I had the pleasure of interviewing a Master's in Computer Science student recently. As you may know, the MS with Thesis student is part of a dying tribe. Why bother to hang around in a dinky lab for years when you can graduate with the bare minimum courses and enter the high flying corporate world? What interested me in her resume was the description of a game theory competition involving wireless toy cars.
The game is as follows. There are N wireless enabled toy cars and M base stations. Each of the M base stations emit 100 packets to each of the cars. A car can receive the packet directly from the base station or from another car if it is too far away. On receiving a packet destined for another, each car can decide to either drop, forward or misdirect it. As part of the game design, each car should broadcast the number of packets received by it and the action taken. The cars take random paths around the base stations. Each car's aim is to obtain the maximum number of packets destined for it.
If you are the owner of a car, what will you do for winning the game? Interestingly this game is very similar to the iterative Prisoner's Dilemma devised by Robert Axelrod. Axelrod surmised in his Evolution of Cooperation that selfish players can come together to cooperate if they find that it improves their chances in the long term. For several years he organized a competition pitting computer agents with each other. Each of these agents would then have to play against all others. The one which obtains the maximum points would be judged the winner. Several competitors from various corners of the world participated. Strategies ranging from AI, game theory and statistical inference were employed. Guess which strategy won?
Apparently "tit for tat" was the winner. The agent started by cooperating with its opponent. From then on the next step would be based on the opponent's response.
So the next time you are unsure on what to do with an unknown rival, try this approach. Start by extending cooperation to him. If he is also a "tit for tat" player, you would then end up cooperating him. If he screws you, you will get your revenge next time. This is not rocket science. We have known it all these years as evident from the proverb "Fool me once, shame on you. Fool me twice, shame on me".