The new forums will be named Coin Return (based on the most recent vote)! You can check on the status and timeline of the transition to the new forums here.
The Guiding Principles and New Rules document is now in effect.
The Prisoner's Dilemma
Windbit
Registered User regular
Registered User regular
My youth pastor gave me the book Mere Christianity to read, and I've read his C.S. Lewis' arguments for an absolute moral law. According to the idea of an absolute moral law, morality couldn't have evolved on its own, and must have been created by a moral law giver. He phrased his argument like this:
According to game theory, in examples such as the Prisoner's Dilemma, however, what's good for society is good for you personally, as well. If we all betrayed each other, we would all lose. Moral constraints maintain balance and ultimately benefit everyone.
http://en.wikipedia.org/wiki/Prisoner%27s_Dilemma#Morality
[i]C. S. Lewis[/i] wrote:"Human beings, after all, have some sense; they see that you cannot have any real safety of happiness except in a society where every one plays fair, and it is because they see this that they try to behave decently. Now, of course, it is perfectly true that safety and happiness can only come from individuals, classes, and nations being honest and fair and kind to each other. It is one of the most important truths in the world. But as an explanation of why we feel as we do about Right and Wrong, it just misses the point. If we ask: 'Why ought I to be unselfish?' and you reply 'Because it is good for society,' we may then ask, 'Why should I care what's good for society except when it happens to pay me personally?' and then you will have to say, 'Because you ought to be unselfish'-- which simply brings us back to where we started."
According to game theory, in examples such as the Prisoner's Dilemma, however, what's good for society is good for you personally, as well. If we all betrayed each other, we would all lose. Moral constraints maintain balance and ultimately benefit everyone.
http://en.wikipedia.org/wiki/Prisoner%27s_Dilemma#Morality
Windbit on
0
Posts
Simple, realistic truth: evolutionary psychology. We ought'n to be selfish because we are social creatures.
We had that show here in the US - it was called Friend or Foe.
It's the interviews at the end where the loser says stuff like, "Well, I have to hand it to them, they played the game", etc. I don't think my comments would be quite so civil.
Yes, but nobody would want to see you talking, and all they hear is "bleeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeep".
I'll never forget the next part of his story. He said he looked into that student's eyes and had the chilling realization that the boy could not comprehend that his actions were immoral.
Well, that's what happens when you meet an actual honest-to-god sociopath.
What I remember happening is that with small amounts, like at the start when a team would blow every single question and they'd be given $200 to play with so they'd have something to fight over, players would be more likely to pick Friend, because, it's a small amount, who cares? But as things progressed, and the pots got larger, players would start seeing dollar signs and pick Foe. I think I saw a Friend/Friend combo in the final round twice during the entire life of the show.
I hate that stupid show. I always want to slug the person who decides to Split. Then I want to slug the show's creators, because they've come up with a gameshow that basically doesn't give any money away half the time.
the article linked from wikipedia on the show suggests that the frequency of friend/foe picking was money invariant.
what i find more interesting is the variation of the prisoner's dilemma given in friend or foe would result in myself ALWAYS picking foe, not because i'm greedy, but just because it doesn't make sense to pick friend.
Seriously, Kennedy did that once. I think it wound up Foe/Foe.
EDIT: Money didn't matter? Okay, I kinda saw different, but if someone figured it out with this math stuff that's so popular with the kids these days, I won't argue. As for the dilemma variation, I agree. I'd always go Foe. I could not figure out what advantage picking Friend had. Is it just me being a nice guy? This is a game show. I'm here for money. Lots of it. And whether I get any of it is completely out of my hands; it is in the hands of my partner. Likewise, whether he gets money is out of his hands, I'm playing God with him. Both players are basically relying on the other player being stupid.
I'd agree that group-forming has a genetic component, as it is seen in all manner of living organisms... and so does territorial and competitive behavior. I'd say that these behaviors are intended to increase fitness: getting more of one's genes into the population. And it seems as though we all posess them in varying degrees.
Those individuals that are more group-oriented are going to seek to control others just as those who are more territorial and competitive in order to achieve the end of increasing fitness. It's not that "group behavior benefits everyone". Group behavior benefits those who exhibit it.
Religion reinforces group values... morality is an excellent device for prescribing behaviors that are beneficial to the group. When group values are percieved as superior, those who hold them may have greater fitness, while more competitive or territorial individuals may be shunned or imprisoned so that their genes are taken out of the population.
However, this does nothing to address the source of religions. Though construction by humans as a clever way to bring about acceptance of group values seems the most obvious reason to many of us... some people may think otherwise. And we can't prove them wrong.
Pointing out that morality exists neither proves nor disproves divine influence.
Check this article out: The Total Perspective Vortex. It doesn't have that much to do with morality, per se, but it does have to do with how our brains work. According to this, psychopaths are people who see the people as it really is, and normal people delude themselves into thinking that their lives are better than they really are.
Or don't, and accept my feeble summary: We have ethics because, as social creatures, the best way to satisfy our natural, evolution-given desires is to, for the most part, be nice to each other.
EDIT: Although he does have somewhat troubling chapter that suggests we should kill or permanently incarcerate all verifiable sociopaths.
This is the way I see it.
If you say Foe, the other guy gets nothing. No matter what. You might get something, or you might get nothing.
If you say Friend, the other guy will get something. No matter what. You might get something, or you might get nothing.
The question is: are you an asshole?
The program that one was relatively simple. Some others had very complex sets of rules but the winning one didn't. The rules were essentially this:
1) Cooperating is the default
2) Every time the other program chooses to not cooperate, you punish them by not cooperating the next time
3) after each punishment, you return to cooperating
Pretty simple teaching algorithm. Cooperating, in this setup, is by far the best solution because it averages out to the fewest years if taken all of the time. It is, on average, the best scenario for both players and therefore, should be viewed as the best outcome (if you're always going to say no, there is no incentive for the other player to ever say yes).
By punishing the other player, it disincentivizes saying no (cuz then the other player automatically gets 25 years each time he says no).
Pretty good rules. And pretty good rules for life, imho.
Er... no. It stems from people being perfectly rational. Regardless of what your friend/opponent does, you always come out ahead if you screw him over. The rational course of action is to fuck the other guy in the ass. The course of action that realizes the greatest net benefit, however, is the one that screws you.
So no, you pretty much missed the point of the Prisoner's Dilemma.
Generally, something like your program, that's a repeated scenario, will end up as a Grim Trigger game in real life; the moment one person decides not to cooperate, the other person will always pick "don't cooperate" from then on.
That, and it's simply about not being a dick. I'd simply trust the other guy and hope he's smart enough to realize that a six month sentence totally beats risking everything for a five year sentence.
Seriously. If everyone was out to get everyone, and if screwing the other guy was the best way to go, then both prisoners would try to screw each other over and thus spend five years in jail.
If both prisoners are morally sound, then spending six months really wouldn't be bad at all compared to lying and making the other guy spend ten years, or, again, having BOTH unnecessarily spend five years in prison.
This is going off the classical example, obviously.
"Look. Whether or not I get any money is completely up to you. I have no control over it. Whether or not you get any money is completely up to me. You have no control over it. I would prefer that you get some money rather than no money, because I'm not an asshole, so I'm going to vote to split it. You can do what you wish."
And then when the guy won the money, which he probably would, because people are assholes, I would just smile sadly and shake my head at him, and say, "You have five grand, but I have the knowledge that I'm not a dick." And I would feel better about myself.
It's oversimplifying it a bit to say there's no reason to cooperate whatsoever, isn't it? Wouldn't be much of a dilemma then. Everyone would always defect and nobody would win anything. Whatever your opponent does, you do better (or at least no worse, depending on the specifics) by defecting, but the same logic applies to your opponent, while if you both cooperated you'd both win. Saying there's no reason to cooperate is ignoring one half of the dilemma.
Puzzle League: 073119-160185
No, in the true prisoner's dilemma, if you are simply trying to maximize your payoff (or minimize punishment) the dominant strategy is to always betray your counterpart. That means that whether your opponent chooses to betray you or not, your payoff is better if you betray them. However, the traditional dilemma precludes communication and the chance of later retribution.
The iterated prisoner's dilemma is a bit different, where you will have multiple rounds of choices to betray or not. There a tit-for-tat strategy as mentioned is almost the best, depending upon how much you know about your counterpart. More altruistic strategies that maintain retribution tend to be the best in terms of selfish payoff if you have more than a few iterations.
That has important implications for naturally arising morality and altruism.
If you both vote to cooperate, you both get $X. If one of you votes not to cooperate but the other one votes to cooperate, the guy who voted to cooperate gets nothing, and the one who voted not to cooperate gets $Y., where Y > X. If you both vote not to cooperate, you both get nothing.
Now, let's look at this from Contestant A's perspective:
A can decide to cooperate. If he does, and B does too, they both get, say, hypothetically, $1000. If B chooses not to cooperate, A get's nothing.
If A decides not to cooperate, and so does B, A gets nothing. But if B chooses to cooperate, A gets $2000.
So, regardless of what choice B makes, from a monetary perspective, A is better off choosing not to cooperate. If B chooses to cooperate, A gets an extra $1000 from not cooperating. If B chooses not to cooperate, A gets exactly the same that he would have otherwise, $0.
I guess you have to decide how much "feeling guilty over being a dick" is worth to you, but based on the rules of the game, I wouldn't feel guilty at all, because, shit, that's how the game is played.
The reason why not-dickish strategies work in real life is because it isn't going to be one case of prisoner's dilemma in a vacuum. There will be knowledge of your past actions affecting how others act towards you. So if you squeal or betray you might end up in the east river or at very least trash trust others have in you.
As for "Friend or Foe", it is a bit different from the traditional prisoner's dilemma because you can communicate and conspire with your counterpart and that betrayal is only a semi-dominant strategy. That is, if your opponent chooses Foe you get the same payoff regardless of what you do, instead of being better off if you return the favor.
Well, strictly speaking A is not 'better off' choosing not to cooperate if that's what B has chosen - as you say, A will then get nothing either way. And by your logic, B will certainly choose not to cooperate, so there's no way A will win any money whatever he does. In fact he might win, precisely because it's more complicated than just that, and choosing to cooperate is a defensible strategy.
Puzzle League: 073119-160185
The way it works is that the team accrues a pool of money. Then comes the dilemma part.
If both players pick "Friend", the pool is split evenly. If one picks "Friend", and the other "Foe", the "Foe" gets the pool all to themselves. If both pick "Foe", the pool is forefeit.
I guess we're talking in circles here, but my point is that I agree with you that cooperating is the dominant strategy (well, I presume I would if I knew exactly what dominant meant in this context anyway). But I think it's more nuanced that that analysis suggests. IANA-game-theorist, but here's essentially my point:
1. If both players cooperate, nobody wins anything. This is the worst outcome (or at least, equal worst from anyone's perspective).
2. The only way anybody can win is if somebody cooperates.
3. Assuming both players are rational, the only way anyone can win anything is therefore if a rational player chooses to cooperate.
4. Therefore, cooperating can be classified as a rational thing to do because otherwise we have the unsatisfactory situation that rational players will always achieve the worst outcome.
Sorry if I've misused any technical terms, by the way.
Puzzle League: 073119-160185
I mean, seriously, after spending several years in jail, do you really think you're still gonna be friends with the guy who went in with you?
A has to chose between two options with the following results:
1) 50/50 chance of getting $2000 or getting nothing
2) 50/50 chance of getting $1000 or getting nothing
Looking at it this way of course does not include the intangible benefit of "not being a dick", but since intangible benefits are by their nature difficult to measure and in this case subjective, I haven't included it in the above.
Maybe I gave that impression, but it's not what I meant to say. All I mean is that, by your argument, the only rational way for the game to be played involves nobody having any chance of winning any money. This seems to me to be a contradiction.
Puzzle League: 073119-160185
Well, one thing that makes the show version a bit different is that the players can confer, which allows for a bit of brinksmanship. In the "canon" scenario, the prisoners are in separate interrogation rooms, with no means of communication.
In the canon scenario, though, you have past knowledge of the person's character, which is probably more valuable than a few minutes during which to tell a person, "Hey, I have no reason to not fuck you sideways, but I totally promise not to." Morality plays a distinct role in canon. In the case of the game show, it's just a functional of greed and naivete.