Originally Posted by psionl0
You don't understand what I mean, which is not "cracking" SHA256. I'm simply talking about a filter that would reduce the average number of trial nonces that would need to be tested to find one that produced the required number of leading zeros in the hash. For all I know, some sort of neural net might find a usable pattern in past blocks. I'm not saying that I expect that to happen; I'm saying that's one of the reasons I'm not buying any bitcoins -- rather far down the list, actually.