If I were a statist Marxist I'd strongly point to the fact that the Soviet union *didn't even use computers to do its central planning*: how the hell are you even supposed to be doing that?
Project Cybersyn was the millennial blossom that got snuffed out bc chance.
But today we have supercomputers and our economies aren't *that* more complicated, planned economies would just work.
@niplav “let computers plan the economy” is an idea that people seem to be constantly reproposing and which was debunked before computers were even a thing
@sophon
I have not read Hayek
@niplav ok well hayek was based 
@sophon
But I don't think anyone has *tried*, right? And Coase's theory of the firm + the fact that markets are probably not performing magic gestures at "there's an equivalence here"
@niplav hmm, not elaborating on this, sorry 
except to say that if something like cybersyn did “work” we would have had larger problems to deal with than economics 📎
@niplav in any case, I am opposed to planned economies not just for practical reasons but because they are an inherent infringement on freedom. become unplannable
@sophon i agree w/ the ideological disagreement w planned economies on liberty grounds, and also contend that while i too am a Project Cybersyn stan planned economies suffer from the classic scaling problem of centralization, which is that of realtime information-processing. there's just too much local information missing at the CPU to make good decisions at the far reaches of the system
@sophon consider that a hub-and-spoke architecture (which a planned economy is) has to sense local conditions in markets, transmit this information to the central hub, process it to generate a plan, push it back out, actuate the plan, and then repeat the process ad-hoc; the further the spokes are the more latency is introduced, the less intervention resolution the system has
@sophon @niplav i meant to be tagging you too but i'm still figuring out masto tbh 🤦♂️
hub-and-spoke systems fail to scale past a certain point, and a classic solution to this classic problem is delegating local decision-making to local actors, ie pushing computation to the edge, which is precisely what markets do
@pee_zombie @sophon
strongly agree that the frame is good
however a bunch of the computation in markets is redundant or (and I think this doesn't have a good computation equivalent?) caught up in zero sum games (and what's the computation-equivalent of externalities?)
i don't think that central planning is better than markets, but i think anti-central-planning people are underestimating the feasibility
@niplav externalities = side effects? but they're by definition out of scope for the optimization problem; if we want to minimize them, they can be brought in-scope
redundant computation is not something inherently to be avoided unless efficiency is explicitly an optimization goal, and even then its only to be minimized to the extent its not necessary to solve the problem; consider proof of work
@pee_zombie
externalities=subspace optimization
@icedquinn @niplav Where did you pull this from?
@icedquinn @niplav One does not need to have a detailed supermodel of human behaviour to plan an economy. Plans in the USSR where a long thing with many revisions that started with proposals from the local administations. For any product, one can run a combination of asking people how much they want/needed and look at past plan and production batches to adjust the number.
There's also the equivalent of "how much have I sold" feedback: you simply check how many leftovers you at to the point of mass distribution, or if you run out of a product earlier than anticipated.
@icedquinn @niplav Dunno where you heard that.
As I said above, plans had many stakeholders
@icedquinn @niplav Did they indicate the year?
@niplav From what little I know they had a series of skrews up that ultimately ended up with them not using computers as you said.
In the early 50's, they didn't think digitalization would go that far, so they focused mostly on mechanical computers. Then when elettrical computation became the clear winner, they attempted to catch up, and even if they eventually made their own computers (the original Elektra microarchitecture) they were behind, and so they opted to reverse engeneer wester designs instead (mostly DEC stuff I think).
That is until CPUs became waaay to complex to reverse engeneer in a decent timeframe.
Do yeah, they skrewed themselfs up. Moral of the story: just make a RISC processor.
cyberstalinism