Prisoner's Dilemma

What would you do in the Prisoner’s Dilemma?

  • Cooperate
  • Defect
0 voters

The thread about the “unexpected egg” reminded me of this game. Prisoner’s Dilemma is a fairly well-known game from the field of ‘Game Theory’. The rules are straightforward but their consequences are quite interesting and paradoxical. Here I’ll use the simplest version of the game, and you can check out the Wikipedia if you want more details: http://en.wikipedia.org/wiki/Prisoner’s_dilemma

In this two-player game, your aim is to maximise your points (let’s use dollars), gained from a third-party banker. The aim is not to “beat” the other player, nor to “help” him, but only to get as much money as possible for yourself. The two of you are in exactly the same situation, which is that you have to choose between two options: cooperate or defect. Neither of you knows what the other player is picking, as communication is forbidden. When both of you have decided, your selections are revealed together and these choices determine what you earn. There are four possibilities:

  1. You both cooperated - each gets $3.
  2. You both defected - each gets $1.
  3. You cooperated and the other defected - you get $0 and the other gets $5.
  4. You defected and the other cooperated - you get $5 and the other gets $0.

So what would you do - cooperate or defect? If you cooperate, then you’re hoping the other player will cooperate as well, so as to earn $3, otherwise - if he defects - you get nothing (and he gets $5). But if you think the other will cooperate, then you should defect, since that means you will earn the maximum $5. In fact, if the other player is defecting, you should still defect yourself, because that will earn you $1, which is better than $0 for cooperating. Therefore, no matter what the other player picks, you’re better off by defecting. So the logical choice is to defect. Unfortunately, the other player will think the same, and so defect as well, resulting in you getting just $1. Whereas if both of you were “illogical” and cooperated, each of you would have earned $3!

What’s your choice in this game, and can you give a reason for it? Do you think it’s really “illogical” to cooperate?

If you cooperate, there are three possible outcomes: you get either $0 or $3. So your average income is $3/2 = $1.50. If you defect, you get either $1 or $5, so your average income is $6/2 = $3. So you had better defect.

If it were a serious tactic and involving a real life situation, logic would have no room, Only honor and integrity should be used. Since it is just a game I would have to presume that I am going to be betrayed. I would get 0 dollars all the time. At the end of the game I would take 3/4 of the money from the other person for being without honor or integrity. :smiley: I win either way.

Hmm, not sure how valid this method of calculating the options is… I thought psychology would come into play somehow.

Hehe that’s cute. You have to insert that as a footnote to the rules in very small prints.

This is right.

I think defecting is the obvious answer. It guarantees you $1 with the potential of $5. I don’t really see the benefit in co-op.

Cooperate. Most people will opt for cooperation on the first turn and, since humans practice damned-near absolute reciprocity, we can continue to cooperate and really rank up the points until the last turn, at which point defection is the clear option.

This is especially important if we are playing in ‘groups’, so I am not merely trying to make as much money for myself but also to get more money than anyone else playing the game in a separate chamber (which more accurately starts to model the world).

I like Xunzian’s analysis. Purely logical game-theoretic analysis doesn’t give a conclusive answer, since there are at least two logical approaches – the traditional maximizing payoff approach and the ‘superrational’ approach suggested in the wikipedia article. (I highly recommmend the Hofstadter essay on superrationality in Mathemagical Themas.) Our answer should therefore depend on what we know about the other person; and if we know nothing, it should depend on the expected behavior of a person chosen at random from the population. This is as much a sociological problem as a game-theoretic one.

I do agree that people will usually be inclined to cooperate, especially in an iterated game or a game with competition from other groups. So I would play cooperation/tit-for-tat.

Kriswest, ever with the practical solution. Excellent. :smiley:

I would also opt for co-op, because assuming you partner isn’t particularily malicious it would yeild the best results. Think about it, if there is unanimous co-operation then you are guaranteed $3 a game; if both you are your partner are playing the odds there are various chances of winning more or less per game. It is the only guaranteed way to succeed with serious capital gains. If you play the game ‘strategically’ then while you may come out ahead it becomes much more tricky. You cannot just constantly defect (this is the worst option for both) so you have to attempt to cooperate to maximize gains. This would then become a question of distribution (which CANNOT be neatly distributed), but if the president to always choose to cooperate ad infinitum then your gains will be meximized.

Will to power says defect.

Defect,defect and defect.

It seems most vocal people favor a philosophy of poverty and weakness. The actual numbers show a more pleasing picture, though not by much.

please… you kill your opponent and the banker, then take all the money.

-Imp

Now your talking some sense. :evilfun: =D>

In one study they did 40% of the responses were co-op (although they did this will the actual prisoner dilema).

Boring moral dualistic interpretations. :unamused:

Why? You have no way of knowing what the other will choose, so you have to reckon with a 50/50 chance.

To further the statement, that is right though that action is only best when the opponent makes the choice as if it was at random or we only had to choose once between the options once.

If not co-operate is the most beneficial.

No no what a waste. You don’t kill the golden goose, you just take what you earned from being betrayed then wait for them to get more money Leave them some so that they will be able to give you another chance, you know; benevolence and goodness and so on. Also they can use it to earn more faster. You know darn well you ought to be able to milk them a few times at least. :sunglasses:

Considering that at the time of this post, 6 have voted for co-op, and 6 for defect. An even playing ground. Just going with common sense, most of the resulting pairings would end up with a defector winning (since it is technically more likely for him to be paired up with a co-op and vice versa).