Separately, and this may be like bringing coal to Newcastle, but are you familiar with the work of Cailin O'Connor and Jim Weatherall, including on curation?
They gave a presentation in January that was a helpful overview.
Happy birthday!! May you have a speedy recovery and enjoy it, even if belatedly. I very much look forward to your “implausibly long” essay, and rest assured you have left us with enough material here to keep us busy until your next birthday and beyond.
I was thinking of you today, particularly, as I wound up engaged in exchanges on a fascinating Paul Krugman post over what the Democrats (in my view) need to correct so that disaffected people would feel better represented by the party and consider voting for it again. Part of this was dropping unpopular ideas, such as men in women’s sports.
A lively and somewhat contentious conversation ensued and, while it took having all of what little I have left of wits about me to think how to respond, it was also illuminating to see where the trouble spots are. One among them was an implication that people who don’t think, eg, that men should be playing on women’s sports are simply being sucked in by the R “culture wars” attack machine. While you are in no way responsible for how I tried to engage productively, I will say that stretching my mind (particularly at the advanced age of 76, now) through yours and Michael Hannon’s essays was a great help to keep calm, carry on, and see whether I could start to build a bridge of some sort.
Game theoretic analyses are done under strong assumptions. Besides the assumption of perfect rationality, the crucial assumptions are about common knowledge. In game theory, something is common knowledge between two people, a and b, if for all n in the set of integers (1,2,3,…with no limit), it is true that (a knows that b knows that)^n a knows all elements of the game. And symmetrically for b. The elements of the game include every possible belief and all possible payoff-relevant private information that every other player could have, and the set of possible [(beliefs about beliefs about)^n, n= m1,2,3,..with no limit] about these beliefs and private information (these define a player’s type), and subjective priors over the high order of infinity of possible types. Of course, the information each player has at each point they make decisions must be common knowledge as are the decisions available to them.
WHY would I bring this up? Because, there is no topology on games, very small changes in the game can lead to very large and discontinuous changes in outcomes. When there is communication in games, we must assume that the players have a language with precise meanings that are common knowledge.
Game theory can illuminate some situations, but reliance on it leads to very sterile intellectual ground.
Agreed. I find that game theory is useful for creating interesting games, but poor for understanding the situations humans typically find themselves in as real “games” have vastly more layers of complexity and hidden information than any analytical tractability can handle.
Partially agreed. I find game theory useful to distinguish what types of situation offer virtually sustainable cooperation and to understand incentives for mixed strategy. Most situations are either one shot or infinitely repeated games. Maybe a reconciliation is that we underestimate the number of situations that are in fact Bayesian games, rather than perfect information games. On the other hand, we overinflate the number of situations where solving with Bayesian models of a game (i.e., forming beliefs) is optimal.
I disagree with “Most situations are either one shot or infinitely repeated games. “ I would instead say that most situations are single shot or repeated games nested inside other games, themselves linked to other games and embedded in still more, to the point we can’t even see them all. Game theory struggles with context and these massively linked games, which humans know as “life”.
I agree with that, we play nested games. We could also argue that no game in the lab is fully a non-nested game, hence all games are nested, which would ultimately make the concept of nested games useless unless we define a gradation of nestedness.
Anyway, no model is perfect, etc.
Some games are too complex to be solved. The mega-meta-game of life and any meta-game are of those categories, plus we have to make assumptions of rationality.
I'm a neuroscientist/cognitivist so I don't give too much credit to game theory in solving real-life problems. I still find that incentives pressure some situations enough so that individuals play rationally, making game theory a useful lense, despite being partial. Or game theory can inform you when game theory is clueless.
Agreed yea. I just remember the heyday of game theory when it was going to solve human interactions l, and now what, 60 years on it is not so much. People so often forget that not everything is a prisoner’s dilemma, and often prisoner’s dilemma’s aren’t even that. Flattening things down to game theory models tends to obscure more than it enlightens, but also tends to make people think they know what is going on when they don’t.
I hope you feel better soon!
Separately, and this may be like bringing coal to Newcastle, but are you familiar with the work of Cailin O'Connor and Jim Weatherall, including on curation?
They gave a presentation in January that was a helpful overview.
https://youtu.be/2EfS5pQY7Aw?si=btCUjZ9RICjZ8WHi
Happy birthday!! May you have a speedy recovery and enjoy it, even if belatedly. I very much look forward to your “implausibly long” essay, and rest assured you have left us with enough material here to keep us busy until your next birthday and beyond.
I was thinking of you today, particularly, as I wound up engaged in exchanges on a fascinating Paul Krugman post over what the Democrats (in my view) need to correct so that disaffected people would feel better represented by the party and consider voting for it again. Part of this was dropping unpopular ideas, such as men in women’s sports.
A lively and somewhat contentious conversation ensued and, while it took having all of what little I have left of wits about me to think how to respond, it was also illuminating to see where the trouble spots are. One among them was an implication that people who don’t think, eg, that men should be playing on women’s sports are simply being sucked in by the R “culture wars” attack machine. While you are in no way responsible for how I tried to engage productively, I will say that stretching my mind (particularly at the advanced age of 76, now) through yours and Michael Hannon’s essays was a great help to keep calm, carry on, and see whether I could start to build a bridge of some sort.
Happy Birthday! Hopefully you will recover quickly. All the best, John.
Happy Birthday
Happy birthday, Dan! And I hope you feel better soon.
Happy birthday Dan! 🥳🥳
Dan,
Game theoretic analyses are done under strong assumptions. Besides the assumption of perfect rationality, the crucial assumptions are about common knowledge. In game theory, something is common knowledge between two people, a and b, if for all n in the set of integers (1,2,3,…with no limit), it is true that (a knows that b knows that)^n a knows all elements of the game. And symmetrically for b. The elements of the game include every possible belief and all possible payoff-relevant private information that every other player could have, and the set of possible [(beliefs about beliefs about)^n, n= m1,2,3,..with no limit] about these beliefs and private information (these define a player’s type), and subjective priors over the high order of infinity of possible types. Of course, the information each player has at each point they make decisions must be common knowledge as are the decisions available to them.
WHY would I bring this up? Because, there is no topology on games, very small changes in the game can lead to very large and discontinuous changes in outcomes. When there is communication in games, we must assume that the players have a language with precise meanings that are common knowledge.
Game theory can illuminate some situations, but reliance on it leads to very sterile intellectual ground.
Agreed. I find that game theory is useful for creating interesting games, but poor for understanding the situations humans typically find themselves in as real “games” have vastly more layers of complexity and hidden information than any analytical tractability can handle.
Partially agreed. I find game theory useful to distinguish what types of situation offer virtually sustainable cooperation and to understand incentives for mixed strategy. Most situations are either one shot or infinitely repeated games. Maybe a reconciliation is that we underestimate the number of situations that are in fact Bayesian games, rather than perfect information games. On the other hand, we overinflate the number of situations where solving with Bayesian models of a game (i.e., forming beliefs) is optimal.
That being said, happy bday to Dan.
I disagree with “Most situations are either one shot or infinitely repeated games. “ I would instead say that most situations are single shot or repeated games nested inside other games, themselves linked to other games and embedded in still more, to the point we can’t even see them all. Game theory struggles with context and these massively linked games, which humans know as “life”.
I agree with that, we play nested games. We could also argue that no game in the lab is fully a non-nested game, hence all games are nested, which would ultimately make the concept of nested games useless unless we define a gradation of nestedness.
Anyway, no model is perfect, etc.
Some games are too complex to be solved. The mega-meta-game of life and any meta-game are of those categories, plus we have to make assumptions of rationality.
I'm a neuroscientist/cognitivist so I don't give too much credit to game theory in solving real-life problems. I still find that incentives pressure some situations enough so that individuals play rationally, making game theory a useful lense, despite being partial. Or game theory can inform you when game theory is clueless.
Agreed yea. I just remember the heyday of game theory when it was going to solve human interactions l, and now what, 60 years on it is not so much. People so often forget that not everything is a prisoner’s dilemma, and often prisoner’s dilemma’s aren’t even that. Flattening things down to game theory models tends to obscure more than it enlightens, but also tends to make people think they know what is going on when they don’t.
True!
Happy birthday! If you were born March 2 1993 then you will hit your billionth second on November 8th this year. A milestone worth celebrating.