Properties of winning strategies

Over the Christmas period my co-authors and I got news of the acceptance of a paper:

“Properties of winning Iterated Prisoner’s Dilemma strategies”. u This the last paper from Dr. Glynatsi’s PhD thesis. It is quite a neat piece of work as it revisits Axelrod’s original rules for good performance in the Iterated Prisoners Dilemma. Whereas Axelrod’s rules were somewhat of a generalisation from the particular: he ran 2 tournaments that were quite similar. The work in this paper does the opposite as it makes use of the Axelrod library to generate a dataset of 45,686 tournaments without different combinations of strategies, different number of players and different tournament parameters (turn length, noise etc…). This dataset was then analysed to conclude with the following guidelines:

1. Be “nice” in non-noisy environments or when game lengths are longer
2. Be provocable in tournaments with short matches, and generous in tournaments with noise
3. Be a little bit envious
4. Be clever
5. Adapt to the environment (including the population of strategies).

---