My recent look at JavaScript as a contender for statistical modeling got me thinking about the different methods used to create random variates. All computers algorithms create Type 1 randomness, which is to say, completely deterministic once you either figure out the underlying algorithm or once you see every number in the algorithm’s period. Jumping outside of software to the hard world around us, it seems possible to create Type 2 or even Type 3 randomness, at least from perspective of an observer who can’t base their predictions on real-time analysis of the generating mechanism (ie, they can’t watch it tick).
My favorite example of a real-world solution to randomizing is shown in the video at top. More details about the construction of the device are here.
What’s your favorite (hardware or virtual) randomization device?