I believe SHA512 is not broken. That guy was just lucky. The pattern may be something else, like looking at the previous bets, and not just dependent on a win or loss.
I'm sure you know what to do or what to check if the server seeds have been compromised. I don't think they have. How does your site store it?
My guess, he's just a good gambler.
The RNG is SHA512 on the seeds and nonce. There is no known mathematical analysis except to guess when you are going to win and when you are going to lose. But the use of the same seeds with an incrementing nonce does imply that the dice, in this case, do have some sort of memory.
Again, that is gambler's fallacy, but as a player, I believe it to be true in this case. However, I am not going to debate or argue it with anyone here. I've said what I've said in the other thread about my martingale sequences on 87%. And I've come up positive after almost a million bets.
This new whale must have a variation on this system.
But these are just anecdotal ... he did what he did, I did what I did, doesn't prove anything.
Well.. SHA-256 has been demonstrated to have non-random properties when 'step reduced'. I believe some of the javascript sha algorithms have the option to set the number of rounds applied.
A lower number of rounds can be used in order to reduce CPU time - and I think it's worth reviewing whether enough rounds are being applied in this case.