Politics and game theory
I admitted to being a constitution nut. I'm also a game theory nut!
Southampton University Electronics and IT researchers entered an 'Iterated Prisoners' Dilemma' competition. They won. The game has deep significance for the practice of politics, and the way they won has fundamental implications for how alliances are built and how they operate.
The prisoners' dilemma is a mathematical and philosophical game with two players and three rules. The two players are hypothetical prisoners, kept in separate cells and accused of the same crime. There's a deal on the table, which presents each prisoner with the option to turn state's witness, or to deny their involvement in the crime. Their decision is influenced by the three rules:
The idea of performing this game multiple times allowed the Southampton University group to develop the concept of signalling. The University entered sixty different prisoners (actually little computer programs) in the run-off competition. The idea was that in any given iterative game their prisoners would play NOT to win, but to discover if the other player was a Southampton University entry. This is done by playing 'in Morse code', so to speak, by playing in a way that is recognisably Southampton-y.
If the prisoner decided the other prisoner was also a Southampton prisoner, the two programs had a decision-making system that allowed one of the two to sacrifice itself so that the other could win. But if the prisoner decided the other prisoner was not a Southampton prisoner, it automatically denied involvement in the crime for the rest of the game, which was its way of diminishing any advantage the other prisoner could win.
As the article states, Southampton prisoners came first, second and third - but the remaining fifty-seven entries all remained near the end of the entries, having sacrificed themselves for the benefit of their comrades.
Now, it's my belief that this has profound implications for political science. Think of the prisoners' dilemma as a series of games played between political groups who have no knowledge of one another's true intentions regarding an alliance between them. If both groups endorse the alliance, they allow one another to share a part of the political advantage. If they both reject the alliance, they will lose all advantage of co-operation and damage each other. If one endorses the alliance but the other rejects it, the rejector gains the advantage by sacrificing the other on the 'altar of expediency'.
What price the Coalition For Change, guys?
Southampton University Electronics and IT researchers entered an 'Iterated Prisoners' Dilemma' competition. They won. The game has deep significance for the practice of politics, and the way they won has fundamental implications for how alliances are built and how they operate.
The prisoners' dilemma is a mathematical and philosophical game with two players and three rules. The two players are hypothetical prisoners, kept in separate cells and accused of the same crime. There's a deal on the table, which presents each prisoner with the option to turn state's witness, or to deny their involvement in the crime. Their decision is influenced by the three rules:
- If they both turn state's witness, they both get two years in prison.
- If they both deny their involvement, they both get four years in prison.
- If one player turns state's witness, and the other denies involvement, the witness goes free and the other prisoner is imprisoned for five years.
The idea of performing this game multiple times allowed the Southampton University group to develop the concept of signalling. The University entered sixty different prisoners (actually little computer programs) in the run-off competition. The idea was that in any given iterative game their prisoners would play NOT to win, but to discover if the other player was a Southampton University entry. This is done by playing 'in Morse code', so to speak, by playing in a way that is recognisably Southampton-y.
If the prisoner decided the other prisoner was also a Southampton prisoner, the two programs had a decision-making system that allowed one of the two to sacrifice itself so that the other could win. But if the prisoner decided the other prisoner was not a Southampton prisoner, it automatically denied involvement in the crime for the rest of the game, which was its way of diminishing any advantage the other prisoner could win.
As the article states, Southampton prisoners came first, second and third - but the remaining fifty-seven entries all remained near the end of the entries, having sacrificed themselves for the benefit of their comrades.
Now, it's my belief that this has profound implications for political science. Think of the prisoners' dilemma as a series of games played between political groups who have no knowledge of one another's true intentions regarding an alliance between them. If both groups endorse the alliance, they allow one another to share a part of the political advantage. If they both reject the alliance, they will lose all advantage of co-operation and damage each other. If one endorses the alliance but the other rejects it, the rejector gains the advantage by sacrificing the other on the 'altar of expediency'.
What price the Coalition For Change, guys?
posted by mal @ 9:46 AM
<< Home