The latest edition of Scientific American has a freely available feature article on how our decisions are often irrational in game theory terms, but can still be more beneficial than the supposed rational choice.
Game theory tries to understand choices when individuals are working independently and each choice affects the other person’s gains or losses.
In other words, it asks the question ‘considering I don’t know what choice the other person is going to make, what is the best option to maximise my own outcome?’.
This was famously the basis of the American Cold War policy of stockpiling huge amounts of nuclear missiles.
Obviously it would be better if there were fewer nuclear weapons in the world, but if the USA decided to reduce the number of missiles, how could it trust the Soviets to do the same?
Game theory suggested that the best option was to have so many weapons that they could destroy the other country. This way, if the other country reduced their stockpile, they were safe, and if they didn’t, both countries were equally armed.
If this were the case, the potential outcome for starting a nuclear war would be the destruction of both countries. As each wanted to avoid this fate, the idea was that it resulted in a stable but uneasy standoff.
Without a hint of irony, the policy was called MAD, short for Mutual Assured Destruction.
While this is perhaps an extreme example of game theory in action, it can be applied to many situations in which gains and losses are dependent on another person’s choices.
In essence, it’s a mathematical take on a psychological guessing game.
The SciAm article looks at how there are many situations where game theory predicts the most rational outcome, but which may actually lead to much less gains for everyone than if people make an irrational response.
One version of the most rational outcome is the Nash equilibrium, named after Nobel-prize winning mathematician John Nash, who was also the subject of the film A Beautiful Mind.
This is where everyone has settled on a choice where no one has anything to gain by choosing something else.
As the article discusses, this rarely happens in practice, however, and in many cases people just take the risk that they may get screwed over and maximise their benefits as a result.
This suggests that game theory can be a narrow view of human interaction (for example, it doesn’t account for the role of dialogue in the arms race).
This was also a criticism made by Adam Curtis, producer of documentary series The Trap, who argued that game theory had given a cynical and oversimplified view of human psychology that has been disastrously applied to politics.
Whether you buy Curtis’ political view or not, it’s a fascinating example of how trying to model psychological decision making can have a huge influence on world politics.
Curtis’ documentary is variously available online, but unfortunately, video streaming sites are blocked from work, but it seems to turn up quite frequently on a Google search.
And if you want more on economics and rationality, ABC Radio National’s The Philosopher’s Zone just had a programme on the ethics of economic rationalism.
UPDATE: The Trap episode 1, episode 2 and episode 3 are available on Google video. From some reason episode 3 is in three 20 minutes chunks, but the next chunk is linked from each page.
Link to SciAm article ‘The Traveler’s Dilemma’.
Link to The Philosopher’s Zone on economic rationalism.
Mind Hacks 
Maybe somebody who understands game theory a little better can help me here, but I’m a little confused by the “Traveler’s Dilemma” article from Scientific American.
It seems to me that there’s a substantial difference between this game and the Prisoner’s Dilemma – there’s no real negative component to coming out on the losing end, and, therefore, no real incentive to compete with your partner. At worst, you wind up where you started (unless we make the assumption that players are really buying into that back story regarding the broken vase when they play). In this context, it’s no surprise that so many people play the ‘dominated’ strategy, b/c that strategy gives you the most chance to cumulatively receive the maximum payout:
1. Both choose 100. Total payout=200.
2. One chooses 100 while the other chooses 99. Total payout=102+98=200.
All other strategies risk receiving less cumulative reward.
Without incentive to compete, the two players are naturally cooperating to achieve the greatest cumulative result. Yes, that means that the players are employing social considerations, but it’s only a surprise if you assume that the players are seeking to maximize individual rather than total payout.
Or am I missing something?