CSCE 475/896

Handout 9: Agony, Agony, Agony

October 2, 2003

In class, we talk about the Prisoner’s Dilemma game:

		Column player
		Cooperate	Defect
Row player	Cooperate	3,3	0,5
	defect	5,0	1,1

In this case, (3,3) has the maximum social welfare, because both cooperated and escaped with some probability. But if I defected and let the other guy stay behind, I would get 5! According to the welfare maximizing and Pareto efficient strategy, we should do cooperate, cooperate. But the only dominant strategy equilibrium and Nash equilibrium is the one where both agents defect. But, if both defect, it will not work. But if you do not defect, then you will be too worried that the other guy will defect. Thus, it is an agonizing dilemma.

And here comes agents. On paper, the problem is not solvable. But agents can monitor their environments, can profile their neighbors, and that would help one make better decisions. For example, if the row player finds out that the column player is an honorable person—i.e., always keeps his promises, then the row player can decide to defect with higher confidence.

You can conduct this simple experiment. Show your friend this game and ask him/her to make a decision—to cooperate or to defect. And tell him/her that you will answer their questions about other player. Pay attention to the questions that they ask. Are the questions about the “characteristics” of the other player?

If yes, then essentially, they are trying to model or profile that other player. And that is one of the areas where agents are good at, and where sound theorems in economics can be most useful and applicable in real-world problems.