If you're using random bits as a way of gathering evidence about random phenomena, consider using the (binary) digits of pi. Then it's win-win: either pi leads you to correct conclusions about random phenomena (so you won't look too foolish) or pi leads you astray (and you'll be remembered as the person whose work led to the discovery of patterns in the digits of pi). On the other hand, if your work leads to the discovery of flaws in some RNG, then your work will be self-effacing in the long term: thanks to your discovery of flaws in that RNG, the RNG will lapse into disuse and tumble into obscurity, dragging with it the fact that it was your work that unearthed the flaw. See mathoverflow.net/questions/26942/is-pi-a-good-random-number-generatorfor a more sober discussion of this topic. Jim Propp On Monday, September 30, 2013, Hilarie Orman <ho@alum.mit.edu> wrote:
Hardware RNGs are very random, except when they aren't. And if they aren't, your adversary may well know about it before you do.
http://crypto.2013.rump.cr.yp.to/55e2988c4ed3c9f635c9a4c3f52fa0b1.pdf
How random are hardware-generated sequences? With biases removed and waiting out correlation times, they are completely random. Even then, a little bit of bias or correlation won't compromise cryptography.
_______________________________________________ math-fun mailing list math-fun@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/math-fun