Post by mhoye but seasonally spooky

Maybe 25 years ago now, I remember Don Knuth writing about withdrawing to retool some hashing algorithm he'd been working on, explaining that while all their testing appeared to produce a promisingly uniform randomness, somebody clever had using tried using a linear sequence of integer seeds - 0,1,2,3,4.... that in those conditions the hash showed alarming signs of being very nonrandom, but nobody had thought to test it like that before.

Thinking about that a lot this week. For some reason.

I wish I could find a proper link to cite this, but [gestures vaguely].