The importance of being "nice"

Another article in this series called "Give to get" was based on work done by Robert Axelrod as recorded in his book The Evolution of Cooperation*. If you have not read that article or want to look at it again click here.

This present article examines one of the fascinating conclusions that emerged from the series of experiments Axelrod organised in order to understand the dynamics of cooperation. But before we look at what it was he found, we need first to explain the approach that Axelrod used.

The iterated Prisoner's Dilemma

In the article "Give to get" I described the hypothetical situation of two criminals arrested under suspicion of having committed a crime together, but where the police do not have enough proof to convict. The two prisoners are placed in separate cells, isolated from one another, and the police visit each of them and offer a deal – the prisoner who offers evidence against the other one will be freed.

The details and consequences of the deal are as follows:

1. If one prisoner betrays the other by giving evidence to the police, the defector will be freed. The one who remained silent, on the other hand, will receive the full punishment under the law.
2. If both decide to betray, both will be punished, but less severely than if they had refused to talk.
3. If neither prisoner accepts the offer, both will receive only a small punishment because of lack of proof.

Axelrod devised a computer tournament in which a variety of "cooperation strategies" are matched against one another, and also against a computer-generated random strategy called RANDOM. Each game consists of exactly 200 moves. On each move, the competing strategies elect whether to "cooperate" or to "defect" i.e. whether to remain silent, or betray the other prisoner.

Points are awarded in this game as follows: both players get 3 points each for mutual cooperation and 1 point each for mutual defection. If one player cooperates while the other defects, the defecting player gets 5 points and the cooperating player 0. So there is a temptation to defect in order to gain the 5 points (the equivalent to being set free in the original Prisoner's Dilemma). But a key element of the iterated Prisoner's Dilemma is that your co-player has the opportunity to get back at you over a series of 200 moves. So if you defect you can expect some sort of retaliation.

The first tournament

Axelrod invited professional game theorists to send in entries and received 14 entries from five disciplines: psychology, economics, political science, mathematics and sociology. Each entry was a set of rules (a "code of ethics", if you like) expressed in the form of a computer program. Clearly, each entrant had to decide how to begin a game (i.e. whether to cooperate or defect), and then how to react to what the other player did on the next and subsequent 198 moves.

The game was run as a round robin, with each strategy paired against each of the other 13 strategies, against the computer-generated RANDOM and against itself. The round robin was run 5 times and the points awarded to each strategy were added up and averaged over the 5 games.

The winning strategy

Many of the entries were very complicated, but the cooperation strategy that won the first tournament was a simple program called TIT FOR TAT. In a game of 200 moves the highest possible score is 1000, although most game scores ranged between 200 and 600 points. Over the five runs TIT FOR TAT averaged 504 points per game.

TIT FOR TAT was submitted by Professor Anatol Rapaport of the University of Toronto. It was the simplest of all the programs submitted and it turned out to be the best! TIT FOR TAT starts out by cooperating on the first move, and thereafter does whatever the other player did on the previous move. The strategy was already known to elicit a good degree of cooperation when played with humans. As an entry in a computer tournament it has the advantage of being difficult to exploit by other strategies and doing well when it plays itself.

Axelrod's analysis and conclusion

The description and analysis of how the various strategies fared and Axelrod's extension of this to how cooperation amongst individuals could have "evolved" in the natural world and what happened between German and British soldiers in the trenches during World War I is fascinating, and I would recommend the book as a very good read. But for partnering practitioners the strongest message to emerge came from Axelrod's analysis of the various types of strategies over a sequence of tournaments, which involved some modifications to the original idea.

The conclusion Axelrod reached was that "nice" strategies always do better in the long run. A "nice" strategy is one that cooperates to begin with, and only defects after the other player has defected. A "nice" strategy is never the first to defect: if you like, it trusts the other player until such time that it has reason not to trust.

Axelrod's advice to a player in the iterated Prisoner's Dilemma takes the form of four simple suggestions:

1. Don't be envious (of the other player's success)
2. Don't be the first to defect (i.e. be "nice")
3. Reciprocate both cooperation and defection
4. Don't be too clever

If you abide by these rules you will do well, indeed better than most, in the long run in the tournament of cooperation. Could these be the four pearls of wisdom to guide your strategy when managing your business partners?

*The Evolution of Cooperation by Robert Axelrod is available on Amazon at this page.

Partnering Points on the importance of being "nice"

• Business partnerships are about "win-win" scenarios, not "zero sum" games. So if your partner is doing well, the chances are that you will also do well. So don't be envious of your partner's success and don't do anything to undermine it. If you have constructed your partnership correctly, you will also benefit from your association with this success.
• Start out with the belief that your partner can be trusted. Don't assume that your partner is out to get you at the earliest opportunity, and don't, whatever you do, let your partner down before your partner has let you down. In Axelrod's analysis, the single best predictor of how well a cooperation strategy performed was whether or not it was "nice".
• But don't be a walkover. If your partner does something that is not in the spirit of the partnership, you must do likewise. In that way you will not be taken advantage of. But also importantly: if your partner subsequently repents and becomes highly supportive again, immediately respond in kind.
• Finally, don't plot and scheme, attempt to second-guess your partner's moves, or try to get away with being less than fully supportive of the partnership. So much effort is wasted, particularly by middle management in larger companies, in working out dubious tactics for shafting partners. In the end it doesn't pay to be too clever. Just get on with the partnership and make sure it works – for both you and your partner.

You are free to use material from istarisArticles in whole or in part, as long as you include complete attribution, including website link. Please also send an email to enquiries@istaris.co.uk, notifying us where the material will appear.

Next article
Previous article