What computer science can teach economics (2009) • Hacker News Board

Re: What computer science can teach economics (2009)

#21

Post by curuinor » Mon, Feb 08, 2016, 11:37 PM UTC

TeMPOraL wrote: post #17

curuinor wrote: post #2

H. Simon was among the founders of artificial intelligence and a very important microeconomicist, and I have often thought that these disciplines are not fundamentally different. What problem does backpropagation solve? Credit assignment. What problem does the market solve? Credit assignment. How do they do it? Localization of computation.

I think it's more general than that. Both backpropagation and the market introduce a feedback loop . That's the source of the behaviour. You can see similar phenomena in other closed-loop systems. That's why I think control theory is one of the best things humans invented ;).

Positive feedback loop, specifically. Tesauro noted power-law times in BP learning and you can go off and look at highly skew distributions in BP weight assignments on your own. Simon had a little program of research on skew distributions in economics (you cannot throw a stone but you hit one: money, firm size, investments, trades, income, GDP, yadda yadda), and how they are best explained by positive feedback loops.

Re: What computer science can teach economics (2009)

#22

Post by pash » Tue, Feb 09, 2016, 12:23 AM UTC

Economists have been investigating these concerns for decades. There is work dating from the late 1960s on the computability of general equilibria under uncertainty, and concerns about computability have played a role in the influential literature on bounded rationality. A small sub-field of the discipline, usually called "computable economics", began to appear in the mid-1990s, when K. Vela Velupillai [0] began to publish papers on the topic.

Velupillai is a central figure in computable economics, and you will find references to much of the work that's been done in the area if you search for his name or browse Google Scholar for his papers [1] (and the papers that cite his).

0. https://en.wikipedia.org/wiki/Vela_Velupillai

1. https://scholar.google.com/scholar?q=KV+Velupillai

Re: What computer science can teach economics (2009)

#23

Post by clavalle » Tue, Feb 09, 2016, 12:44 AM UTC

icebraining wrote: post #20

clavalle wrote: post #18

Earlier quoted context omitted.

That is interesting but seems to put too much onus on the poor individual. >poverty makes people feel powerless and blunts their aspirations What if it is more than a feeling? What if they are not irrational but seeing quite rationally what transactions await them? Let's assume that two sides of a potential transaction are rational, and further assume that each side will maximize the value they have after the transac…

Yes, a non-poor person has a better BATNA, therefore they can demand more from the other party. But I think the studies show that the poverty has an effect even beyond that.

I guess my problem with studies that show the other effects of poverty is that they seem to paint the poor as just hapless, irrational, creatures; subhuman and deserving of their plight since they can't reason their way out of it.

And, who knows, they may be, but I'm going to have to see a lot more proof than what I've seen so far which tend to be of the vein 'well, look at their circumstances. You wouldn't make that decision that led them there. I wouldn't make that decision that led them there. Obviously they are not rational like you or me'.

It is very strange that everyone in the economic model of the world is assumed to be rational but the poor.

Re: What computer science can teach economics (2009)

#24

Post by icebraining » Tue, Feb 09, 2016, 1:31 AM UTC

clavalle wrote: post #23

icebraining wrote: post #20

Earlier quoted context omitted.

Yes, a non-poor person has a better BATNA, therefore they can demand more from the other party. But I think the studies show that the poverty has an effect even beyond that.

I guess my problem with studies that show the other effects of poverty is that they seem to paint the poor as just hapless, irrational, creatures; subhuman and deserving of their plight since they can't reason their way out of it. And, who knows, they may be, but I'm going to have to see a lot more proof than what I've seen so far which tend to be of the vein 'well, look at their circumstances. You wouldn't make that…

While your concern is valid - patronizing and disrespecting the poor is rampant - I don't think that's a consequence of observing this effect; there's usually a prior belief. After all, the evidence is actually against the "subhuman" hypothesis - it shows that the reduced quality of the decisions is specifically a result of the environment, not an innate deficiency of some individuals.

Also, the assumption that everyone is rational is no longer unquestionable, ever since behavioral economics started gaining prominence.