The Iterated Prisoner’s Dilemma is a very famous game, sometimes linked to real-world economic or political decision-making. For many years, a simple “Tit for Tat” strategy was considered optimal: Do unto others as they do unto you, which had a satisfying simplicity and leads to cooperation.
Last year, two very famous scientists (Bill Press and Freeman Dyson) discovered a new strategy for Iterated Prisoner’s Dilemma which seems to improve on Tit for Tat, and allows the smarter player to outperform, using what they call “extortion.” (If IPD is treated as a model of real-world interactions, this seems sad. 
) I stumbled upon this recently, and thought it interesting enough to share…
My sinuses are killing me, so I’m not going to try to read that article, which I would be bored by halfway through it anyway. Could you summarize what that new strategy is?
I’ve my own health problems and laziness and also hope some mathematician will produce the intuitive explanation to help me! 
But briefly, opponent’s strategy (assuming his memory extends to only one prior move) can be described with two simple parameters. The paper gives two simple functions of those parameters to give the counter-strategy. A key fact, though I’m not sure of the proof, is that opponent cannot improve his outcome by remembering more than one prior result.
An interesting fact – though it may be discussed in articles about the paper rather than the linked-to paper itself – is that Dyson stumbled on the solution while debugging an IPD simulater. The solution involves matrixes with zero determinant, which were generating divide-by-zero faults (or such) in the simulation!