Nice, forgiving, tough and clear – all at the same time

Imagine a situation  where two people are arrested on suspicion of a crime. They are interrogated separately so there is no communication between the prisoners who have two choices; to defect (that is,  inform on the other suspect) or to cooperate (with the other suspect, not the police) by saying nothing. I have seen various descriptions of the various outcomes; if they both remain silent, they both go free or suffer a minimal penalty, if they both defect, they both suffer a major punishment while if one defects and the other cooperates, the cooperator suffers the major punishment but the defector goes free and may even reap some reward. This scenario is called the Prisoner’s Dilemma and is, I think, well known in the ‘Game Theory’ world; I believe that it also received much interest in the ’60s and ’70s in the context of the Cold War arms race.  If the situation only arises once, the rational decision is to defect but if the situation arises repeatedly a different optimum emerges.

In the 1970s an academic at Michigan called Robert Axelrod set up a tournament  in which the entrants were required to submit  computer program to play the part of one of the prisoners. The programs were paired of against each other, in the way of a cup competition; each round consisted of 200 cycles of the scenario. The program that won was submitted by an academic from the University of Toronto called Anatoly Rapoport who adopted a very simple strategy called ‘TIT FOR TAT’. TIT FOR TAT cooperated in round one and  then chose to do exactly what the other prisoner had done in the previous round for every round thereafter. This outcome surprised Axelrod and he ran a second tournament sometime later where the challenge was to beat ‘TIT FOR TAT’; not one of 62 entrants succeeded.

‘TIT FOR TAT’ can be characterised as:

• ‘Nice’ in that it never defects first
• ‘Forgiving’ in that it rewards cooperative behaviour
• ‘Tough’ in that it punishes uncooperative behaviour
• ‘Clear’ in that opposing programs (the other prisoner) can work out the pattern pretty easily.

So what is the relevance to business? Perhaps it is in the pertinent questions that are raised about effective behaviours in competitive situations? Who are the prisoners and who the authority structure that can ‘reward’ or ‘punish’? Is business a prisoner in the competitive marketplace and when does cooperation become anti-competitive? Or are business colleagues all prisoners within the business world? What role does performance management have in encouraging or discouraging behaviours? Does it foster nice, forgiving, tough and clear behaviours if these are seen as desirable?

Since Axelrod’s competitions, I believe that there has been considerable development in modelling of the Prisoner’s Dilemma. Eric Beinhocker describes an evolution of ‘TIT FOR TAT’ embodied in a strategy called Fair. This addresses the situation where both prisoners adopt ‘TIT FOR TAT’ and there is the potential for ‘lock in’ to either mutual cooperation or mutual defecting. Axelrod began to look at strategies for situations where the game history suggested that the other prisoner could be bluffed. Lots of interesting stuff and good reading material for those who are so inclined. The Prisoner’s Dilemma is quite well covered in the literature and there is a good general description in a book called ‘The Origin of Wealth’ by Eric Beinhocker, already mentioned ( see pages 221 to 233) and in ‘Complexity’ by Mitchell Waldrop (pages 262 to 265).

The Prisoner’s Dilemma is an example of a ‘non zero sum’ phenomenon; the net loss of freedom if both cooperate is minimal while the net loss of freedom if one or other defects or if both defect is an order of magnitude greater.

Cooperation then gives the best overall outcome for two prisoners but it is unlikely that the majority of cooperations within business will operate on a one to one basis. Inevitably our cooperations are many to many and these can probably be represented usefully as a series of networks with multiple nodes. It is important for the business then to understand how networks operate in order to ensure that our knowledge and competency networks are cooperative, robust and resilient. There is quite a body of academic research in this area which is relevant at many levels within business and the research is important in order to extract  simple rules from complex phenomena (for instance, be nice, be forgiving, be tough, be clear).