| |
Imagine the following scenario: two criminals in league with one another have been apprehended by the authorities. The authorities don’t have enough evidence to convict either one of them of anything other than a minor crime unless they can get one of them to confess, so they take each prisoner into a separate room and offer them a deal: if they will rat on their companion, the authorities will go easy on them. If you were one of the criminals, what would you do? If you keep quiet and your accomplice spills the beans, you’re going to jail for a long, long time while your accomplice gets off the hook scot-free. However, if you spill the beans and so does your accomplice, you’ll both have to share the sentence fifty-fifty. And if neither one of you says a word, you’ll both only have to face relatively minor charges.
Obviously the best outcome for you and your accomplice as a group would be for both of you to keep quiet and get out of the locker in no time, but how can you trust your accomplice to make that same choice? If they betray you and strike a deal with the prosecution, you’ll be stuck rotting in jail while they are out partying on the beach. Do you try to cooperate with your accomplice and risk getting stabbed in the back, or do you bet that they are going to betray you first and try to protect yourself right from the start?—they are a lying, cheating criminal after all.
The above scenario is called the Prisoners Dilemma and is popular in game theory. It is frequently referenced because it poignantly demonstrates the difficulties of cooperation as well as the rewards. Cooperation can provide great benefits to both parties, so one might think that everyone should be willing to cooperate, but unfortunately, things are not that simple. While cooperation can be beneficial, there is also always the temptation to cheat. And if you cooperate but the other side doesn’t, you get burned. These dynamics are explored in another form of the Prisoners Dilemma, this one played out by computers.
A professor of political science named Robert Axelrod invited psychiatrists, sociologists, game theorists, and political scientists to submit a computer program to play in a contest.* Here are the rules in a nutshell: two computer programs at a time are pitted against each other in a round robin tournament. At the start of each turn, they each independently have to decide whether to cooperate with the other program or to defect. The choice each program makes is then revealed. If both programs chose to cooperate, they each score a three. If both programs chose to defect, they each score a one. And if one side chose to cooperate and the other side chose to defect, the one which cooperated receives zero points while the defector receives five points. Based on this value system, it pays to cheat (that is, to defect) but only if the other side lets you get away with it. If the other side retaliates with defections on subsequent turns, both sides start scoring one’s—not good! since the winner of the tournament is the program which has the highest average score after playing all the other programs.
So who do you think won this tournament? Believe it or not, it was a very simple program called TIT-FOR-TAT. If you cooperated with it, it cooperated with you. If you defected, the next round it defected. If you started cooperating again, it started cooperating again. And it was this program which beat out over sixty other programs in the tournament, some of which were quite sophisticated—too sophisticated for their own good it turns out!
TIT-FOR-TAT was simple, but it was also difficult to exploit. It was nice, but not too nice. Forgoing the Golden Rule, “Do unto others as you would have them do unto you,” TIT-FOR-TAT employed Lex Talionis, “An eye for an eye, a tooth for a tooth.” While TIT-FOR-TAT may not have been a perfect saint, it did always start off nice, that is to say, it would always cooperate on the first round and keep cooperating until the other side defected. In these ways, TIT-FOR-TAT was pleasant and congenial, but if you ever cheated on it, it would always seek immediate retribution. You could take advantage of it, but then you would have to pay the price the very next round.
TIT-FOR-TAT won because it was a cooperator and could reap the rewards of cooperation at the same time that it could avoid being cannibalized by cheaters, that is to say, it could avoid being taken advantage of by programs which gave no cooperation back. A “nice” program which paid no attention to being wronged would soon be eaten alive by “mean” programs which paid back its cooperation with defections. In contrast, TIT-FOR-TAT paid constant attention and took immediate defensive measures as soon as anyone tried to take advantage of it. In the end, TIT-FOR-TAT won because it found the right balance for living in the “real world.” It was fundamentally nice, yet tough enough to withstand those who weren’t.
Eusocial Group seeks to use insights such as these in determining what its own course of action should be in the world. Of course, the real world is much more complex than anything being simulated by this computer tournament, but this tournament is nevertheless helpful in elucidating some of the issues involved and provides useful illumination as a part of a larger debate. Ultimately, the Prisoners Dilemma and the computer tournament are just a sample of the sorts of materials Eusocial Group draws upon in its quest to create itself into a higher-level group. Making recourse to fields such as sociobiology, anthropology, evolutionary psychology, neurobiology, and game theory, Eusocial Group believes that knowledge and understanding derived from these fields can be advantageously applied to creating better social networks. Potentially combining these insights with new technologies such as the internet and social software, Eusocial Group is on a mission to make itself into a group for the future!
|
|
