Game Theory 101 (#62): Repeated Games and the Prediction Problem

[@Gomer._.]

It basically caught up with me on this episode because I've been watching this for the philosophy and absolutely don't understand the math, and it's finally really caught up to me where I can't conceptualize why anything would be ok here and for what reason

[@Gomer._.]

But I'm enjoying this so much I'm going to become a math matitican

[@PunmasterSTP]

Repeated games? More like “This helps make big brains!” Thanks again for all these wonderful videos.

[@jvgama]

At least it invalidates some values for the discount factor: if there is some cooperation, the discount may not be above a certain value... Not nearly what one would hope for, however... :( And in theory you have not exausted the possibilities for the next US president (Paul Ryan, any independent like Bloomberg, any third party candidate like Jill Stein, etc.), ehehe.

[@Gametheory101]

+jvgama Yeah, that's basically what I teach my students. The cutpoint for grim trigger gives the minimum threshold of the discount factor to generate outcomes that aren't the reversion outcome (i.e., mutual defection). So if you can use comparative statics to find what changes make that minimum go up or down, then you have some empirical leverage. But this is a lot weaker (and a lot noisier if you try to take to data) than a clean comparative static that says "x is increasing in y."

[@01Versatran]

Seems like there's something wrong with the concept of subgame perfect equilibrium if it allows all these weird equilibria. Nobody would actually do these things.

[@danielklein5560]

I think thats were beliefs come in. Or maybe giving preference to certain SPE over other and assuming players arent indifferent between them.