Is testing for randomness completely futile?
@Bartholomew
Quote: "....truly random numbers...."
Perhaps a thought experiment will help. Suppose you have a fair coin, and suppose you toss the coin a number of times. And suppose, in one trial, the coin comes up heads ten times in a row. This result is "truly random"!! The fact that the result does not pass some test for "randomness" does not make the result "less random".
Longer tests will likely restore the stream of results to something closer to 50/50 (regression to the mean), but that may take a very long time! ...and in the mean time, the stream will show characteristics which will fail some test for "randomness".
Which is a long way of saying that streams of "random numbers" -- irrespective of their source -- may always fail these tests, unless of course the stream is infinitely long.
Biting the hand that feeds IT
