I’ve been interested in random number generation since I did my first Monte Carlo computer simulation more than half a century ago. I became aware that the problem is that computers flawlessly do the same things over and over again. The random number generators available then did NOT generate random numbers, they generated a random number sequence over and over again. It was possible to increase the length of the sequence, but it was not possible to prevent its repetition. This becomes a problem in such things as encryption, passwords, and also Monte Carlo simulations. It’s great to see new solutions for this.
True. It is incredibly hard to get truly random numbers from a computer. This does have some implications for cryptography, though most modern cyphers get around this by using more bits than they otherwise would.
Here's an interesting way to get around generating random data for a crypto seed: DiceKeys. Note: the link is actually demoware. The actual dice keys are physical dice. It's an interesting way to generate a significant amount of entropy. There are 25 letters on the dice (A-Z minus Q). Each side is numbered 1-6. When shaken into the container, there are also 4 possible orientations for the individual dice. All told, that can produce 196 bits of entropy. Put another way, there are 124,127,134,662,179,891,202,329,100,571,859,806,502,566,406,865,813,504,000,000 different ways to orient the keys in the box.
More info at DiceKeys.com. I bought a couple to play with.