A classic experiment in the fields of computer science and game theory is the simulation of the Prisoner's Dilemma. The original experiment showed that the optimal behavior for this particular competitive game is Tit-For-Tat, which was revolutionary because it was an entirely cooperative behavior, not competitive or greedy like everyone expected.
Richard Dawkins explains it well http://www.youtube.com/watch?v=48EWLj3gIJ8
I will be running a similar simulation event sometime in December, so if you're a coder or want to learn to code to submit something, write em up and post them here
- Read a single line, this will be your opponent's ID
- Process decision based on previous experience
- Write a single line to STDOUT with your decision as "cooperate" or "defect"
- Read a single line, this will be your opponent's decision as "cooperate" or "defect"
- Update your decision making process based on result
Example stream of 3 games -STDIN 8STDOUT cooperateSTDIN cooperateSTDIN 7STDOUT cooperateSTDIN defectSTDIN 5STDOUT defectSTDIN cooperate
- You will be given 500M of disk space, 500M of memory, one CPU core, and STDIN and STDOUT will be hooked up to the simulation environment.
- You may store data you uncover about your opponents and use them in the decision making process.
- You may take up to 5 minutes to process your decision (though I expect it to take much less time).
I will try to recreate a few of the basic examples: tit-for-tat, always-defect, etc, for comparison.
If this goes well I may put this up as a permanent installation