3 ms·
Veratasium just made an approachable video on this subject recently, I definitely recommend. It rebukes the whole 'oh no I just heard what a Nash equilibrium i
by hughesjj 3y ago
Veratasium just made an approachable video on this subject recently, I definitely recommend. It rebukes the whole 'oh no I just heard what a Nash equilibrium is and now I'm worried cooperation is impossible long term' fears some people interpret his thought experiment (prisoners dillema) to have. Tldr it's rare that you can't coordinate, communicate, or have exactly one trial.
https://youtu.be/mScpHTIi-kM?si=v8FvrfMCHioLJET5 https://youtu.be/mScpHTIi-kM?si=v8FvrfMCHioLJET5
- hot_gril 3y agoIs the longer tl;dr that with multiple trials and possible miscommunication, the winning individual strategy is tit-for-tat, and the winning group strat is some softer tit-for-tat with one forgivable betrayal?
- JoshTriplett 3y agoThe winning multi-round group strategy in a game in which you can identify and remember different people is, roughly, "cooperate with cooperators, defect with defectors". https://ncase.me/trust/ https://ncase.me/trust/ is well worth playing to get an idea of how this works. (Also, by way of acknowledging that reality is more complex than simulations like the Prisoner's Dilemma, in reality you might also have options like "stop interacting with defectors in order to avoid lose-lose situations" or "warn others about defectors, and listen to others' warnings but weight them based on whether the entity giving the warning is honest". And conversely, defectors get more options like "pretend to be a different entity with no history" or "pretend to be a known trusted entity".)
- hot_gril 3y agoYep, I remember playing that minigame a while back, it's a nice one. Also the "wisdom of crowds" one on that same site.