AI and Communism: Narratives of AI Behind the Iron Curtain • Hacker News Board

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#101

Post by sudosysgen » Mon, Jul 05, 2021, 3:14 PM UTC

betwixthewires wrote: post #97

sudosysgen wrote: post #89

Earlier quoted context omitted.

The problem of maximizing your self-interest in a market is general enough that finding a solution to it can require solving an NP-hard problem. This is actually true, real financial problems are NP-hard. The market does not optimize for resource distribution, it optimizes for profit. It is tyrannical in that it removes your agency, as a member of the market you have no other choice but to optimize for profit in many…

Do you have an example of an NP hard problem a person would have to solve on the process of optimizing for self interest? A market optimizes for whatever the participant wants, profit is simply a mechanism by which the overall market performs price discovery. You always have a choice to not optimize for profit, I make trades all the time where profit is not my goal. Just choosing a job and being kindly allowed a smal…

I'm on my phone right now, but there is a paper on the exact P=NP efficient market question.

I don't assume that planning an economy would require solving NP-hard problems to outperform the market because I am certain that the market itself cannot solve them efficiently.

As for resources, us as a society already dedicate trillions of dollars a year to market planning - its called profits and capital gains, as well as the entire finance and real estate industry. As long as the planning system requires less than a trillion dollars a year to run even without being more efficient than the market at whichever cost function we choose it will be more efficient overall.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#102

Post by betwixthewires » Mon, Jul 05, 2021, 5:35 PM UTC

sudosysgen wrote: post #101

betwixthewires wrote: post #97

Earlier quoted context omitted.

Do you have an example of an NP hard problem a person would have to solve on the process of optimizing for self interest? A market optimizes for whatever the participant wants, profit is simply a mechanism by which the overall market performs price discovery. You always have a choice to not optimize for profit, I make trades all the time where profit is not my goal. Just choosing a job and being kindly allowed a smal…

I'm on my phone right now, but there is a paper on the exact P=NP efficient market question. I don't assume that planning an economy would require solving NP-hard problems to outperform the market because I am certain that the market itself cannot solve them efficiently. As for resources, us as a society already dedicate trillions of dollars a year to market planning - its called profits and capital gains, as well as…

> I don't assume that planning an economy would require solving NP-hard problems to outperform the market because I am certain that the market itself cannot solve them efficiently.

That sounds to me like an ideological position, which is fine of course.

But the question arises, if solving distribution problems requires a participant to solve NP hard problems as you've said (I'd appreciate a link to that paper whenever you can get around to it, I seriously do want to read it and am not just shouting "source", same goes for the distributed planning in anarchist planned economies we were discussing in another thread) then how would that not be true for a machine? Either the problem is NP hard or it isn't, regardless of what entity is trying to solve it.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#103

Post by betwixthewires » Mon, Jul 05, 2021, 5:39 PM UTC

lucian1900 wrote: post #100

betwixthewires wrote: post #87

Earlier quoted context omitted.

Do you have a source that food was destroyed to prevent it's price cratering and that caused starvation? Usually if there's demand for food it's price won't crater, and the fact that people were starving tells me there was said demand. Of course if that is the case that that happened, that's an external (government) intervention in a market. Planning is distributed, in market economies. Central planning by definition…

Before the revolution, landlords in agricultural areas were already selling food to the industrial cities, with prices subject to market forces. After the revolution, the state planned this trade and set fixed prices for food so industrial workers could afford it. Since these prices were lower than what the landlords wanted to extract, they destroyed food rather than sell it to the state. “Fraud, famine and fascism”…

> After the revolution, the state planned this trade and set fixed prices for food so industrial workers could afford it.

So the problem stemmed from central planning? Sounds like a good example for my point.

> ...some amount of centralisation is useful for optimisation...

See, my point is that this is not the case, and beyond that, that planning cannot accomplish more optimal distribution of resources that markets, and I'm using analogous network topology to demonstrate how this can be proven to be true.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#104

Post by lucian1900 » Mon, Jul 05, 2021, 5:56 PM UTC

betwixthewires wrote: post #103

lucian1900 wrote: post #100

Earlier quoted context omitted.

Before the revolution, landlords in agricultural areas were already selling food to the industrial cities, with prices subject to market forces. After the revolution, the state planned this trade and set fixed prices for food so industrial workers could afford it. Since these prices were lower than what the landlords wanted to extract, they destroyed food rather than sell it to the state. “Fraud, famine and fascism”…

> After the revolution, the state planned this trade and set fixed prices for food so industrial workers could afford it. So the problem stemmed from central planning? Sounds like a good example for my point. > ...some amount of centralisation is useful for optimisation... See, my point is that this is not the case, and beyond that, that planning cannot accomplish more optimal distribution of resources that markets,…

No, the fixed prices solved the problem of people starving due to not being able to afford food when market rates went up. Collectivisation later fixed the problem of sabotage by large landowners. Along with mechanisation, the poor periodic environmental conditions were also compensated for and thus the century-old periodic famines were ended.

Central planning could optimise further than the profit of individual landlords, since the USSR clearly prospered.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#105

Post by sudosysgen » Mon, Jul 05, 2021, 7:22 PM UTC

betwixthewires wrote: post #102

sudosysgen wrote: post #101

Earlier quoted context omitted.

I'm on my phone right now, but there is a paper on the exact P=NP efficient market question. I don't assume that planning an economy would require solving NP-hard problems to outperform the market because I am certain that the market itself cannot solve them efficiently. As for resources, us as a society already dedicate trillions of dollars a year to market planning - its called profits and capital gains, as well as…

> I don't assume that planning an economy would require solving NP-hard problems to outperform the market because I am certain that the market itself cannot solve them efficiently. That sounds to me like an ideological position, which is fine of course. But the question arises, if solving distribution problems requires a participant to solve NP hard problems as you've said (I'd appreciate a link to that paper wheneve…

It's not an ideological position - I don't think that P=NP and I don't think that the Church-Turing hypothesis is false so I don't think that markets can solve NP problems efficiently and thus I don't think markets are efficient.

NP-hard problems can be approximated. You get the machine to approximate a solution that is better than what the market could do or thereabouts.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#106

Post by nradov » Mon, Jul 05, 2021, 7:55 PM UTC

lucian1900 wrote: post #104

betwixthewires wrote: post #103

Earlier quoted context omitted.

> After the revolution, the state planned this trade and set fixed prices for food so industrial workers could afford it. So the problem stemmed from central planning? Sounds like a good example for my point. > ...some amount of centralisation is useful for optimisation... See, my point is that this is not the case, and beyond that, that planning cannot accomplish more optimal distribution of resources that markets,…

No, the fixed prices solved the problem of people starving due to not being able to afford food when market rates went up. Collectivisation later fixed the problem of sabotage by large landowners. Along with mechanisation, the poor periodic environmental conditions were also compensated for and thus the century-old periodic famines were ended. Central planning could optimise further than the profit of individual land…

No, fixing prices below the market rate inevitably causes shortages. There is no way to prevent this. Central planning has always failed.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#107

Post by nradov » Mon, Jul 05, 2021, 8:00 PM UTC

sudosysgen wrote: post #91

nradov wrote: post #72

Earlier quoted context omitted.

The Soviet famines were intentional genocide. Communist Party leaders always had plenty to eat. https://victimsofcommunism.org/

This used to be believed by academic historians, until the opening of the Soviet Archives in 1992, and now the consensus is that it wasn't, it was just really bad data. The link you're citing had multiple authors resign due to academic dishonesty and is funded by the US government, therefore making it state propaganda.

People died. That was not due to bad data.

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#108

Post by lucian1900 » Mon, Jul 05, 2021, 8:04 PM UTC

betwixthewires wrote: post #99

sudosysgen wrote: post #90

Earlier quoted context omitted.

The conclusion doesn't follow from the premises. As for food being always produced in market economies, see the Bengal famine. Food wasn't destroyed in the Holodonor, it was simply redistributed. Similar famines happened under capitalism, all it takes is for a crop to come up first or for cash crops combined with the anarchy of production to lead to a crop that is beneath the requirements. So no, similar famines did…

Optimal does not mean perfect. Externalities exist. For example, a meteor could hit the great plains and cause a famine. The market cannot solve that, but it can optimize available resources. Markets emerge spontaneously, like evolution. Evolution functions more optimally than if there were theoretically some god pulling levers. Not perfect, optimally. Markets function better than planned economies for the same reaso…

Deliberate design always beats evolution, given the same resources. Why should we be at the mercy of random chances to improve our lives when we could directly steer production for our use?

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#109

Post by betwixthewires » Mon, Jul 05, 2021, 9:58 PM UTC

sudosysgen wrote: post #105

betwixthewires wrote: post #102

Earlier quoted context omitted.

> I don't assume that planning an economy would require solving NP-hard problems to outperform the market because I am certain that the market itself cannot solve them efficiently. That sounds to me like an ideological position, which is fine of course. But the question arises, if solving distribution problems requires a participant to solve NP hard problems as you've said (I'd appreciate a link to that paper wheneve…

It's not an ideological position - I don't think that P=NP and I don't think that the Church-Turing hypothesis is false so I don't think that markets can solve NP problems efficiently and thus I don't think markets are efficient. NP-hard problems can be approximated. You get the machine to approximate a solution that is better than what the market could do or thereabouts.

Why can't the market approximate NP problems at least as good as a central planning entity?

Re: AI and Communism: Narratives of AI Behind the Iron Curtain

#110

Post by betwixthewires » Tue, Jul 06, 2021, 12:39 AM UTC

lucian1900 wrote: post #108

betwixthewires wrote: post #99

Earlier quoted context omitted.

Optimal does not mean perfect. Externalities exist. For example, a meteor could hit the great plains and cause a famine. The market cannot solve that, but it can optimize available resources. Markets emerge spontaneously, like evolution. Evolution functions more optimally than if there were theoretically some god pulling levers. Not perfect, optimally. Markets function better than planned economies for the same reaso…

Deliberate design always beats evolution, given the same resources. Why should we be at the mercy of random chances to improve our lives when we could directly steer production for our use?

That's a hard assertion with no real evidence, and it sounds like a truism.

Do you think climate change is an existential crisis? If you do there's flat evidence right there that deliberate design doesn't always beat evolution.

The idea that humans can engineer our way out of everything is hubris. Natural emergence seamlessly takes every variable into account. With man made things, they are designed within a scope so as a result there are always externalities. Literally every problem people are arrogant enough to think they can solve with central planning is an unintended consequence of attempts at central planning.

These unintended consequences are the direct result of part of what I'm talking about. Natural, emergent systems take everything into account because the entire system performs the information processing on behalf of itself. A central planner is incapable of that.