Maybe. But the proof also gives a lot of cute ideas for how to go further with this sort of demo. Since adding random variables amounts to doing convolutions wi

Maybe. But the proof also gives a lot of cute ideas for how to go further with this sort of demo. Since adding random variables amounts to doing convolutions with their distribution functions, if you had a Fourier transform package in JS you could probably show people some nice visualizations of adding together various random variables, and what it does to their probability distributions. This might give more insight into the generality of the CLT: you take a strange curve, convolve it with itself a bunch of times, and in the convolutions and rescaling it becomes first a skewed distribution, then a normal one.