Re: game theory

["Devine, James" <jdevine@xxxxxxx>]

I wrote:
> this is an excellent statement of the game-theoretic way of thinking,
> seen in its starkest way in the kind of paranoia that characterized
> John Nash. It also points to the often-unnoted psychic costs of
> thinking that way.

Ted Winslow writes:
> The delusional aspect concerns a great deal more than paranoid
delusions about the intentions of others e.g. the conception of self
and others as calculating machines, the complete inability to take
account of and understand cultural distinctiveness, etc., etc.<

I wouldn't say that game theory itself is necessarily paranoid. Nor does it necessarily involve conceiving people as calculating machines, totally abstracting from cultural distinctiveness. (Due to lack of time, I won't comment on the "etc." or the other "etc.") Just as in mainstream economics, there are differences of opinion among game theory practitioners about what game theory is and how it should be used. (I rely on Hargreaves-Heap and  
Varoufakis, _
Game theory: a critical introduction_ (Routledge, 1995), David Kreps, _Game Theory and Economic Modeling_,  and William Poundstone's _Prisoner's Dilemma_). 	
	

I'm not an expert on game theory (and I've never even played one on TV). But I think that the bad stuff that Ted associates with game theory might best be associated with John Nash, John Von Neumann, and the Cold War RAND culture that decided that GT was a cool tool. I've never found game theory to be very useful in my research; nor does it seem very harmful. A lot of it seems like an academic game. My feeling is that its main harm comes when people reify it and use it as an ideological weapon, as some of the RANDites did. I'd blame this dark side of the GT force much more on the Cold War than on GT itself. And I blame the Cold War on... but I digress. 

I would agree with Ted that we should reject Nash equilibrium except as an abstract notion that might (in some circumstances) provide a useful contrast with reality. It's very similar to the macro (and bogus) concept of "rational expectations": people expect the economy to produce the results the model predicts it will produce and so act on these expectations. Thus, in equilibrium the model produces what they expect  (always assuming that the economy = the model). (RatEx says people's expectations work this way on average; Nash equilibrium is _defined_ as having them work exactly.) 
 
But the idea of Nash equilibrium and GT don't necessarily say that people are calculating machines. It could be interpreted as saying that in certain circumstances (in "games") people act _as if_ they were calculating machines -- or that people might be assumed to act this way as a first approximation to reality (simplifying reality in order to try to understand it). In certain oligopolistic market situations, profit-seeking firms[*] are pushed to act in this way. Similarly, the cops-and-robbers life inhabited by the fictional Tyrone Ten Eyck encourages this kind of behavior. The Cold War pushed the power elites to train and hire people who thought this way.  On the other hand, a social situation like a family or a church congregation or an anarchists' convention might be very hard to understand by assuming that people are calculating machines. Part of a practitioner's job is to figure out when the use of GT is appropriate.
 
Even if people are calculating machines (and they're not -- or at least I'm programmed to think that they're not), that doesn't mean that culture plays no role. The values to the participants of the rewards in the game matrix can and do reflect the culture that those people were brought up in. If a meat-eater and a vegetarian are each given a pork chop by the play of the game, each would assign different values to the reward. Thus, different numbers would show up in each individual's box. 
 
A major problem with GT, however, is that it (as far as I know) doesn't see culture as endogenous, something that develops from the societal "game." Playing a prisoners' dilemma game over and over again might cause one to become like Ten Eyck or a Hobbesian, valuing any marginal advantage over others, struggling to survive at any cost, even beginning to eat pork chops (and like them!) In a lot of circumstances, people in experimental prisoners' dilemmas actually learn how to cooperate with each other. What GT misses, I think, is that this learning process may actually change their utility functions -- ahem! -- I mean personalities, ethical values, and world-views. So the development of cooperation is more than a matter of learning to communicate with each other by their actions. 

Even if we assume that Nash equilibrium should rule the roost and that people all value the rewards equally and in the same way, that doesn't mean that culture plays no role. Most games don't have unique Nash equilibria. And just as with RatEx, the cool results aren't as dynamite if the equilibrium isn't unique. (_All_ of the "nice" results of RatEx come from the assumption of a unique equilibrium lurking in the background.) If the equilibrium isn't unique, culture must play a role in deciding which equilibrium results. (Yeah, it's wrong to think of society or parts of society as ever being in equilibrium. But sometimes it seems that there are major forces pushing toward some sort of equilibrium. Some institutions last for years without qualitative change. Again, the practitioner must decide when use of the equilibrium concept is appropriate. It's not omni-appropriate as the neoclassicals assume. [Reading the book by Westlake got me into all sorts of parenthetical thinking {because he (or rather, his protagonist [J. Eugene Raxford]) does it (in THE SPY IN THE OINTMENT)}] [Is this parenthetical over-kill or what?])
 
Anyway, as part of an informal democratic process, conventions might develop that people generally follow, allowing them to collectively "choose" between equilibria. Alternatively, they might be imposed from above by kings or corporations. A capitalist society limits the choices and the "games" that people find themselves in, while molding our conventions, culture, personalities, ethical values, and world-views. 

>Isn't it true that, outside of economics, the main support for the
development of game theory has come from the US military?   This
produces the more obvious Strangelove aspect, Herman Kahn etc.
Markowitz is himself a Cowles, Rand person, isn't he?  His company
seems mostly to be involved with war gaming simulations for the
military.<
 
All sorts of social scientists and philosophers use game theory. I don't know to what extent they use it for evil rather than for good. That's an empirical question. Though given the corrupt nature of capitalist academia, I'd guess that much or most of it is used for evil. But that doesn't say _a priori_ that game theory is bad. An experimental economist I know tells me that he thinks that GT is pretty useless -- except as a guide for how to structure experiments. His experiments, among other things, show that people aren't calculating machines... 
 
Jim Devine

[*] Reading the GUARDIAN today, I noted a reference to the "state-run Anatolian news agency." That phrase seems apt, since it helps the reader understand any biases that might arise due to state interests. But why not refer to the "corporate-run New York TIMES" or better yet, the "profit-driven New York TIMES." The GUARDIAN, I believe, is a not-for-profit.