Game Theory

[Wake Up]

Just like to kick start a debate on game theory, if it is true and if so can it be applied to communism. I'm hardly an expert on the subject but it is a powerful theory and wish to know more...

In a nutshell game theory is that all humans are acting for themselves. In helping others they either advance there own cause or make themselves feel better.
A Classic example would be buying a girl/guy a drink at a bar. Nice think to do but their are ulterior motives at work here. Even if you are a genuinely nice person there may well be a small amount of so called 'game' theory at play here.


Personally I think it exists in different amounts in all people. Hence in my example a person may just be looking for sex but they may also be genuine, or a mix of the two.

Now if true does this effect leftist theory?
My first thoughts are , no it doesn't. Humans will keep on playing the game, no matter the social conditions. But our blocks on things like greed will stop the game being abused.

Anyone's 10 cc would be appreciated.

[Rosa Lichtenstein]

I think this should be in 'Research'.

However, if you abandon the idea that we all act in the way that game theorists say we do (and there is precious little evidece to suggest we do), then the theory falls apart.

It's a neat mathematical theory, but it relies on a bourgeois view of humanity: we are all social atoms, and are all ultimately selfish.

[JimFar]

If you scroll through the thread at:
http://www.revleft.com/vb/latest-issue-science-t44084/index.html (http://www.revleft.com/vb/../latest-issue-science-t44084/index.html)

There is a bit of discussion on game theory. Some of the
Analytical Marxist writers like John Roemer and Jon Elster
were rather taken with game theory.

[Rosa Lichtenstein]

Indeed, Jim, that was a direct consequence of their methodological individualism.

[Dean]

Just like to kick start a debate on game theory, if it is true and if so can it be applied to communism. I'm hardly an expert on the subject but it is a powerful theory and wish to know more...

In a nutshell game theory is that all humans are acting for themselves. In helping others they either advance there own cause or make themselves feel better.
A Classic example would be buying a girl/guy a drink at a bar. Nice think to do but their are ulterior motives at work here. Even if you are a genuinely nice person there may well be a small amount of so called 'game' theory at play here.


Personally I think it exists in different amounts in all people. Hence in my example a person may just be looking for sex but they may also be genuine, or a mix of the two.
This last paragraph exemplifies how woefully inaccurate this is. Each and every human action is self-interested, this is known. The reason being, you do not do things unless there is some compulsion that exists within you driving that. That would be an interest.

The problem with Game Theory is that it presupposes not that we are self-interested, but that that interest is somehow at odds with the interests of your fellow man. While leftists recognize that humans are self-interested, the "game theorists" believe that humans are selfish.


Now if true does this effect leftist theory?
My first thoughts are , no it doesn't. Humans will keep on playing the game, no matter the social conditions. But our blocks on things like greed will stop the game being abused.

Anyone's 10 cc would be appreciated.
The problem is that "the game" is a supposedly anarchistic society where all human beings act equally against each others' interests. It is the pinnacle of paranoia, where every single member of society is at total odds with each other member. Cooperation is ignored.

The theory was developed to help ward off nuclear war. The concept was to try to develop a foreign policy which insured that nuclear war was unlikely as possible, but if it had to happen, the U.S. would strike first.

Interestingly, their study has shown that almost nobody follows 'game theory' - a survey was done to test the theory - people were told (paraphrased):
You have happened upon a Jewel lost by a master criminal. He is willing to give you money for the jewel, but you both agree to put the jewel and the money in locations hundreds of miles apart. You are to hide the jewel and call him, who will tell you the location of the money. But just when you are about to call, you realize that if you give him a false location, you can go and take the money and still keep the jewel. But you know he is probably thinking the same thing. So the only 'rational' option is to lie to him: at worst, you still have the jewel, and at best, you have the money and the jewel. If you are honest, you risk losing everything.
Participants were given a choice: lie to him, or cooperate. An overwhelming majority choose to cooperate, and they found that the only groups of people who choose the 'rational' approach were economists and sociopaths. It is worth noting, too, that if everybody did choose the 'rational' option, the deal could never be made - the seller and buyer both suffer.

[Ratatosk]

Just like to kick start a debate on game theory, if it is true and if so can it be applied to communism. I'm hardly an expert on the subject but it is a powerful theory and wish to know more...

In a nutshell game theory is that all humans are acting for themselves. In helping others they either advance there own cause or make themselves feel better.No, game theory is a tool for analyzing situations in which there are actors which are trying to maximize their profit, in some sense of the word. It can be applied or misapplied, but it's just not true to say that a mathematical theory makes assumptions about how humans act.

The fact is, however, that human and human organizations often do act so as to maximize their profit, for example capitalist firms do. The point is not that they people are perfect utility-maximizers, the point is to be able approximately analyze what kinds of behaviour might arise under such and such conditions.

This last paragraph exemplifies how woefully inaccurate this is. Each and every human action is self-interested, this is known. The reason being, you do not do things unless there is some compulsion that exists within you driving that. That would be an interest.Why would it be have to be an interest? Why not habit or why could the action have no particular reason? Not to mention that the theory of inner compulsions preceding actions sounds much like the old Cartesian theory of the will.

You're redefining "self-interest" to make it completely vacuous.

Interestingly, their study has shown that almost nobody follows 'game theory' - a survey was done to test the theory - people were told (paraphrased):
You have happened upon a Jewel lost by a master criminal. He is willing to give you money for the jewel, but you both agree to put the jewel and the money in locations hundreds of miles apart. You are to hide the jewel and call him, who will tell you the location of the money. But just when you are about to call, you realize that if you give him a false location, you can go and take the money and still keep the jewel. But you know he is probably thinking the same thing. So the only 'rational' option is to lie to him: at worst, you still have the jewel, and at best, you have the money and the jewel. If you are honest, you risk losing everything.
Participants were given a choice: lie to him, or cooperate.Regardless of whether the conclusion is true or not, I don't understand how an experiment where people are told to imagine that they are in such a dilemma would imply anything about their actual behaviour in similar situations.

That's kinda like asking people how they would behave if they were prison guards vs. Zimbardo.

The problem is that "the game" is a supposedly anarchistic society where all human beings act equally against each others' interests. It is the pinnacle of paranoia, where every single member of society is at total odds with each other member. Cooperation is ignored.That's completely ridiculous - the theory of all games other than zero-sum games with two players is essentially a theory of what patterns of cooperation will arise. In fact, games where cooperation is not beneficial are called "inessential" and set apart as an uninteresting, trivial case.

An overwhelming majority choose to cooperate, and they found that the only groups of people who choose the 'rational' approach were economists and sociopaths. It is worth noting, too, that if everybody did choose the 'rational' option, the deal could never be made - the seller and buyer both suffer.Are you saying that there is some kind of consensus between game theorists that lying is the rational choice? Because I sincerely doubt that.

[Guerrilla22]

I think this should be in 'Research'.

However, if you abandon the idea that we all act in the way that game theorists say we do (and there is precious little evidece to suggest we do), then the theory falls apart.

It's a neat mathematical theory, but it relies on a bourgeois view of humanity: we are all social atoms, and are all ultimately selfish.


Yeah game theory is something your liberal arts profs will teach because they use it themselves. In quite a few instances the conclusions they come to could also reached just by using common sense. Although there isn't nearly as much prestige involved when you simply explain why something probaly is rather than write a paper using all sorts of equations to back up your arguments.

[black magick hustla]

game theory was pioneered by nash, a schizophrenic sociopath. Game theory was also what led to the whole mutually assured destruction policy, where both the right and left wing bourgeosies were piling enough nukes to carbonize the world.

[Dean]

Why would it be have to be an interest? Why not habit or why could the action have no particular reason? Not to mention that the theory of inner compulsions preceding actions sounds much like the old Cartesian theory of the will.

You're redefining "self-interest" to make it completely vacuous.
Regardless of whether the conclusion is true or not, I don't understand how an experiment where people are told to imagine that they are in such a dilemma would imply anything about their actual behaviour in similar situations.
No, I am not redefining "self-interest." It is very simple. Self-interest are the desires and goals that you have as a human - even if it is "habit" there is a cumpulsion for it; things don't come from nothing.


That's kinda like asking people how they would behave if they were prison guards vs. Zimbardo.
That's completely ridiculous - the theory of all games other than zero-sum games with two players is essentially a theory of what patterns of cooperation will arise. In fact, games where cooperation is not beneficial are called "inessential" and set apart as an uninteresting, trivial case.
Are you saying that there is some kind of consensus between game theorists that lying is the rational choice? Because I sincerely doubt that.
Yes, there is. It's funny how you think that you need to defend these nonsensical theories which assume a very simplistic human being or value system.

As for the element of cooperation, non-cooperation is the most applied form of the theory when it comes to real-world situations. So I think it does hold more weight, though the cooperative elements still involve external coercive factors to maintain their agreements.

Nash's game theory assumes that the only equillibrium human beings can exist within is mutual distrust and manipulation. The general assumption of game theory is that human beings act within a predefined set of rules which exist as responses to given situations. It's asinine and childish, to say the least, since it refuses to analyze the broader political existence of the subjects and a slew of other variables.

In any case, I think its really funny that, when actually confronted with a game common to the theory, you brush it off.

[MarxSchmarx]

Just a second.

Game theory is just another approach to solving an optimization problem. What precisely is it that game theory offers that other optimization approaches don't?

If none, then when we are confronted with the question: what is wrong with taking an opportunistic approach (what works, what convinces others, etc...) rather than dogmatically sticking to one approach?

In certain scenarios in our daily lives I can see a role for game theory emerging out of such an opportunistic approach, e.g., in union contract negotiations. But to expect it to apply across the board to virtually all conceivable optimization problems, much less human interactions, is mysticism.

[Hit The North]

You can find a copy of part one of Adam Curtis's excellent documentary mini-series, The Trap here: http://video.google.com/videoplay?docid=404227395387111085

In it he documents the policy repercussions of how the power elite in the USA and UK employed Game Theory. It's a fascinating, if contentious, account.

[Ratatosk]

game theory was pioneered by nash, a schizophrenic sociopath.Among others, yes - and this is relevant how?

Game theory was also what led to the whole mutually assured destruction policy, where both the right and left wing bourgeosies were piling enough nukes to carbonize the world.Are you saying that this would not have happened if game theory hadn't been invented?

No, I am not redefining "self-interest." It is very simple. Self-interest are the desires and goals that you have as a human - even if it is "habit" there is a cumpulsion for it; things don't come from nothing.First of all, a compulsion is not necessarily an interest/desire/goal, second of all, it would be retarded to say that since I just stabbed myself in the pancreas, I must have done so out of "self-interest". Only someone in the grip of an absurd theory would claim that.


Are you saying that there is some kind of consensus between game theorists that lying is the rational choice? Because I sincerely doubt that.Yes, there is.Evidence? It's patently not true for the iterated PD and I'm not sure whether it's true for the non-iterated PD.

It's funny how you think that you need to defend these nonsensical theories which assume a very simplistic human being or value system.It's funny how you think mathematical theories assume something about human psychology. I wonder, do we also need to assume a "very simplistic human being" when we apply game theory to evolutionary biology or computer science? Do logicians who deal with game semantics also make assumption about human psychology? These people - who all use game theory - certainly don't make assumptions about human beings or value systems, much less simplistic ones.

As for the element of cooperation, non-cooperation is the most applied form of the theory when it comes to real-world situations. So I think it does hold more weight, though the cooperative elements still involve external coercive factors to maintain their agreements.So? Either it's being applied correctly, in which case there is no problem, or incorrectly, in which case the problem is with the application. But it's simply false to claim that:

The problem is that "the game" is a supposedly anarchistic society where all human beings act equally against each others' interests. It is the pinnacle of paranoia, where every single member of society is at total odds with each other member. Cooperation is ignored.
Nash's game theory assumes that the only equillibrium human beings can exist within is mutual distrust and manipulation.First of all, I don't understand why Nash equilibruims would have to be based on mutual distrust and manipulation, second of all, game theory neither began nor ended with Nash.

The general assumption of game theory is that human beings act within a predefined set of rules which exist as responses to given situations.No, there is no such assumption (the transition from the extensive to the normal form is just a mathematical convenience) - or rather, the assumption is that the players are able to foresee what the possible plays would look like if they did this or that. It's fucking trivial that people actually cannot do this, but that doesn't mean it's not a useful and to some extent necessary simplification.

It's asinine and childish, to say the least, since it refuses to analyze the broader political existence of the subjects and a slew of other variables.Neither does any economical theory (afaik), so?

And of course it would be ridiculous to expect a mathematical theory to be able to deal with, for example, one's political leanings and the effect they might have on one's behaviour. Sure, trying to apply game theory to situations where one's "broader political existence" significantly influences the decisions made would be idiotic. But I don't understand why you're blaming this on game theory. It's a bit like blaming economy for not being able to explain why some people prefer to go to the cinema rather than to a concert.

In any case, I think its really funny that, when actually confronted with a game common to the theory, you brush it off.I pointed out that your argument based on an "imagination" experiment was simply not valid. I didn't deny that people often choose to cooperate - especially since there are actual experiments to demonstrate that, not just "what would you do if" experiments.

Game theory is just another approach to solving an optimization problem. What precisely is it that game theory offers that other optimization approaches don't?Game theory is not an optimization problem, because you have several competing functions to be maximized. It's conceptually something rather different.

In certain scenarios in our daily lives I can see a role for game theory emerging out of such an opportunistic approach, e.g., in union contract negotiations. But to expect it to apply across the board to virtually all conceivable optimization problems, much less human interactions, is mysticism.Yeah, except I doubt anyone expects it to apply to all conceivable "optimization problems", much less human interactions, so I don't understand who your criticism was directed at.

[Dean]

I don't have the time or the inclination to ague with you on this. The topic is interesting, but your method of debate is belligerent and offensive, and frankly I don't want to get involved with more assholes.

[Ratatosk]

I was being no more offensive than you. Feel free to come back whenever you feel like backing up your assertions.

Given that the only offensive word in my whole post was "retarded" (in 'it would be retarded to say that since I just stabbed myself in the pancreas, I must have done so out of "self-interest"', where it was perfectly justifiable and it wasn't even necessarily directed at you, unless of course you accept this ridiculous statement as true), my working hypothesis is that you were just looking for a cop out from having to answer my objections.

ETA: also, lol @ "your method of debate is belligerent and offensive ... I don't want to get involved with more assholes".

[Guerrilla22]

Game theory was also what led to the whole mutually assured destruction policy, where both the right and left wing bourgeosies were piling enough nukes to carbonize the world.

Yeah the US government still employs "experts" to game certain scenarios for purposes of making good policy. They had numerous people game the invasion of Iraq for example.

[Ratatosk]

Yeah the US government still employs "experts" to game certain scenarios for purposes of making good policy. They had numerous people game the invasion of Iraq for example.These war simulations had nothing to do with game theory. The only link between these simulations and game theory is the word "game".

(Not to mention that the advice these simulations gave was afaik correct.)

[MarxSchmarx]

Originally Posted by MarxSchmarx
Game theory is just another approach to solving an optimization problem. What precisely is it that game theory offers that other optimization approaches don't?
Game theory is not an optimization problem, because you have several competing functions to be maximized. It's conceptually something rather different.
I don't want to get too technical, but I think you are missing the forest for the trees.

Game theory is (often) a special case of multi-criteria optimization. The goal is to maximize benefits while minimizing costs or risk, and the only difference with certain forms of (one-dimensional) optimization, is that the objective function is a weight of metrics that frequently imply an inherent trade-off (for example, famously, the iterative versus single prisoner's dilemma). The goal of game theory is to identify the set of strategies where you can't improve in one metric of performance without sacrificing another (so-called Pareto optimal sets). More general multi-dimensional optimization does not have this trade-off between potential metrics, but in any event game theory is a special case of such optimization problems.



Quote:
In certain scenarios in our daily lives I can see a role for game theory emerging out of such an opportunistic approach, e.g., in union contract negotiations. But to expect it to apply across the board to virtually all conceivable optimization problems, much less human interactions, is mysticism.
Yeah, except I doubt anyone expects it to apply to all conceivable "optimization problems", much less human interactions, so I don't understand who your criticism was directed at. At the risk of repetition, game theory offers very, very little in normative prescriptions. It has some, very limited potential, as a descriptive account in some fields, but otherwise has proven to be of limited value. The criticism is therefore directed against those who assert that game theory is more than an analytically tractable brain puzzle.

[Die Neue Zeit]

^^^ Why were you just talking to yourself? ;)

[Ratatosk]

I don't want to get too technical, but I think you are missing the forest for the trees.

Game theory is (often) a special case of multi-criteria optimization. The goal is to maximize benefits while minimizing costs or risk, and the only difference with certain forms of (one-dimensional) optimization, is that the objective function is a weight of metrics that frequently imply an inherent trade-off (for example, famously, the iterative versus single prisoner's dilemma). The goal of game theory is to identify the set of strategies where you can't improve in one metric of performance without sacrificing another (so-called Pareto optimal sets). More general multi-dimensional optimization does not have this trade-off between potential metrics, but in any event game theory is a special case of such optimization problems.Well, I partly concede the first point and disagree about the second. I thought optimization theory included only the maximization of an objective function subject to some constraints (which is clearly not what's happening in game theory), but apparently the label is applied to areas like game theory and control theory, too. In that case, however, I don't understand your complaint - game theory applies to a particular class of optimization problems, just like linear programming or control theory. So?

What's definitely not true, though, is the claim that the "goal of game theory" is to find Pareto optimal sets. That's simply false - Nash equilibria are not guaranteed to be Pareto optimal, neither are their various modifications. There are lots of other concepts of solution than Pareto optimality - which is the trivial starting point of game theory, rather than its goal.

At the risk of repetition, game theory offers very, very little in normative prescriptions. It has some, very limited potential, as a descriptive account in some fields, but otherwise has proven to be of limited value. The criticism is therefore directed against those who assert that game theory is more than an analytically tractable brain puzzle.Well, let's distinguish two things here - defending game theory from mistaken criticisms of game theory based on popular misconceptions (such as that it's based on the belief that everyone's an evil, manipulating social atom or that game theorists believe it to be applicable to all human behaviour or that it disregards cooperation) and defending game theory as such. So far, I've tried to do the former. I'm not particularly qualified to do the latter, but there definitely are valid areas of application and it's of course an interesting theory to study on its own. If someone is dogmatically trying to apply game theory to areas where it's not applicable, let's point it out, instead of attacking the theory itself. Looking at the last few issues of the International Journal of Game Theory and other similar publications, there doesn't appear to be any shortage of applications.

ETA: one more thing: humans clearly don't always act rationally, but surely that's not an argument against studying rational behaviour? A valid argument against game theory would be that it misconceives what constitutes rational behaviour (and people have argued that, and it resulted in various modifications of game theory), not the trivial platitude that humans don't always act rationally. That would be like arguing that we don't need to study psychological disorders, because not everyone is insane.

[MarxSchmarx]

^^^ Why were you just talking to yourself? http://www.revleft.com/vb/../revleft/smilies/wink.gif

whoopsie! thanx and fixed.


I thought optimization theory included only the maximization of an objective function subject to some constraints In the broadest sense of the term "constraints", yes.


I don't understand your complaint - game theory applies to a particular class of optimization problems, just like linear programming or control theory. So?...Looking at the last few issues of the International Journal of Game Theory and other similar publications, there doesn't appear to be any shortage of applications.
I was laying the groundwork before going a step further. That, moreover, there is no domain of concrete application to social praxis where game theory, and game theory alone, applies. Although I am not familiar with the literature you mention, as far as I know the uniqueness of game theory's applications is not (understandably) well known. I suspect game theory's prescriptions are not unique, but that's just a hunch. As such game theory is largely an intellectual curiosity and a useful heuristic starting point, nothing more nothing less. I do think we can design uniquely useful policies in systems engineering and applied operations research out of, for example, linear programming, genetic algorithms and other optimization schemes you mention. I'm yet to be convinced this is the case for game theory, and I feel this is precisely because much of the discoveries game theorists make are essentially just more optimization methods.

The only real case I've heard where game theory does directly apply is to internet traffic management, but I don't know if there is another formulation of this problem.



What's definitely not true, though, is the claim that the "goal of game theory" is to find Pareto optimal sets. That's simply false - Nash equilibria are not guaranteed to be Pareto optimal, neither are their various modifications. There are lots of other concepts of solution than Pareto optimality - which is the trivial starting point of game theory, rather than its goal.
Perhaps the problem is one of semantics, in particular, our understanding of "the goal" of game theory, and, secondly, of how the Pareto set is construed. Of course you are correct; I chose my words poorly so I'll try again. In the applications, rather than the study, of game theory, the goal should be to identify games that equilibriate around a pareto-set (broadly defined), and, failing that, to explain why the equilibria are suboptimal and identify corrections or needed modifications to the theory. This what I meant by the "goal" of game theory.

I am unsure whether any equilibrial state can, as a generality, be locally suboptimal. An example might illustrate my misgivings. The Nash equilibrium in a two person game can fall outside of the pareto set if the metrics/objective functions are composed of the payoffs of the two players. However, this is only one definition of the pareto-set. The trivial pareto-set of one objective function described by the payoff of one player is, from this perspective, pareto-optimal (as any change in this strategy will, by definition, lead to lower payoffs). More generally, I conjecture that single-dimensional optimality, which, from my understanding, game theoretic equilibria are, are special cases of pareto-sets. To be sure, if the objective functions are composed of the payoffs of other players, then these are not pareto-optimal. I realize this is not how the phrase "pareto optimality" is used in the game theory literature so I do apologize for that.


one more thing: humans clearly don't always act rationally, but surely that's not an argument against studying rational behaviour? A valid argument against game theory would be that it misconceives what constitutes rational behaviour (and people have argued that, and it resulted in various modifications of game theory), not the trivial platitude that humans don't always act rationally. That would be like arguing that we don't need to study psychological disorders, because not everyone is insane.Agreed, it's not an argument against studying rational behavior, but it is an argument against expecting the theory to be able to provide much insight into social practices. I hate to sound like a broken record, but if game theory can only explain a tiny fraction of social phenomena in very limited and contrived examples, or by expanding the definition of "rationality" to the point where it morphs into an ambiguity, and doesn't offer much in the means of concrete human enrichment, can it justify the bundle upon bundle of libraries it has filled?

It's worth emphasizing that viz. your analogy, there are two ways we can study psychological disorders. One is the case study approach, looking at each case and describing its symptoms and potential cures, kind of like early natural history.

The other is to approach the study of psychology disorders with a coherent and unified theory of how and why psychological disorders arise, which should certainly take into account their rarity. I think such a comprehensive approach is more promising from a predictive and prescriptive standpoint, and ultimately a more fulfilling use of psychologists' time.

Similarly with optimization problems in society. An opportunistic approach that takes problems on a case-by-case basis is well and fine, and game theory seems to have found a home under this rubric. However, this doesn't address its core deficiency (which may be shared by other optimization approaches), namely, that it has all too limited a domain of application precisely because of the general inapplicability of some of its more commonly applied foundational assumptions.