The problem with Iterated Golden Balls (with elimination)

The Wizard is hosting what I believe is the first iterated game of Golden Balls ever played in the public domain. This is a very interesting sociological experiment, in the vein of iterated Prisoner's Dilemma. Some very strong and informative strategies have been developed to play iterated Prisoner's Dilemma, shining a light on some aspects of cooperative human behaviour.

I worry the same will not be said for this version of iterated Golden Balls.

Golden Balls is close enough to prisoner's dilemma that the socially repugnant choice (steal) dominates the socially beneficial one. Each player has an obvious EV incentive to 'Steal' every round: they either do no worse, or get 2x the money.

Facing elimination each round creates a strong incentive to try and get the other player to say Split, but you have no way to coerce this action. The Nash Equilibrium all end up with at least one eliminated player.

The problem with iterated Golden Balls with elimination is that once a player steals and moves to the next round, their intent to steal at least some of the time is revealed. In Prisoner's Dilemma, you can punish the player by stealing back the next game, and then return to cooperating after some time. With elimination, the punishment that is key to encouraging cooperative strategies results in the elimination of at least one player. Probably both.

And so, at every round, facing a player (or team) that has stolen at least once, you are reminded that your weakly dominant strategy is to steal, and that, as some have pointed out in the thread, you have a social responsibility to punish/steal. The team that stole in the previous round is not ignorant of this. Therefore, I predict almost inevitable failure of the group. In other words, the Wiz's money is safe.

There is a simple fix, which I suspect the Wizard has already considered for game 2. Play repeated games against the same or different opponents, splitting or stealing 1 point, and give the prize money to the person with the most points at the end of X rounds. But now, one has to wonder if this will be different enough from iterated Prisoner's Dilemma to be considered a breakthrough sociological experiment.

One thing occurred to me. Play it as an iterated game, but each point is a share of pool, rather than out-right elimination. 1 for a share, 2 for a successful steal, 0 and 0 for any other result. The key twist for GoldenBalls is that Mutual Steal and Stolen from is the same result.

I thought I was having fun in a game. I didn't realize I was a guinea pig in a social experiment!
Time to push the button, I want treats.

Quote: RaleighCraps

I thought I was having fun in a game. I didn't realize I was a guinea pig in a social experiment!
Time to push the button, I want treats.

Odd.. each time I push the button, I get shocked. What gives?

Quote: thecesspit

One thing occurred to me. Play it as an iterated game, but each point is a share of pool, rather than out-right elimination. 1 for a share, 2 for a successful steal, 0 and 0 for any other result. The key twist for GoldenBalls is that Mutual Steal and Stolen from is the same result.

I thought of this too, however the Wiz would never agree to it since I think the whole point of his experiment was to show that it was unlikely that everyone would SPLIT enough and make it to the final round. Why else would he want to give away $200 of his own money?

Quote: Ayecarumba

Odd.. each time I push the button, I get shocked. What gives?

Did it occur to you that my treat is watching you get shocked?