I think it actually was. Central planning is a Bad Thing because Soviets did it and it went bad, so nowadays its inferiority is treated as an axiom. But back then optimization theory was just starting to develop and was never actually applied in practice, and we didn't have enough computing power, nor instant communication infrastructure. I think we should revisit the concept of central planning with algorithms doing global optimizations based on real-time signals.
as a person who lived in yet another centrally planned country, let me tell you this, plain simply - it doesn't work. soviets failed, but so every other nation that went that way. let's call it slow but steady financial rotting that will bring any country down, given enough time. there are plenty of reasons for that, they are always the same, but for me one is most important - they all work with ideal person who wants to help the system, instead of disconnected greedy selfish individuals, that real society consist of.
this is of course valid for humans, who have all kinds of flaws, some more than others. so you propose to give power to software, who by principle always has... flaws. imagine how juicy this thing would be for hackers, who would have billions-big incentives from shady business people to nudge decisions in their favors... silly naive people :)
What I would suggest - let some state of the art simulation run along the humans, and evaluate its decisions after some time passes, to clearly see who made the right one, in long term. after there is some long term perception that computer is clear winner (definitely not below 10-20 year mark of steady performance), we can move that way. and always, always have some sort of semi-neutral commitee to oversee and accept/reject all important stuff (with necessary reasoning, not on whims of current most powerful one). make it as transparent as possible. -- do you think this is real? no way, when looking at how shady today's politics are done, be it US, europe or any other place...
The issue with central planning is that it isn't robust - if the planners mess up, everyone is screwed. If we could inject a method by which a failed plan would take down only the planners (and not consumers), it would help a lot. For example, we could allow large centrally planned organizations to exist, but if they fail to meet consumer needs they get dissolved and consumers can go to a different planning board.
Wait, we already have such central planning boards - Uber, Fedex, Walmart, etc. My mistake.
On the other hand, such distributed planning is incredibly inefficient - how much resources are wasted by Uber competing with Lyft competing with Taxis, by FedEx competing with UPS competing with post office, etc.? The point is, it's a tradeoff - efficiency with single points of failure vs. resilience at a cost of huge waste.
Goods are produced that no one needs. The market gets randomly stuck at local minimas that lead, for instance, to throwing away tons of food because it's more profitable to do that than to donate to those in need. You get tons and tons of duplication by companies competing with each other over a limited set of customers, which also feeds the advertising industry that wastes resources on zero sum games. In current model, the economy is regulated implicitly, by feedback loops fighting against each other creating a dynamic equilibrium. It's inherently less efficient than regulating explicitly. My point is that while humans are not capable of regulating on such scale and thus have to rely on implicit methods, optimization algorithms run on real-time data feeds may in fact be able to regulate explicitly.
Getting stuck in local minima happens in centrally planned systems too.
You are correct that things like advertising are (sometimes) zero sum - specifically, advertising based on irrational criteria. (Advertising which transmits useful information, e.g. "taxis suck and Uber exists", is not zero sum.)
But you also gloss over one of the big benefits of a non-centralized system - correctly finding the utility function. In a centrally planned system, your utility function needs to properly identify what tradeoffs people would make if they had the choice. This is not easy, and modern computing doesn't help very much here.
In contrast, consumer choice via price systems handles this nicely.
Say you're the engineer responsible for maintaining the Central Planning And Optimization Algorithm. What motivates you not to calibrate the algorithm in such a way that your family and your buddies benefit the most?
Nothing. Of course I'll be tempted to set things up so that those that matter to me have their problems solved. That's unfortunately human nature, that's what happens now (people can game distributed systems too), and I have no idea how to design a system that would be gaming-proof and perfectly egalitarian.
But even assuming that some people will abuse the system (as they do today and we still live with it), it seems plausible to me that the economy would still run better if everything was driven by global optimization algorithm fed with real-time data.
I've never understood why anybody would think that central planning would ever work. The very idea is total hubris - before you even get to the decision making process, you need accurate data to work from. It should be painfully obvious that data about (and especially provided by) humans is at best an approximate representation of reality. Often, it is completely wrong. Real-time noise is still noise.
If you want proof of this, consider why work-to-rule[1] works. While it is obviously intended to be used as a type of strike or similar direct action, the idea is that you are technically doing you job too accurately. The work continues to function thanks to the fact that almost nobody does everything they are technically supposed to do. We speed on the way to work, we overlook minor mistakes, we fix instructions and reinterpret orders. Reports are regularly sent up the chain of command that "yes, project $FOO was finished on time" even though everybody involved knows it wasn't.
The idea that you can take known-bad data, apply it to a fantastically complex model, iterate that model to add unknown amounts of chaotic noise, and get usable results is completely insane. Given the realities of how bureaucracies work, that noise will simply be re-interpreted until it says whatever the people involved want it to say. In the end, it is still a situation where the people making decisions are distanced from the consequences of their decisions.
If you want to fix some of the problems we have, I suggest that we need a lot more of the opposite of central planning: local decision making authority. A control system tends to work better when it can get accurate and immediate feedback, and in human terms that means experiencing - or at least directly seeing - the consequences of any decision. We, as a species, are notoriously terrible at evaluating any decision where the consequences are not felt immediately and personally.
What will they actually optimize for is an excellent question. We have plenty of central planning, its just most people don't like it very much. Look at the intentional destruction and elimination of the american middle class or the general population of Greece or how the criminal justice system treats black people in the USA. Statistically, they're unlikely to optimize for you, for most values of you.
Also don't forget the somewhat more intractable philosophical problem of whats actually good for you vs what you superficially want vs what you actually want, and our recent history of spending trillions on PR to intentionally confuse people about those three. I think we can safely assume that it'll be very cheap and easy to keep almost everyone unhappy about "the" central plan, which will be weaponized by factions.
Looking at how the disease of Puritanism has infected most of our culture pretty deeply, you can expect a similar meme infection in a centrally planned economy. So even assuming there is a "True" ideal state, we'll have infections and need an immune system. Merely computing faster or more complicated models is like assuming yeast are diseased and whales are disease free merely because of size and complication differential. What we do know from biology is the more mono-cultural and rigid something is, the more likely it is to get wiped out, and the analogy to central planning would be it'll inevitably get wiped out by meme diseases, so the best long term central plan might ironically be to not centrally plan.
Large corporations typically implement a form of central planning - that's what applications such as Oracle Hyperion and IBM Cognos are for - managing complex multidimensional models that integrate actual historical values with multiple forecast scenarios.
Another valid analogy with large corporations is if central planning was the optimum strategy, then extreme micromanagement from the top would be the optimum business strategy. Surely it would take top level supercomputers and programmers to run an entire country, but any random MBA with an Excel spreadsheet could run a business more successfully than a business without TPS reports and content-free numbers. So observationally, micromanagement and management by metric number should be a sign of a highly profitable and successful company, yet its usually associated with dying companies or companies only kept alive by what boils down to corruption (monopolies, licensing, regulation, revolving door between .com and .gov, etc).
If there was a solid argument WRT scaling explaining why the result would be different at extreme scale vs business scale then the picture might be brighter for central planning.
There are also growth rate issues where your two examples are unfortunately companies experiencing an implosion of mindshare, people run from those companies as fast as they can, not run toward them. That is probably just an unfortunate coincidence but does look bad for central planning PR. There must be some "cool companies" that centrally plan and are successful? Or maybe there aren't?
Worth noting that Google seems to use Oracle Hyperion - you pretty much have to use tools like Hyperion for consolidated financial reporting once a company gets past a certain complexity level.
I think that any large business is already too large for the top to micromanage things. No human has enough time and free computing power in their head to do that. That's why I believe central planning run by humans is unworkable - we're too slow to handle the complexity.
But then again, successful micromanagement happens - it's usually called "strong leadership", "CEO with vision", etc. Apple with Jobs and Amazon with Bezos come to mind as past examples (not sure how Amazon is doing now, but there was a lot of praise written for Jeff's micromanagement), and the most visible current one would be Tesla and SpaceX under Elon Musk.
I would recommend looking to nature for patterns. Large networks of loosely interdependent organisms work best. Once one species (e.g. humans) become too dominant the whole system begins to fall apart.
There is no central planning in nature. It could be argued that central planning is an unnatural state. A far more robust solution would be to focus on having an economy made up of many small/medium size organizations and individual makers. Currently our economy with our "too big to fail" institutions is very fragile, as single points of failure can have catastrophic global effects. It would be much more healthy to eliminate the possibility of devastating single points of failure.
I could have sworn the old idea was that factories would shift from working for the executive suite to working for the workers. That is, make the workers the shareholders and board.
I'm not sure about appeal to nature here; biology seems rather centralized in a lot of aspects. An organism may be made of many cells, but each of those cells have a specific function. Kidney cells do not compute, and brain cells do not contract to pump blood. When a cell starts doing whatever is best for it locally, we call that "cancer". Centralized management arises naturally whenever there's a need for efficiency. So while there's a lot of distribution going on, I wouldn't say centralization is unnatural. It seems common enough on many different scales.
The body as you have described it is one node (an organization in my example) in the wider network. For that node to function it needs to have all of its components working in unison but if the node fails other nodes can adapt and the system keeps on chugging.
This layering of central and distributed seems common on all levels. If you go up from your example, you have an ecosystem that needs to work in unison. But if it fails, there are always other planets...
I'm as leftist as they come and I'm against central planning.
Bitcoin, removing gatekeepers, technologies that reduce/remove barriers to entry like 3d-printers, secure free-flowing information networks look much more promising for a future with empowered humans, free from debt and corporate/government overlords. Artificial scarcity, planned obsolescence and utter waste of resources both human and planetary could be solved if you combined tech like self-building machines, advanced recycling and permaculture principals.
I don't consider myself leftist (or rightist or any -ist) at all, I try to evaluate all the concepts on their merits. At the time of writing my original comment it seemed to me that computer-based central planning could be a good solution for some problems. Responses here brought up some arguments that made me reconsider the extent to which such central planning could be successfully applied. Maybe inefficiency is a price worth paying for increased resiliency and ability to adapt to environment without explicit specification.
I'm definitely not buying into the principle that everything needs to be distributed, libre and democratic. There are too many obvious failure modes there. Empowering everyone is good, but empowering everyone to the point when a random nutjob can kill millions with a home-made bioweapon doesn't. Nor the coordination prolems like tragedy of commons, oh so present around us everywhere, and one of the primary reasons things are being regulated.
This is actually an interesting hypothesis - has computing advanced enough to make central planning possible. I am not sure how you would test it though.
The biggest problems with central planning were never quite as much about the lack of computing power, but more about the incentives that go with that power structure. Politically, the central planner gets to reward all his friends and punish his enemies and deprive them of resources and opportunities. Even aside from the obvious implications on political liberty, getting the opportunity to do business in that environment is a matter of political connection, not merit, leading to stagnation.
I suppose, though, that you could at least use sufficiently advanced algorithms to monitor supply levels and inventories and sales to adjust prices so you can bring back some measure of the price-signaling mechanism that central planning has traditionally lacked... if the guy in charge isn't tempted to fiddle with prices for political reasons. Looking at some extant central-planning huggers like Tom Steyer (who wants to investigate/punish/tax oil firms because California regulations make gas more expensive than elsewhere in the nation) you can color me skeptical of the premise.
Then there's the fact that no one can know what any given person wants to buy at a given time, and therefore central planning is doomed to failure because you can't centrally plan "the right" prices for everything, and then you'll end up vastly misallocating resources, and there will be widespread destitution ala Soviet Union.
In conclusion, central planning ultimately just completely fucks up any economy. Check out Venezuela for a recent example.
So if 100% central planning results in a total catastrophe, then there should be no central planning at all, because any extent of central planning is harmful compared to less (or ultimately none at all).
So if 100% central planning results in a total catastrophe, then there should be no central planning at all, because any extent of central planning is harmful compared to less (or ultimately none at all).
I don't think your argument works. If I drink a swimming pool full of water, I will die. Does that mean I should drink no water at all?
Maybe it was poorly phrased. Here's another attempt.
Any degree of central planning is harmful to us because: 1) it always represents an opportunity cost compared to people making choices themselves, and 2) the choices made by central planners are not made to benefit the masses (because otherwise they'd just let people make the choices themselves)
What would be the point of being in a position to make decisions for other people, if you were just going to do your best to benefit others, instead of yourself?
This problem applies to a non-central planned economy too. Someone somewhere has to make a decision of what people want to buy and make it in advance of the item being purchased (hence inventory). This is why most business use central planning. If they get the planning wrong they lose money and potentially go out of business. The question is can this be scaled up now economy wide now that we have much more advanced computational tools to predict and react swiftly to demand changes. I don’t know :)
However you might want to define "central planning", there's a massive difference. The business people are using their own money, but central planners use everyone else's.
Actually most large businesses are using someone else's money - the shareholders. The same sort of agency problems that exist in centrally planned economies also exist in large public companies.
I didn’t think we were even arguing, just discussing an interesting idea :)
Our current economic system is remarkably similar to the centrally planned economy the socialist dreamed of creating. Most productive enterprises have no controlling owners - the ownership of most public companies is highly dispersed and much of it in the hands of pension funds, etc. About the only difference between a socialist “paradise" and what we have now is that the indirect ownership of the means of production is not evenly distributed within society.
The reason central planning fails (in societies or in companies) is because the people doing the work at the "bottom" have a lot more knowledge about how things really work than the people making the decisions at the top. A lot of that knowledge isn't even easy to capture - it's tied up in skills, techniques, oral history and obscure trade secrets.
In other words: the problem isn't likely to be speed/power of the planning computers, but how you get the information that needs to be fed into them.
I think this could also explain why managers who worked their way up from the bottom tend to do well, whereas professional executives who are helicoptered in to companies fail.
> I think this could also explain why managers who worked their way up from the bottom tend to do well, whereas professional executives who are helicoptered in to companies fail.
While intuitive, I have doubts about the truth of this statement. Citation needed.
I don't think it's principally processor bound. I think that the "intelligence" of a market is more in aggregating information than in running a complex algorithm. We probably have the computation resources to replace a market economy, but we have no way that I can see to get all the information needed into the computer.
We need to get better at writing software before we do that. The idea of putting software as it's written now in charge of planning a national economy is terrifying.
Well - the idea of putting the people who regularly crash economies in charge of writing that software is terrifying.
A good model would be an improvement on what we have now. But you need to talk to heterodox economists to get the good models, and that's unlikely to happen for political reasons.
See e.g. this http://crookedtimber.org/2012/05/30/in-soviet-union-optimiza... which is a discussion of the difficulties of computerized central planning and argues that -- even allowing for approximate solutions being good enough, and even allowing for computers to get much faster -- we don't have algorithms that would allow for real-time central planning of a real economy, and even if we did there would be substantial further problems in actually doing it.
(It's in the context of Francis Spufford's book "Red Plenty" -- which is actually pretty much about computerized central planning, I believe -- and it's by Cosma Shalizi who is a very smart guy, knows a lot about economics, and is sufficiently leftist to be more than averagely sympathetic to the idea of a planned economy.)
Problem with financial sector began when US Treasury dropped gold standard as basis for dollar. Since then FED has unrestricted ability to create "money" from the void. This leads to market bubbles, where the real money, earned on real-life activities such as manufacturing being lost in financial-market-style gambling. It is remarkable that so called "crysises" started to happen more often as financial sector grew.
"Our hypothesis is that it arises because finance tends to favour relatively low productivity industries as such industries usually own assets that are relatively easy to pledge as collateral."
My alternative hypothesis would be brain drain from production R&D into finances.
It's a bit unclear to me how the authors establish the direction of causality. Isn't it an equally valid interpretation that low productivity industries with lots of collateral cause the financial sector to grow?
The reason is that if you try to fit a quadratic regression to an S-curve, you'll come up with a downward sloping parabola (if the data points live near the right side of the S, if they live near the left you'll get an upward sloping one). That's simply a limitation of quadratic regression. Oops.
50 comments
[ 4.6 ms ] story [ 154 ms ] thread[1] https://en.wikipedia.org/wiki/Project_Cybersyn
this is of course valid for humans, who have all kinds of flaws, some more than others. so you propose to give power to software, who by principle always has... flaws. imagine how juicy this thing would be for hackers, who would have billions-big incentives from shady business people to nudge decisions in their favors... silly naive people :)
What I would suggest - let some state of the art simulation run along the humans, and evaluate its decisions after some time passes, to clearly see who made the right one, in long term. after there is some long term perception that computer is clear winner (definitely not below 10-20 year mark of steady performance), we can move that way. and always, always have some sort of semi-neutral commitee to oversee and accept/reject all important stuff (with necessary reasoning, not on whims of current most powerful one). make it as transparent as possible. -- do you think this is real? no way, when looking at how shady today's politics are done, be it US, europe or any other place...
Wait, we already have such central planning boards - Uber, Fedex, Walmart, etc. My mistake.
You are correct that things like advertising are (sometimes) zero sum - specifically, advertising based on irrational criteria. (Advertising which transmits useful information, e.g. "taxis suck and Uber exists", is not zero sum.)
But you also gloss over one of the big benefits of a non-centralized system - correctly finding the utility function. In a centrally planned system, your utility function needs to properly identify what tradeoffs people would make if they had the choice. This is not easy, and modern computing doesn't help very much here.
In contrast, consumer choice via price systems handles this nicely.
But even assuming that some people will abuse the system (as they do today and we still live with it), it seems plausible to me that the economy would still run better if everything was driven by global optimization algorithm fed with real-time data.
If you want proof of this, consider why work-to-rule[1] works. While it is obviously intended to be used as a type of strike or similar direct action, the idea is that you are technically doing you job too accurately. The work continues to function thanks to the fact that almost nobody does everything they are technically supposed to do. We speed on the way to work, we overlook minor mistakes, we fix instructions and reinterpret orders. Reports are regularly sent up the chain of command that "yes, project $FOO was finished on time" even though everybody involved knows it wasn't.
The idea that you can take known-bad data, apply it to a fantastically complex model, iterate that model to add unknown amounts of chaotic noise, and get usable results is completely insane. Given the realities of how bureaucracies work, that noise will simply be re-interpreted until it says whatever the people involved want it to say. In the end, it is still a situation where the people making decisions are distanced from the consequences of their decisions.
If you want to fix some of the problems we have, I suggest that we need a lot more of the opposite of central planning: local decision making authority. A control system tends to work better when it can get accurate and immediate feedback, and in human terms that means experiencing - or at least directly seeing - the consequences of any decision. We, as a species, are notoriously terrible at evaluating any decision where the consequences are not felt immediately and personally.
[1] http://libcom.org/organise/work-to-rule
Also don't forget the somewhat more intractable philosophical problem of whats actually good for you vs what you superficially want vs what you actually want, and our recent history of spending trillions on PR to intentionally confuse people about those three. I think we can safely assume that it'll be very cheap and easy to keep almost everyone unhappy about "the" central plan, which will be weaponized by factions.
Looking at how the disease of Puritanism has infected most of our culture pretty deeply, you can expect a similar meme infection in a centrally planned economy. So even assuming there is a "True" ideal state, we'll have infections and need an immune system. Merely computing faster or more complicated models is like assuming yeast are diseased and whales are disease free merely because of size and complication differential. What we do know from biology is the more mono-cultural and rigid something is, the more likely it is to get wiped out, and the analogy to central planning would be it'll inevitably get wiped out by meme diseases, so the best long term central plan might ironically be to not centrally plan.
If there was a solid argument WRT scaling explaining why the result would be different at extreme scale vs business scale then the picture might be brighter for central planning.
There are also growth rate issues where your two examples are unfortunately companies experiencing an implosion of mindshare, people run from those companies as fast as they can, not run toward them. That is probably just an unfortunate coincidence but does look bad for central planning PR. There must be some "cool companies" that centrally plan and are successful? Or maybe there aren't?
But then again, successful micromanagement happens - it's usually called "strong leadership", "CEO with vision", etc. Apple with Jobs and Amazon with Bezos come to mind as past examples (not sure how Amazon is doing now, but there was a lot of praise written for Jeff's micromanagement), and the most visible current one would be Tesla and SpaceX under Elon Musk.
There is no central planning in nature. It could be argued that central planning is an unnatural state. A far more robust solution would be to focus on having an economy made up of many small/medium size organizations and individual makers. Currently our economy with our "too big to fail" institutions is very fragile, as single points of failure can have catastrophic global effects. It would be much more healthy to eliminate the possibility of devastating single points of failure.
I'm definitely not buying into the principle that everything needs to be distributed, libre and democratic. There are too many obvious failure modes there. Empowering everyone is good, but empowering everyone to the point when a random nutjob can kill millions with a home-made bioweapon doesn't. Nor the coordination prolems like tragedy of commons, oh so present around us everywhere, and one of the primary reasons things are being regulated.
I suppose, though, that you could at least use sufficiently advanced algorithms to monitor supply levels and inventories and sales to adjust prices so you can bring back some measure of the price-signaling mechanism that central planning has traditionally lacked... if the guy in charge isn't tempted to fiddle with prices for political reasons. Looking at some extant central-planning huggers like Tom Steyer (who wants to investigate/punish/tax oil firms because California regulations make gas more expensive than elsewhere in the nation) you can color me skeptical of the premise.
In conclusion, central planning ultimately just completely fucks up any economy. Check out Venezuela for a recent example.
So if 100% central planning results in a total catastrophe, then there should be no central planning at all, because any extent of central planning is harmful compared to less (or ultimately none at all).
I don't think your argument works. If I drink a swimming pool full of water, I will die. Does that mean I should drink no water at all?
Any degree of central planning is harmful to us because: 1) it always represents an opportunity cost compared to people making choices themselves, and 2) the choices made by central planners are not made to benefit the masses (because otherwise they'd just let people make the choices themselves)
What would be the point of being in a position to make decisions for other people, if you were just going to do your best to benefit others, instead of yourself?
Our current economic system is remarkably similar to the centrally planned economy the socialist dreamed of creating. Most productive enterprises have no controlling owners - the ownership of most public companies is highly dispersed and much of it in the hands of pension funds, etc. About the only difference between a socialist “paradise" and what we have now is that the indirect ownership of the means of production is not evenly distributed within society.
In other words: the problem isn't likely to be speed/power of the planning computers, but how you get the information that needs to be fed into them.
I think this could also explain why managers who worked their way up from the bottom tend to do well, whereas professional executives who are helicoptered in to companies fail.
While intuitive, I have doubts about the truth of this statement. Citation needed.
A good model would be an improvement on what we have now. But you need to talk to heterodox economists to get the good models, and that's unlikely to happen for political reasons.
See e.g. this http://crookedtimber.org/2012/05/30/in-soviet-union-optimiza... which is a discussion of the difficulties of computerized central planning and argues that -- even allowing for approximate solutions being good enough, and even allowing for computers to get much faster -- we don't have algorithms that would allow for real-time central planning of a real economy, and even if we did there would be substantial further problems in actually doing it.
(It's in the context of Francis Spufford's book "Red Plenty" -- which is actually pretty much about computerized central planning, I believe -- and it's by Cosma Shalizi who is a very smart guy, knows a lot about economics, and is sufficiently leftist to be more than averagely sympathetic to the idea of a planned economy.)
My alternative hypothesis would be brain drain from production R&D into finances.
Isn't that the famous study which turned out to be [edit: almost] completely wrong because of errors in the excel formulas? https://en.wikipedia.org/wiki/Growth_in_a_Time_of_Debt#Metho...
The one paper they cite which isn't paywalled uses a rather flawed methodology: http://www.bis.org/publ/work381.pdf
The same methodology also shows that too many doctors, telephones and or R&D technicians inhibits growth: http://www.iie.com/publications/pb/pb15-9.pdf
The reason is that if you try to fit a quadratic regression to an S-curve, you'll come up with a downward sloping parabola (if the data points live near the right side of the S, if they live near the left you'll get an upward sloping one). That's simply a limitation of quadratic regression. Oops.