Wednesday, January 24, 2007

Emergent Design: Prisoner's Dilemma mods

Emergent gameplay is when complex (and often unexpected) behavior arises from the interaction between simple rules.

In class yesterday, I witnessed emergent game design: some students in the class managed to create some very interesting mechanics even though I had not planned that as part of the lesson.

We were talking about the Prisoner's Dilemma, one of the few things from the field of game theory that's actually relevant to making games. Prisoner's Dilemma isn't all that compelling as a game itself, but it often finds itself incorporated as a byproduct into larger social games.

A student offered this modification to make the basic game multiplayer:
  • Each player has the same two choices: Cooperate or Defect. However, if you Defect, you also choose a target -- one of the other players.
  • If you Cooperate, you get 2 months in jail.
  • If you Defect, you only get 1 month in jail, but the target gets an extra 2 months. Cumulative.

Very simple, but I like it as a designer because it has some interesting properties. If you play selfishly, there's no reason to Cooperate; Defecting is better for you, personally, no matter what. However, the total number of months spent in jail among all players increases if you Defect, so in the big picture Defecting hurts everyone.

In practice, players found that opening with Cooperation (assuming multiple trials) was a good strategy; anyone who Defected on the first round made themselves a target for retribution on subsequent rounds. If you were the only Defector at the table on round one, you were pretty much going to get killed for the rest of the game.

An unexpected meta-strategy occurred: once you do start Defecting, it's better to keep your focus on the same person rather than spreading it around. You make fewer enemies that way, and increase the chances that you won't be in Last Place.

I didn't learn much about teaching from this exercise (except to respect my students' game design skills, which I already knew) but the game designer in me was fascinated.

1 comment:

Meg said...

I think my strategy would be to pound on the same person another person is picking on. If both people are targeting the same guy, he can't target both of us and we'll only receive 1/2 the damage we give. If three people or half the class, are picking on the same guy it's even better.

PS I always like reading your lesson plans.