The first prisoner’s dilemma tournament last week was so much fun, so we’re gonna do another, with more teams this time! To watch or play, join this FB group.
A prisoner’s dilemma strategy is a description of logic about when to cooperate and when to defect, based on how other people cooperate with or defect against you. “Tit for tat” is a popular strategy that starts by cooperating and then copies the last move from the other player. It’s the strategy of “perfect fairness”. Other strategies are nicer, or meaner, and to better/worse against different kinds of strategies. A strategy can feel a lot like a worldview… that’s why this game gets so interesting! This link has a bunch of strategies, starting simple and getting sorta fancy.
Some famous strategies, left here without explanation:
Here’s how the 1st tournament worked:
- 2 teams: To play, you joined one of 2 teams (Ant or Beetle) and picked a strategy to play against every person on the other team. You could talk to your team to get ideas and try to coordinate strategies. Let’s say you eventually picked “complete randomness” as a strategy — that means you’d cooperate 50% of the time and defect 50% of the time.
- 3 rounds: Each round the game played your strategy against each player on the other team 100 times. So if, say, one of the players on the other team was “always cooperate”, you’d probably get 50% cooperate-cooperate and 50% defect-cooperate (since you’re random).
- Points work like this: cooperate-cooperate = 3 points for each player, defect-cooperate = 5 points for you, 0 for them, cooperate-defect is the opposite (0 for you, 5 for them), and defect-defect is 1 point each. So, after playing someone 100 times, your score for the round would be 50% 3 points and 50% 1 point = 200 points. If both sides cooperated 100% of the time you’d get 3 points * 100 = 300 points. Et cetera.
- How scores are tallied: Every round you play every player on the other team, and so does everyone else. I use a script to do this so it happens automatically. Each team gets the average score of everyone on it and I report back how each player, and each team did that round. (But you only know the strategies your team used.)
- Between rounds: You can change your strategy between rounds to try to have the best set of strategies to amass points.
- How to win: After 3 rounds, the team with the most points is declared the winner! Last time, that was Team Ant. And the people within each team with the most points are MVPs! You can see results here.
We learned some interesting things during this trial tournament, and are gonna tweak a few rules to make it even more interesting.
- More teams: We’re gonna allow more than 2 teams (3–5, I think). This will make it so you can’t extort (or be extorted) entirely by another team because you don’t have to rely on them for all your team’s points. Other dynamics/conflicts of interest will emerge I’m sure.
- Self-organized: Teams will have some players assigned randomly and but some ability to recruit and self-organize prior to the game. This will help people get aligned on strategy ahead of time and make sure they want to be on a team together. We have 4 team captains so far.
- Democratic: All the rules of the tournament can be changed (including this one) by making a proposal and getting majority vote. Last game we changed the number of rounds and slightly edited the goal of the tournament. This is where meta prisoner’s dilemma can be played.
Want to learn more and possibly play in the next tournament?
Team registration starts Sunday, July 22nd, and the games will start early August! Whether you want to play or just watch, join this Facebook Group.
If you want a quick primer on why prisoner’s dilemma is so interesting as a tool for understanding how cooperation evolved, this is one of the better intros:
🎟 Register for the tournament before July 29th! https://facebook.com/groups/dilemma