4 ms·

The idea is not to assume that everyone will be superrational or trustworthy at the beginning of the game. I believe it’s possible for the superrational player

The idea is not to assume that everyone will be superrational or trustworthy at the beginning of the game. I believe it’s possible for the superrational players in a game to become the majority by expressing the superrational strategy in terms that will appeal to rationally self-interested players and demonstrating a superrational strategy working at some non-trivial scale.

ColanR 7y ago
Even so, that idea merely expresses the hope that the prisoners can be educated out of their mistrust - which still seems like saying, 'the dillemma doesn't exist, if we fix the prisoners'. (for the sake of argument, it looks like we're both assuming a n>2 group of prisoners.) Also, the point of the dillemma is that only one prisoner has to act selfishly. If a subset of the group does cooperate, it doesn't matter because the one left out will take advantage of them all...and that risk increases exponentially as the size of the group increases, and makes the decision to trust even a subset of the group irrational.
- notduncansmith 7y ago
  In the prisoners’ dilemma, it is true that even a single bad actor has the ability to disrupt the rest. One limitation of that thought experiment is that it assumes defection is guaranteed to achieve a certain outcome. I think it’s reasonable to assume that any cooperative strategy has some computable “defection tolerance” and a variant on the prisoners’ dilemma that accounts for this could have very different outcomes than the classical formulation. Furthermore, once you start playing multiple times with the same players, you establish an algebra of trust that also drastically changes the game. The ability to influence other players between rounds also changes the game.
  - ColanR 7y ago
    > One limitation of that thought experiment is that it assumes defection is guaranteed to achieve a certain outcome. That's a premise, and it's what creates the dillema in the first place. > Furthermore, once you start playing multiple times with the same players, you establish an algebra of trust that also drastically changes the game. The ability to influence other players between rounds also changes the game. Actually, the wikipedia article on the dilemma discusses this. The nash equilibrium is actually for all players to defect. There is no engendered trust. [1] [1] https://en.wikipedia.org/wiki/Prisoner%27s_dilemma#The_iterated_prisoner's_dilemma https://en.wikipedia.org/wiki/Prisoner%27s_dilemma#The_itera...
    - notduncansmith 7y ago
      > The nash equilibrium is actually for all players to defect. Only in games where players know the number of rounds (e.g. a suicide bomber defects because they’ve decided to cap the number of rounds they’re going to play). The same Wikipedia article describes that in studies where the number of rounds is unknown to players, superrational strategies are sustainable and that the “always defect” strategy becomes only one of several paths to Nash equilibrium.

ColanR 7y ago

Even so, that idea merely expresses the hope that the prisoners can be educated out of their mistrust - which still seems like saying, 'the dillemma doesn't exist, if we fix the prisoners'. (for the sake of argument, it looks like we're both assuming a n>2 group of prisoners.) Also, the point of the dillemma is that only one prisoner has to act selfishly. If a subset of the group does cooperate, it doesn't matter because the one left out will take advantage of them all...and that risk increases exponentially as the size of the group increases, and makes the decision to trust even a subset of the group irrational.

notduncansmith 7y ago
In the prisoners’ dilemma, it is true that even a single bad actor has the ability to disrupt the rest. One limitation of that thought experiment is that it assumes defection is guaranteed to achieve a certain outcome. I think it’s reasonable to assume that any cooperative strategy has some computable “defection tolerance” and a variant on the prisoners’ dilemma that accounts for this could have very different outcomes than the classical formulation. Furthermore, once you start playing multiple times with the same players, you establish an algebra of trust that also drastically changes the game. The ability to influence other players between rounds also changes the game.
- ColanR 7y ago
  > One limitation of that thought experiment is that it assumes defection is guaranteed to achieve a certain outcome. That's a premise, and it's what creates the dillema in the first place. > Furthermore, once you start playing multiple times with the same players, you establish an algebra of trust that also drastically changes the game. The ability to influence other players between rounds also changes the game. Actually, the wikipedia article on the dilemma discusses this. The nash equilibrium is actually for all players to defect. There is no engendered trust. [1] [1] https://en.wikipedia.org/wiki/Prisoner%27s_dilemma#The_iterated_prisoner's_dilemma https://en.wikipedia.org/wiki/Prisoner%27s_dilemma#The_itera...
  - notduncansmith 7y ago
    > The nash equilibrium is actually for all players to defect. Only in games where players know the number of rounds (e.g. a suicide bomber defects because they’ve decided to cap the number of rounds they’re going to play). The same Wikipedia article describes that in studies where the number of rounds is unknown to players, superrational strategies are sustainable and that the “always defect” strategy becomes only one of several paths to Nash equilibrium.