This essay reflects my understanding of IPD.
A rather simplistic if feeble explaination of the complexity of IPD.
Ever wished for a technique by which you would have a means of handling all types of people ? Computer Science,social studies and psychology has come together just to provide such a mechanism. This rather unusual concept was the result of a game known as the Prisoner's Dilemma. The actual experiment was conducted by Professor Robert Axelford, a professor of political science and public policy at the University of Michigan.
The situation consists of two prisoners involved in the same crime and imprisoned in separate cells. They have been arrested only on the basis of circumstatial evidence. Both are individually asked to speak the truth about each other's guilt. Their level of punishment depends on the extent of agreement of their stories. The situation can be considerably simplified if another similar situation is considered. Two intelligences are able to barter some object of value(either material objects or information). They do so without the ability to know what is being obtained for the object that they have given. To quantify the entire transaction, it is deemed that values be assigned to mutual co-operation or to defection by the parties. One sided defection is assigned the highest points as it means the success of one party over the other.
Strategies were many and very complex. However, surprisingly the simplest strategy seemed to be the most effective. `Tit for Tat lasted much more than other algorithms. This idea believes in co-operation first followed by mimicing the opponents previous move. This strategy was nice to nice intelligences and evil to evil intelligences. This was soon followed the `Generous tit for tat which also seemed to be rather successful. This strategy based on random acts of kindness along with the harsh `Tit for Tat principle. As the games continued, the entire system seemed to evolve away from defection towards co-operation. Yet, if at this point some exploiters had entered the society, it would cause destruction. At this point, strategies based on the concept of looking back through the game tree became more and more relavant. The `Pavlov strategy belives in continuing a winning strategy and changing it only if losses occurred. This strategy prevented exploiters entering the society.
The eventual results that could be drawn from the experiment were that the longer a system was allowed to experience evolutionary pressures, it would tend towards a co-operative system. Yet, these systems are themselves prone to sudden collapses. The transitions between the two states of peace and chaos are indeed rare but they are also very abrupt. Thus intelligences will either all co-operate or defect in order to survive. The unknown factors involved in dynamic systems can not be accounted for in totality in Computer modeling. Thus, the entry of players not present before can alter the entire state of the model. The model in its incomplete analysis suggests the use of co-operation as a strategy.
Where might this dry academic analysis be applied you may ask ? From topics as far apart as simple inter-personal relationships, politics to extra terrestial life this aspect of Game theory influences everyone. The dynamics of human relationships with respect to individuals, mobs and unknown and known life forms is directly linked to this issue. So, the next time someone smiles at you throw a hundred watt smile back. The next time someone scowls at you, scowl back (occasionaly smiling if you like the `Generous Tit for Tat strategy ). If it does not work, no problem, just blame it on Game theory and walk away. You are still following the best known strategy known to man - blaming it on someone else.