The first time I heard of it was in this Matt Parker video where he helped do this with a school-sized group of kids [0]. An "AI player" made of matchboxes, run by schoolkids, is a fantastically fun idea.
Nice. I notice that the author has some other interesting posts. I like this one on the James–Stein estimator [0] and this one [1] on day length variations.
Thank you for the kind words! Yes, I think you're right about the missing factor of rho. And rho^2 is being drawn from a chi-squared distribution, not a chi distribution. (But the mode I stated is correct for a chi-squared distribution --- I must have omitted the squares when typing this up.)
https://en.wikipedia.org/wiki/Matchbox_Educable_Noughts_and_...
[0] https://www.youtube.com/watch?v=R9c-_neaxeU
[0] https://joe-antognini.github.io/machine-learning/steins-para... [1] https://joe-antognini.github.io/astronomy/daylight
By the way I think there's a missing factor of rho in the numerator in [1] in the sample transformation section. Should be rho^2