Vindauga

Perhaps strangely, one of the things that I keep returning to in my research, or which often piques my interest, are issue in rational choice theory. I am constantly annoyed at the logical solutions presented to dilemmas and problems. But annoyed in a way that makes me want to participate in not ignore the discussion.

While preparing for a recent class I came across (it was suggested to me by a friend) a particular dilemma of which I had been ignorant: The Traveler’s Dilemma. My history in this arena has been largely confined to the Prisoner’s Dilemma. It is actually one of the only things I ever researched completely independently during my graduate career. It was never the topic of any of my classes but it was something that I had learned way back in high school (from Scientific American). In graduate school I started reading much of the contemporary discussion. And I just wanted to beat my head on the table. I still don’t know how everyone misses the point of choosing how to act in these kinds of scenarios.

Let me see if I can be more clear. The Traveler’s Dilemma (TD), apparently invented by Kaushik Basu, is purported to reveal the logical of illogical behavior. Or something like that anyway. Honestly, I said it that way because its got a nice ring to it. Two separate people have had an identical item damaged by an airline. The airline wants to compensate them but needs to find out how much the items were worth. So, a manager suggests making each person write down the value of the item. If the value is the same, then each person gets that amount. If the amount is different then each person gets the lower value, plus or minus some reward or punishment (a reward for telling the truth and a punishment for lying). If we limit the range of values, say 2 t0 100 with a reward of +2 and a punishment of -2, then there is a logical response to what value should picked. Unfortunately, its not the value that we would naively suppose. Its 2.

It works like this. Each person might immediately think “I should write 100″ but then they might also think “ahh, since the other person is going to think that also, I could make more money by writing 99″ (99, being the lower, will be assumed to be true and so will be paid and because I said it, I get the plus to and so I will be paid 101 while the other person will be paid 98). But then the person will think “ah-ha, but I know that and so the other must know it too, so rather than picking 99, I should pick 98, and Then I will make the most amount possible.” Sounds good so far, except that this logic continues: “Right, but since I could realize that, so could the other. I better take 97″ and so it goes all the way back to 2. Which, I should point out is the lowest amount you could bid and since you will both say it, then you will both only get 2.

Thus goes the logic. And apparently nearly all decision theories produce the same, maddening result. Its maddening for a number of reasons but the most obvious is that it is completely irrational to imagine that the best or right or good move is to write 2 when you could have written 100; that somehow writing 2 is the rational choice. Its also maddening because in every experiment done on this, no one ever writes the so-called rational choice. No, before you accuse philosophy of being so hopelessly lost in the clouds that they can’t see how irrational the rational choice is let me just say that this isn’t strictly philosophy’s problems; its really a problem of mathematicians, economists, and psychologists. It exists in philosophy but its not a huge philosophical field. Now, you might want to say that this is just academics mentally masturbating their way into publications and tenure. That might be true but considering how many of the policy decisions in both companies and governments are made by referencing decision theories (especially economic ones) I think it might be excessive to ignore the value of this level of abstract thinking.

The same problem appears in iterated prisoner’s dilemmas. In the prisoner’s dilemma two criminals have been caught are are being interrogated separately. Each must decide whether she will talk (defect/turn) or keep her mouth shut (cooperate/stay). Game Theorists claim that the “best” move is to always defect. They think this because if your partner chooses to cooperate then you can get a higher pay-off by defecting (5 for you with 0 for him). On the other hand if your partner chooses to defect then you would be foolish to cooperate (0 for you and 5 for him) so you should defect as well, which would give you 1 and him 1 (both cooperating will give you each 3). So, it seems that the best move is to always defect. And this becomes even more prominent if you are only going to interact one time with that other; with no possibility of retaliation the other has no motivation to cooperate. Strangely, in a finite set of interactions, we end up with the same conclusion. If we are going to interact 10 times, then it seems the best move is to cooperate. Only, it will occur to the shrewd that defecting on the final hand can yield a higher total if your partner continues to cooperate. But surely your partner will realize that as well and so they will defect on the 10th round so you must defect on the 9th. But I know that you know that I know, so I must defect on the 8th. This ends back at the first round. Again, we must defect on the first encounter. And again, this result is obviously worse than the alternative: cooperating from the start (in fact, its even worse than cooperating from the start and letting your partner cheat you in the end). And yet theorists continue to suggest that this is the rational response.

There are times when I think everyone is just missing the obvious point that “best” does not mean what they seem to think it means. There are other times when I think their big mistake is confusing the comparison. But I always think that they miss the difference between considering my actions as purely reactive and being active. That is, all of these discussions tend to focus on “what the other person is going to do.” And I don’t really care what the other person is going to do. For example, I’m no rat fink. I don’t have to think “what is my partner in crime going to do” in order to make a decision about how I want to behave. I could rat him out and get some time off. Or, I could be afraid that he is going to rat me out. But, I could also decide that I don’t want to be the kind of person that rats others out. I could decide that even though it might mean more time in jail, the best life is one in which I don’t rat people out. And so, the iterated prisoner’s dilemma is not really a dilemma at all.  Likewise the traveler’s dilemma doesn’t have to be about my fear of how the other is going to act but could be about what type of person I want to be.

And more to the point. Any reasonable (notice I didn’t use rational) would, if they were capable of doing the complex reasoning necessary to realize that logic dictated that always defect, would recognize that they other knows it as well and would not defect. Seriously, the solution to this problem shows up in Wargames: “Strange game. The only way to win is not to play.” As soon as you know enough to know where that self-interested thinking takes you, you should know enough to know not to use it.

Perhaps the hard part is getting people enough experience with the problem to be able to avoid it.