Fun note: It's possible to test a programming language's RNG, though it's likely open code anyway. Just use any number of statistical tests. For example, apply a chi-square goodness of fit test. Have the comp generate 100 random numbers. Or perhaps use the slightly more subtle method of testing nearness of numbers in the sequence. For instance, if the RNG is truly random, we expect the chance of the next number rolled being the same as the previous is 1/10, being one off is 2/10, all else is 7/10.

you fail satsitcs examples forever,you forgot the false positive paradox,more info here. http://en.wikipedia.org/wiki/False_positive_paradox

also bronies fail everything forever,so kindly remove the 20% thing,it has no relation to this page at all whatsoever.this is on statistical failure not mematic mutations!