search

Axelrod-Python / axelrod-fortran

Python wrapper library around TourExec Fortran for Axelrod's second tournament.
Other
4 stars 0 forks source link

Representative Strategies #49

Closed meatballs closed 7 years ago

meatballs commented 7 years ago

Axelrod's paper (Axelrod, R. (1980). More Effective Choice in the Prisoner’s Dilemma. Journal of Conflict Resolution, 24(3), 379-403) describes 5 of the strategies as 'Representatives' and claims that they "can be used to account very well for how well a given rule did with the entire set of 63" and that "the scores a given rule sets(sic) with them can be used to predict the average score the rule gets over the full set."

The paper refers to those strategies as S6, S30, S35, S46 and S27 where Sj is the jth strategy. It is likely, but not explicit that 'j' refers to the ranking of the strategies in the second tournament.

It also names the second representative as 'REVISED STATE TRANSITION' by Jonathan Pinkley and two of the others as 'TESTER' by David Gladstein and 'TRANQUILIZER' by Craig Feathers. Given that Pinkley's strategy was ranked 30th, Gladstein's 46th and Feathers' 27th, it seems clear that the Sj notation does indeed relate to the ranking.

In that case, the 5 representative strategies are:

S6: K60R S30: K91R S35: K40R S46: K76R S27: K67R

We should add a representatives list to strategies.py to return this list of strategies.

meatballs commented 7 years ago

The paper gives the formula for predicting the tournament score as:

T = 120 + (S6 0.202) + (S30 0.198) + (S35 0.110) + (S46 0.072) + (S27 * 0.086)