assets rather than doing something useful.
China certainly allows you to get rich from starting a company so long as you're well-connected politically and/or obey the party. There is plenty of benefit to be had in a way absolutely not possible for someone in the USSR.
The computational complexity is a bit of a red herring. Finding the most optimal solution is hard. But if you settle for a close-to-optimal solution, then it can be solved in a much more reasonable time.
It is worrying that in the first "describing history" he describes the main data quality problem being that factory managers and party bosses outright lied to the central planners and to the communist party in Moscow, and then totally disregards that problem in the data quality issues for the future. Elsewhere you find the same: the very people who needed to secure against lying factory managers simply took their cut.
I've heard the same problem described from Chinese people as occurring on an equally large scale in China, even now. That what people the control Beijing has is vastly overstated due to the very widespread lying to acquire resources. Followed by complaints that the lying goes the other way as well: local governments telling home owners that a rail line will be built through their house, "disown them" with documents nobody in government knows about, then a company owned by a party boss builds a new house there and sells it, or doesn't even build anything, just sells the house they just acquired through fraud. Same with businesses.
I'm not sure what the fix is for this, but I can imagine that this is how the game is played in any authoritarian system. I doubt it will be much of a challenge to lie to the "internet of things".
The efforts to sabotage and disband central planning started as early as 1954.
In 1954 an executive order of 14 Oct 1954 reduced the amount of administrative personnel by 450 thousands.
The amount of metrics went down from 9 940 to 6 308 in 1954, to 3 081 in 1955, and to 1 780 in 1958.
Khrushchev moved most of the planning power from central planning institution to the regions and down to the factories and enterprises. What previously was strict targets from the center now became soft suggestions.
Article misses those basic things completely and instead goes into tangential theoretic topics and even goes to state things like that:
Source: the comment #162 in some blog. Talking about quality of the data...Fine you know more History but I don't think that the point of the article is to be historically accurate but rather to provide a theoretical explanation of why centrally planned economy is in fact quite difficult because of combinatorial explosion
It is difficult, this is why it required a whole institution. And not just an institution, but with proper mandate as well.
It also can be necessary.
In fact, Fred Brooks in The Mythical Man-Month, specifically in chapters 3 "The Surgical Team" and 4 "Aristocracy, Democracy, and System Design" explains why strong (even, maybe, authoritarian) central planning is imperative for large [software] projects.
Without proper mandate the work would be undermined constantly. Again, with regard to software projects (which with all accustomed with) a "Broken Ownership" article [0] [1] illustrates (among other things) the necessity of the mandate.
All in all, if you've been stripped of the mandate, lacking resources and your project is sabotaged in other ways, you don't need to look too far in the search of why it doesn't go as ...planned.
[0] https://blog.alexewerlof.com/p/broken-ownership
[1] https://news.ycombinator.com/item?id=37225397
If they were not accurate about history, what else did they get wrong?
V. Gluskov designed OGAS to help with planned economy but for whatever reason it got denied funding. Decades later we're trying to make AI do just that on global scale.
From WIKI:
In 2016, a book dedicated to OGAS was published in the US, entitled How Not to Network a Nation: The Uneasy History of the Soviet Internet, by Benjamin Peters, professor at University of Tulsa. Harvard University professor Jonathan Zittrain commented that the book "fills an important gap in the Internet's history, highlighting the ways in which generativity and openness have been essential to networked innovation".[15] A reviewer of the book at MIT wrote:[16] "Soviet attempts to build a national computer network were undone by socialists who seemed to behave like capitalists".
I understand why it's appealing for techies, but I feel like talks about importance of OGAS are very much blown out of proportion.
Not that the project didn't have merit on itself (we're talking about proto-internet, basically), but its assumed role in planning is exaggerated. If anything, it's a pitch before MVP.
The issues were very much political (denied funding speaks in favor of that), not computational. You can't solve those with math and transistors.
I wonder what a mega-bureaucracy but managed by AI would be like.. :)
I wonder if one could make a fun and entertaining game that's inspired by Soviet central planning and Factorio alike - so much of Factorio is balancing raw inputs, intermediate products, and the need for "final" products (e.g., ammunition, rocket fuel, etc.).
I think the steel tube example already shows that such a “game” would already be unplayable for even the 99.9th percentile HN reader.
For example, it would likely take dozens to hundreds of hours of study focused on steel tubes alone due to the very high dimensionality of real world things. And assuming a minimal forgetting rate of 1 fact for every 1000 facts after every 100 hours awake, they would pretty much be stuck forever on just the relatively simple goods.
Maybe a game with hundreds of times fewer dimensions then the real world, but then it might as well be pure fantasy anyways.
Oh, it would certainly have to be simplified to the level of Factorio. Not a thousand kinds of steel tube, but just one kind.
I suspect that you can definitely use planned economy tools down to a certain scale and then let the markets sort the rest.
At some level, Indian and Chinese economies have their development centrally planned, not just financially regulated, like in the West.
Of course you can. E.g. the steel production in the US has been government-planned (well, DoD-planned) since the WW2, and it's not even that big of a secret, just not something that's mentioned often.
And with modern analytics it’s harder to make up the numbers like they did. Also cheaper to transmit and centralize the data, especially with some IOT type stuff. Doesn’t make it immune to human nature, but modern tech reduces the friction and increases the transparency.
There is a book "The People's Republic of Walmart: How the World's Biggest Corporations are Laying the Foundation for Socialism" describing how e.g. Walmart (there is also Amazon now) does its own large-scale planning.
It's kinda funny that the "socialist calculation debate" has largely settled on "okay, it's probably possible but requires way too much of computational resources and accurate data gathering" — and now we spend insane amount of resources on AI data centers and tracking/surveillance on literally everything. And, of course, all of those resources won't end up being used for solving (or even accurately figuring out) the actual economical problems we face today.
Probably a lesson in all this for management-by-poorly-conceived-metrics dashboard enjoyers
https://bactra.org/weblog/918.html
In USSR, Optimization Solves You.
Markets are optimal but capitalism is not. If you let supply and demand set prices but don't allow extracting value you get an economy that is both fair and efficient. The rule is all profits must be reinvested or paid out to workers and the state can impose price controls if margins are too high.
Yugoslavia tried it and they had the best economy from the eastern bloc growing faster than a lot of the west. China and Vietnam went further in their reforms and allowed private investment but they still kept state owned corporations instead of privatizing everything, and they don't worship markets and property like America does.
I don't think it's an accident that all these mixed economies got ahead of pure socialist or capitalist systems.
> The rule is all profits must be reinvested or paid out
We're already there today with struggles to properly allocate capital etc. If you pay (workers or investors/capital owners) too much, the facilities break down etc. and eventually you can't meet overall demand cashing prices to go up, encouraging higher investment (right when the workers want more payments because revenue's increasing, and maybe their personal costs) but maybe investment everywhere goes up, increasing capacity too much and tanking revenues (the business cycle) etc.
> Yugoslavia
For a lot of the time, East Germany by many metrics (GDP/cap etc. ignoring the (stronger) surveillance state etc.) was ahead of Yugoslavia (but because of a lot of debt, leading to East German stagnation in the 80s). But Slovenia (the wealthiest part of Yugoslavia) even surpassed the UK and Italy's standard of living in the 80s! (Northern Italy was still better off, though.)
If I'm not allowed to extract value, what is the incentive to start or invest in a business? Why bother owning or running a company?
I assume the parent comment meant extract as in vampiric. Collecting rents merely for owning assets rather than doing something useful.
China certainly allows you to get rich from starting a company so long as you're well-connected politically and/or obey the party. There is plenty of benefit to be had in a way absolutely not possible for someone in the USSR.
Do you have any source defending your point better than you can in small comment ? I am interested but a bit unconvinced so far
Ah, the familiar twinge of sorrow and rue at reading the name "Yugoslavia"...
"Red Plenty" the book being discussed in the article is well worth a read.
The computational complexity is a bit of a red herring. Finding the most optimal solution is hard. But if you settle for a close-to-optimal solution, then it can be solved in a much more reasonable time.
(2016) but still worth the read
Ooopsie. Sorry I can't edit anymore. @dang Would be kind enough to add the year ? (I know this is not supposed to work but I know it works)
It has also been discussed in 2020 (https://news.ycombinator.com/item?id=25084479) with 200 comments, and 2017 (https://news.ycombinator.com/item?id=14515225) with 171 comments
It is worrying that in the first "describing history" he describes the main data quality problem being that factory managers and party bosses outright lied to the central planners and to the communist party in Moscow, and then totally disregards that problem in the data quality issues for the future. Elsewhere you find the same: the very people who needed to secure against lying factory managers simply took their cut.
I've heard the same problem described from Chinese people as occurring on an equally large scale in China, even now. That what people the control Beijing has is vastly overstated due to the very widespread lying to acquire resources. Followed by complaints that the lying goes the other way as well: local governments telling home owners that a rail line will be built through their house, "disown them" with documents nobody in government knows about, then a company owned by a party boss builds a new house there and sells it, or doesn't even build anything, just sells the house they just acquired through fraud. Same with businesses.
I'm not sure what the fix is for this, but I can imagine that this is how the game is played in any authoritarian system. I doubt it will be much of a challenge to lie to the "internet of things".
The fix is to realize that if your system requires changing human behavior you should downweight it against alternatives