1. 6
  1.  

  2. 2

    Tagged games because this is a game theory simulation and that seemed the most literal tag for it.

    I submitted this because I think it’s a fairly readable and accessible paper about simulating iterated games with an eye towards trying to explore social behavior.

    From the glossary, the four major strategies are:

    • Selfish: A strategy of defecting against all other agents.
    • Traitor: A strategy of cooperating with agents of a different tag and defecting against agents of one’s own tag.
    • Ethnocentric: A strategy of cooperating with agents of one’s own tag and defecting against agents with a different tag.
    • Humanitarian: A strategy of cooperating with all other agents.

    I think this sort of work is always a fun read, and something you could probably code up (albeit inefficiently) in an afternoon.

    I think it’s also very important to keep in mind the gross simplifications and limits of this kind of work, and this paper again I think is easy to grasp in where its modeling is short of reality.

    1. 3

      Some scattered thoughts:

      • This is one of those cases where having the source code available to play with on our own would be really neat.
      • Throwing this into a Jupyter notebook or something so people can easily tweak the parameters or do a sweep and witness the results would be really slick.
      • Some modeling of “assimilation” (or “conversion”) whereby cells could adopt (or spread) tags at some rate (perhaps as a separate orthogonal strategy to their PD strategy) might be interesting.
      • The paper explicitly chose a one-shot (ahistorical) PD approach for all interactions; if there was a cluster-wide or group-wide memory of interactions that might be a neat addition. Making that an adjustable parameter (or perhaps even still another orthogonal strategy) could be cool.