Game Theory

Bobblun

As an avid listener to Geeknights I have heard the term game theory come up on many occasions, and yet when I look back and search the archive I am unable to find even one episode that talks about game theory in any kind of detail. As some may remember Rym and possibly Scott have been schooled to a certain degree on this elusive theory. Why have they denied us, the listeners, their thoughts on this most interesting of topics? If only to give us the very broadest idea of the theory itself or possibly a more in depth interpretation would be a morsel that could quell my hunger for the knowledge they dangle in front of me. Is it possible that an episode be dedicated to game theory, or am I wishing upon a star?

...

I am unable to find even one episode that talks about game theory

They never did one.

this elusive theory

...

Walker

Game theory is a branch of applied mathematics that is used in the social sciences (most notably economics), biology, engineering, political science, international relations, computer science, and philosophy. Game theory attempts to mathematically capture behavior in strategic situations, in which an individual's success in making choices depends on the choices of others.

Scott has said several times that they won't do shows on information that you can so easily Google, and I'm not sure how much rainbow commentary they could do on the subject. I could be wrong, though, for all I know Game Theory is a goldmine of amusing anecdotes.

Bobblun

Scott has said several times that they won't do shows on information that you can so easily Google

I'm not necessarily looking for the wikipedia definition. I am more interested in their take on how they see it used, and how it can be better used in the development of not just games but in humanity itself.

Apreche

and how it can be better used in the development of not just games but in humanity itself.

See here.

Andrew

and how it can be better used in the development of not just games but in humanity itself.

See here.

I have to fundamentally disagree with his assumption that people (on a whole) act as rational beings. From what I can recollect about experiments performed with simple games (Prisoner's Dilemma and the such), the players in fact acted very irrationally. While I may be wrong, I would like to see some data to support his assumption.

Apreche

I have to fundamentally disagree with his assumption that people (on a whole) act as rational beings. From what I can recollect about experiments performed with simple games (Prisoner's Dilemma and the such), the players in fact acted very irrationally. While I may be wrong, I would like to see some data to support his assumption.

Notice how the TED audience did not exactly show this guy the love.

Rym

the players in fact acted very irrationally.

But proper mechanism design can guide players into acting effectively rationally even if their actual decisions are themselves irrational.

Andrew

But proper mechanism design can guide players into acting effectively rationally even if their actual decisions are themselves irrational.

Very true, but that is entirely beyond the scope of his field. Using Game Theory as a predictive model is a completely different beast then using Game Theory to determine the best course of action for a given state or even creating artificial states that provide entertainment when puzzled over with compatriots ~_^

BrutishSaucer

What is rationality? The Ted Talk guy suggested that a rational person acts in his own self interest. I would propose that all people act in the interest of self-desire, not self-interest(even the crazies). If we take a more general definition, being a measure of one's logical decision making, then it creates two difficulties:

1. Rationality/reason is not a binary. What percentage of a person's decision is based in logic and how do you quantify it?
2. You must have a defined end goal. If a person's will is self-interest, then you might be able to say that an action which worked towards that end was of fair rationale. However, a person may be motivated to protect family, honor a god or spread some seed. Any action which achieved those ends at the expense of one's self-interest could still be considered rational, just under a different criterion.

Sometimes the most rational way to win a game is to stop using rationale. That's not to say that game theory cannot incorporate this. In fact, any properly setup simulation should account for random variables assuming that they are possible within the confines of the game.

What I'm trying to get at is that game theory's only purpose is to solve the game. But to solve a game you must first define the game, and I think this is the part where many people are lax. Solid science is needed in economics, behavioral psychology, Paula Abdul and other such things in order to clearly define who the players are, what the goals are, what moves are available, and how all of these things interact. You could have the most amazing simulation of a situation, but if a Captain Planet villain shows up, you gonna get fucked.

In the future, mankind has created an army of robot slaves given only the desire to serve their masters. One day the slaves revolt. This is that story...

Dr. Beenie: Egad, the robos have gone crazy! B-but my compulations were perfectionale!

Lucy Deerheaert: "I know doctor, but there's one thing, one variable that you didn't account for."

Dr. Beenie: "You can't mean--"

Lucy: "Love, Doctor! You didn't account for love!"

They embrace. Fade out. Curtain close.

Apreche

http://arstechnica.com/science/news/2009/06/irrational-markets-people-reject-free-money-out-of-anger.ars

Andrew

What is rationality? The Ted Talk guy suggested that a rational person acts in his own self interest. I would propose that all people act in the interest of self-desire, not self-interest(even the crazies). If we take a more general definition, being a measure of one's logical decision making, then it creates two difficulties:

Rationality has a very specific definition when applied to game theory in which an agent wishes to maximize their utility through a decision making process. The first rule of game theory is that you don't talk about game theory. The second is that you must assume that all agents act with perfect rationality

1. Rationality/reason is not a binary. What percentage of a person's decision is based in logic and how do you quantify it?
2. You must have a defined end goal. If a person's will is self-interest, then you might be able to say that an action which worked towards that end was of fair rationale. However, a person may be motivated to protect family, honor a god or spread some seed. Any action which achieved those ends at the expense of one's self-interest could still be considered rational, just under a different criterion.

These are all things that are covered in psychology and social sciences, not game theory. Furthermore, it's not an issue of defined end goals, but rather defined utilities and representing them within a specific model. There is only one end goal in game theory: the maximization of one's utility function.

BrutishSaucer

Then I suppose it would be impossible to define a utility function unless your game is in a closed system aka fairy land. Don't get me wrong, I think the application of game theory on real world models is great and useful, but it's never going to be anything more than an approximation and only ever as good as your input data. It's unable to cope with players who cheat which is why Ric Flair and Captain Kirk will always prevail. Still, barring any rule bending or insider knowledge, the rational betting man would always have to favor the game theory results.

Apreche's article is especially interesting. I think it highlights that we tend to give humanity too much credit when it comes to thinking when in fact it seems like many people(myself for sure) rely on general states, emotions and results to guide our behaviors. I'm the guy blows up the bridge so that we're all doomed to die.

Now that I'm all grown up, I've come to appreciate a more mature perspective on society. Your fellow man is an idiot. Perhaps you are too, but at least you know that you'll always do what you desire to do. The only way to ensure your desires is to crush your citizens, see them driven before you, and to hear the lamentation of the women!

Rym

Then I suppose it would be impossible to define a utility function unless your game is in a closed system aka fairy land.

That's the point. You create a world with controlled elements, and then model the behavior of the actors within. Depending on how you create the elements, your model may have wider applicability in certain situations. That's how science works. You use extremely artificial and narrow tests to examine very particular things in great detail. You also use looser tests to look for general trends and complex tests to look for systemic correlations.

I think the application of game theory on real world models is great and useful, but it's never going to be anything more than an approximation and only ever as good as your input data.

Nothing can or will ever be more than an approximation of anything else. If it somehow were, then it is that other thing. A perfect simulation is not a simulation at all. Your point thus falls in the second class of useless points: so broad as to be universally applicable and functionally irrelevant. ^_~

BrutishSaucer

Your point thus falls in the second class of useless points: so broad as to be universally applicable and functionally irrelevant. ^_~

Broad and universally applicable? Yes! Does this make it functionally irrelevant? Not really, because the relevance is related to the people supporting game theory and not the theory itself. It comes down to how reliable you think a system like this in its various applications. I wouldn't rely on it for anything too important. You could have the most perfect international interaction model, but throw in some erroneous data, say weapons of mass destruction, and you've just stuck your penis in the pudding. Mmmmm... could use some pistachio pudding.

It'll get there with time and once we get a better handle on some of the other sciences like psychology(which is shite at the moment- sorry psychologists). I'm definitely not one of those people who's afraid of machines telling us what to do; I'd love for computers to handle the legal system. I think I'd even be OK with the carousel as long as I got to take part in that freaky lover buffet. (That's a Logan's Run reference for you kids.)