Multi Arm Bandit Free

A simple game to demonstrate the multi-armed bandit problem. There are three buttons each with a different probability of winning a token. The probabilities are 0.35, 0.5 and 0.65. You have 100 turns and each time you push a button you either win or lose a token.

This illustrates the exploitation / exploration trade-off: how much do you explore which is the best button (they are shuffled each time you play) versus exploiting which seems to be best?

A quick word of warning: the screen is set to stay on (I find it less annoying this way when using it for teaching), so remember to quit the app when you're finished with it.