I love this article.

I love this article. I work in the casino industry, and I’ve spent a lot of time looking at our own PRNG. We have also investigated fast TRNG systems using quantum teleportation schemes. The TRNG’s we were experimenting with produces 7.99999 99999 99999 bits of entropy out of 8.0 bits possible. The PRNGs are not quite that good. Usually we see about 99.9% of possible entropy, which sounds great, but produces interesting defects in the performance of the machine from a business perspective. In my line of business its the time it takes a casino machine to reach it’s highest confidence on the theoretical profit. The less entropy the longer it takes. After having spent a lot of time porting our private implementation of MT from C++ to Julia, I was curious how the hell did we produce any random numbers on machines like the Commodore 64, the Amiga, the Texas Instruments TRS-80 and so forth. What I discovered was similar algorithms to what you show in your article, but also a funny one. The Commodore 64 used the Sound Interface Device (aka SID), which was a 6581 chip, and set it to a white noise setting, and that produced random numbers between 1 and 256. The entropy must have been horrendous! The SID chip designers eventually went on to form Ensonic and what might be called SID II ended up being built into the ESQ-1 synthesizer. http://www.vintagesynth.com/ensoniq/ens_esq1.php