yWhat is this?z
This is a model file of artisoc. You can participate in the famous "Prisoner's Dilemma" game played by computer agents. Imagine, two suspects of serious crime are arrested for minor offences. If both of them keep silent, they would be prosecuted and get petty punishments. The prosecutor has not obtained evidence of the serious crime, so he places them in different cells and offer plea bargain to each of them. "If you alone confess, you are free." In that case, another will get severe punishment. If both confess, both will be imprisoned, but for somewhat shorter term.
In his seminal book, The Evolution of Cooperation (Basic Books, 1984), Robert Axelrod made a computer simulation of iterated version of this game. He holds a tournament and invites experts of game theory to submit programs.
The model file you have downloaded is based on the Axelrod's one. The major difference is that you yourself do such choices, just as the programs in Axelrod's tournament do. You will face 50 choices with the same partner. If both cooperate, you get 3 points, and if both defect, you get 1 point. If you defect while the other cooperates you get 5 and your partner gets 0. The partner will change every 50 choices. Every computer agent acts based on one of the following strategies.
"Random" : randomly chooses if C (cooperate) or D (defect).
"allD" : always defects.
"allC" : always cooperates.
"TFT" : cooperates on the first choice, and imitates the hand of the other thereafter.
"TFTT" : cooperates on the first choice, and defects only when the other defects 2 times in a row.
"rTFT" : defects on the first choice, and imitates the hand of the other thereafter.
"Friedman" : cooperates on the first choice. if the other defects, keeps defecting thereafter.
"Davis" : cooperates on the first five choices. if the other does not defect then, keeps cooperating. defects otherwise.
"Joss" : cooperates on the first choice. if the other cooperates, cooperate with 90% probability thereafter. otherwise, never cooperate.
"Tullock" : cooperates on the first five choices and calculate defection rate. defects with a probability which is 10% higher than the rate thereafter.
yHow to usez
- Put the names of strategy in control panel. If you put 0, the model will automatically choose the strategy form the strategies above.
- Run the simulation and then choose cooperation or defection using the control panel.
- The cumulative scores will be displayed in the graph. The red bar is yours.
- The simulation does not end so you can play it as long as you like.
- You can change the points you and agents get from each pairs of choices. Right-click the Reward(mutual cooperation), Sucker(you alone cooperate),Temptation(you alone defect), Punishment(mutual defection) in the tree and change the initial value.
- If the C and D buttons do not work, change the value of "wait" in run preferences.
yAbout this modelz
- This model is tested on artisoc (former JAVA KK-MAS). Please visit KKE website(http://www.kke.co.jp/iit/mas/index.html) for details.
- This model is constructed by Katsuma Mitsutsuji, and Kazutoshi Suzuki. Please send bug reports and comments to Kazutoshi Suzuki
- The newest version will be distributed at Yamakage Lab (http://citrus.c.u-tokyo.ac.jp/). The model is subject to change due to change in simulators and bug fixes. Please check out the website.