2. The RNG is not even a RNG. It uses a deterministic but secure hash function in a message authentication code format. The results look random. They are not. They are uniformly distributed.
JD's rolls have 3 components: a server seed, a client seed, and an incrementing nonce. The server seed does not change until you Randomize. The client seed does not change. The nonce goes up by one for every roll.
I believed, and I still do, that you can predict a pattern.
[...]
For lack of a true scientific explanation, I resorted to just saying I had magic seeds. However, my understanding of statistics and probabilities, or the lack thereof, convinced me that I can exploit this predictable or deterministic pattern.
But. No one believes me, and I can't explain myself.
I have the same suspicion and the same problem. I can't even think clearly when sha256 is involved

.
Maybe a tried-and-true PRNG could be used seeded with sha256(server_seed || ":" || user_seed) or whatever) instead of that nonce-postfix-thingy that is implemented, which only changes a couple of bits (down to only one bit rougly every second time) of the sha256 input between rolls. I'm not saying this is a flawed approach (I clearly do not know this), just that it's impossible for me to be sure it's not (partly) predictable.
I'm not very firm with math-stuff, just wanted to assure Dabs he's not alone in his suspicion.
EDIT: I highly doubt a human brain would be able to detect such things, btw. Nakowa is just a lucky gambler.