|
ViggyNash posted:There was a once programming competition based around the Prisoner's Dilema. Do you sell out, or trust the other player? (I can't source it, I heard it on a Radiolab episode though) Did someone say sources? There were two tournaments by the first person to do this; the first one produced the brutally effective "tit for tat" strategy that basically owned everyone else present despite being the smallest program there by a few orders of magnitude, and the second was held specifically to challenge an entire generation of programmers, strategists and ethicists to produce a strategy that could beat tit-for-tat (they couldn't). They concluded that the four things you needed to be effective were to assume good faith, retaliate against betrayal, be willing to forgive, and to act for the general good of all over the good of yourself. The tournaments have been repeated endlessly since. There still isn't a quantifiably better strategy. I don't know how much of such reasoning can apply to RPS though; RPS is like 100% metagame. I'm gonna be delighted to find out, though.
|
#
¿
Feb 12, 2017 15:12
|
|
|
|
| # ¿ Apr 27, 2024 20:08 |
|
|
I don't know what game I expected to usher in the next golden age of LP but
|
|
#
¿
Feb 16, 2017 17:36
|
|
|
The big difference between Rock Paper Scissors and games like Chicken, the Prisoner's Dilemma, the Stag Hunt, Rendezvous, etc, is that RPS does not have a pure strategy Nash Equilibrium, because there is never going to be a situation in which neither player stands to gain by being the only one to change strategies (this is to be expected, really; RPS is both zero-sum and non-cooperative). If you allow mixed strategies, then, well, obviously the equilibrium is going to end up just being choosing entirely randomly between the three, since any weighting towards one choice produces an incentive for your opponent to weight that choice's counter. This is why this tournament has banned the use of random number generators - because it removes the only possible equilibrium state and turns RPS into an exercise in pure metagaming; predicting, or attempting to predict, what your opponent is going to do based entirely on data and knowledge of them (and, of course, appraising of their knowledge of you). A Nash Equilibrium is a state proven (by the reasonably famous economist of the same name) to exist in any iterative finite-strategy game. In an equilibrium state, all players know that no player has anything to gain from being the only one to change their strategy. In most cooperative games, this state is when both players choose complementary strategies; if either suddenly stops, everyone loses. In the Prisoner's Dilemma, it's when both players betray eachother (and not when they both stay silent), because switching from betrayal to alliance only pays off if your opponent does the same. In Chicken, it's literally any state other than the car crash. Etc etc.
|
|
#
¿
Feb 18, 2017 19:18
|
|
|
Man, crow was really holding those guys back.
|
|
#
¿
Feb 27, 2017 13:42
|
|
|
|
|
#
¿
Feb 28, 2017 15:50
|
|
|
The teams are supposed to not read this thread, right?
|
|
#
¿
Mar 1, 2017 14:01
|
|
|
If option C prevails I will add an extra wrinkle to proceedings. It will be fun.
|
|
#
¿
Mar 1, 2017 14:23
|
|
|
Allow me to wrinkle the proceedings further. Crow's transgression is going to punished. This is not up for question. I don't know quite how I feel about the matter, however. This was clearly a violation of the spirit of the game, and yet it has been a source of entertainment and drama in these dark times. How slighted are we, really? I will therefore determine the severity of Crow's punishment according to the depth of the hatred across both teams. There are twenty-one non-treacherous players across both teams, and I will count their betray votes. 0) 6 hours. 1-3) 12 hours. 4-12) 1 day. 13-18) 3 days 19-21) 1 week Fate will decide. E: the teams will not be told any of this.
|
|
#
¿
Mar 1, 2017 15:46
|
|
|
Per the extra rule it seems that the best way to spite Crow would be to all betray, so as to make him triple-lose instead of singular lose.
|
|
#
¿
Mar 1, 2017 15:56
|
|
|
|
| # ¿ Apr 27, 2024 20:08 |
|
|
Honestly I think there could be a good case for doing a classical game-theory-game megathread with continual game theory asides. Go all into the maths behind it and let people play nifty decision games.
|
|
#
¿
Mar 5, 2017 01:11
|
|