3 ms·

> (Saying that, I am now a cryptographer and don't use much graph theory, or any of the economics and game theory I did in my undergradate :-) Game theory in pa

by dsacco 9y ago

> (Saying that, I am now a cryptographer and don't use much graph theory, or any of the economics and game theory I did in my undergradate :-) Game theory in particular, I am highly skeptical of, but that's another story...) You’re a cryptographer and don’t use any game theory? Most cryptographic protocols are, or can be, phrased through game semantics. Game theory is very useful, what are you skeptical about?

Moodles 9y ago
It is true that you can analyse many cryptographic schemes by modelling them as games. In fact, my PhD was largely focused on this exact topic :-) Our discussion really depends on what we define as “game theory” and “graph theory” I suppose. The game theory I studied in economics (albeit in my undergrad) I'm very skeptical of, though. It was about exact payoffs and rational agents and loads of assumptions that really don't work in real-life. I'd spend time literally analysing optimal strategies in board games. It all seemed kind of pointless. People always said you can apply game theory to political science and wars, but it all seemed like common sense and logic, rather than the specific stuff I was actually studying. I remember solving integral equations to calculate Nash equilibrium bidding strategies in third price bid auctions with perfect information and a uniform distribution of value to buyers. The solutions had a lot of maths, but had no bearing on reality whatsoever. What I studied was never really used to make any actual predictions in the real world. It seemed almost as if they were trying to make this subject mathematical to give it some legitimacy. A lot of it is common sense. One could argue that studying it helps you think logically I guess, but so do lots of things. Ariel Rubinstein is a famous economist and actually one of the founders of game theory and gave an excellent critique of the whole space. He is a lot more eloquent than I. You should listen to him and let me know what you think. https://rwer.wordpress.com/2012/11/05/game-theory-a-critique/ https://rwer.wordpress.com/2012/11/05/game-theory-a-critique...
- tw1010 9y ago
  Isn't the (only) problem with game theory that the actor optimization function is inappropriate? Rationality as an assumption is the problem. But everything else, all the derived higher order structure above that seems, in my eyes, to be pretty useful. There are other contexts, like evolutionary game theory, where the optimization function is just darwinian fitness. And in that situation do most professionals seem to agree that the model is perfectly useful. So game theory as a framework seems would be totally fine if we could only figure out a more psycologically plausible goal function, wouldn't you agree?
  - Moodles 9y ago
    Rubinstein makes the point (which I wholeheartedly agree with) that game theory is a lot like fables. They might help you interpret a certain situation in a certain way, but they have no value in actually predicting anything. I'm not sure how it will ever get there. If someone comes up with a goal function and actually makes predictions which work in the real world then yes, I would of course change my mind.
    - tw1010 9y ago
      I used to be just as sceptical as you. (And I still am about traditional game theory.) But then I stumbled into evolutionary game theory, which seems much more robust. Have you read anything about it? What's your take? Apparently ESS has been used to model cancer very fruitfully.
- dsacco 9y ago
  Interesting, I’ll read more. Thanks for the perspective.

by dsacco 9y ago

Moodles 9y ago

It is true that you can analyse many cryptographic schemes by modelling them as games. In fact, my PhD was largely focused on this exact topic :-) Our discussion really depends on what we define as “game theory” and “graph theory” I suppose. The game theory I studied in economics (albeit in my undergrad) I'm very skeptical of, though. It was about exact payoffs and rational agents and loads of assumptions that really don't work in real-life. I'd spend time literally analysing optimal strategies in board games. It all seemed kind of pointless. People always said you can apply game theory to political science and wars, but it all seemed like common sense and logic, rather than the specific stuff I was actually studying. I remember solving integral equations to calculate Nash equilibrium bidding strategies in third price bid auctions with perfect information and a uniform distribution of value to buyers. The solutions had a lot of maths, but had no bearing on reality whatsoever. What I studied was never really used to make any actual predictions in the real world. It seemed almost as if they were trying to make this subject mathematical to give it some legitimacy. A lot of it is common sense. One could argue that studying it helps you think logically I guess, but so do lots of things. Ariel Rubinstein is a famous economist and actually one of the founders of game theory and gave an excellent critique of the whole space. He is a lot more eloquent than I. You should listen to him and let me know what you think. https://rwer.wordpress.com/2012/11/05/game-theory-a-critique/ https://rwer.wordpress.com/2012/11/05/game-theory-a-critique...

tw1010 9y ago
Isn't the (only) problem with game theory that the actor optimization function is inappropriate? Rationality as an assumption is the problem. But everything else, all the derived higher order structure above that seems, in my eyes, to be pretty useful. There are other contexts, like evolutionary game theory, where the optimization function is just darwinian fitness. And in that situation do most professionals seem to agree that the model is perfectly useful. So game theory as a framework seems would be totally fine if we could only figure out a more psycologically plausible goal function, wouldn't you agree?
- Moodles 9y ago
  Rubinstein makes the point (which I wholeheartedly agree with) that game theory is a lot like fables. They might help you interpret a certain situation in a certain way, but they have no value in actually predicting anything. I'm not sure how it will ever get there. If someone comes up with a goal function and actually makes predictions which work in the real world then yes, I would of course change my mind.
  - tw1010 9y ago
    I used to be just as sceptical as you. (And I still am about traditional game theory.) But then I stumbled into evolutionary game theory, which seems much more robust. Have you read anything about it? What's your take? Apparently ESS has been used to model cancer very fruitfully.
dsacco 9y ago
Interesting, I’ll read more. Thanks for the perspective.