Jamie Dallaire on Thu, 20 Nov 2008 22:24:57 -0700 (MST)


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

[s-d] Proto: Subgame/Contest: The Evolution of Cooperation


Cross posting because I figure there could be interest on both sides. If
need be this can be a Werewolves-like endeavour.

Would anyone be interested in playing the following, based on Robert
Axelrod's and WD Hamilton's "The Evolution of Cooperation"? (see links
below)

Basically the subgame would go as follows:

Anyone who wishes to participate is given time to write up a script that
will then play a game against each other participant's script. Each script
is matched against each other script, in a round robin format.

The game played between scripts is as follows (it's studied in game theory,
and it's called "Ultimatum"):

Script A is tasked with dividing 10 points between itself and Script B in
any manner it chooses (9/1, 5/5, 10/0, etc.). The proposed division is then
presented to Script B, who may choose to accept the division (in which case
each Script is awarded the proposed number of points) or to reject it
(neither Script gets any points). In the next round, the roles are reversed.
A match between two scripts lasts for a large number of rounds (e.g. a
couple hundreds) that is randomly determined, unique for each match, and
unknown to the scripts until the end of the game. (The possible bounds may
be known, e.g. always between 100 and 300 rounds) Note that the scripts can
be set up to take the outcome of previous rounds as inputs, e.g. to track
the behaviour of the other script and use it to make decisions.

At the end of the round robin, each script has faced every other script, and
obtained an average score in each match (e.g. Script C averaged 3.4 points
per decision against Script F). The averages from each match played by a
given script are cumulated, and the author of the script with the highest
total is the big winner.

Please, no collusion in script design.

Axelrod initially played this with a different game (Prisoner's Dilemma,
rather than Ultimatum) and solicited entries from people in a whole bunch of
fields (evolutionary biology, political science, economics, mathematics,
psychology, artificial intelligence, you name it). If you do think this
sounds interesting and haven't heard of Axelrod's results, it would be a
good idea to familiarize yourself with the outcome of that project.

Wikipedia Article: http://en.wikipedia.org/wiki/The_Evolution_of_Cooperation
Science Article:
http://www.sciencemag.org/cgi/content/abstract/211/4489/1390
Book:
http://www.amazon.ca/gp/search?ie=UTF8&keywords=the+evolution+of+cooperation&tag=mycroft07-20&index=blended&linkCode=ur2&camp=15121&creative=330641

Note that while redoing Prisoner's Dilemma might be boring, it doesn't have
to be Ultimatum either. A number of other games on
http://en.wikipedia.org/wiki/List_of_games_in_game_theory could be quite
interesting. Including Battle of the Sexes, Nash Bargaining Game (sort of
like Ultimatum), Stag Hunt, etc.

Cheers!

Billy Pilgrim
_______________________________________________
spoon-discuss mailing list
spoon-discuss@xxxxxxxxx
http://lists.ellipsis.cx/mailman/listinfo/spoon-discuss