On Fri, 21 Sep 2001 at 10:28:58 -0700, Hrant H Papazian
<[log in to unmask]> held forth thusly:
> From: Rodger Whitlock
> > There should be a simple function that gives
> > uniformly random numbers in the interval 0-1.
> Aha, I got you! :-) "Anybody who intends to produce random
> numbers though arithmetic means is, of course, in a state of
> sin." The father of computer science, Johnny von Neuman said
> that, and of course he was right. The best known way to try
> to *produce* "random" numbers is through taking the modulus
> of relative primes. But that's only because we have yet to
> see the pattern in it. Just like we have yet to see the
> pattern in the chirping of birds.
David Ibbetson also took me to task on this point. May I assure the
collective that I am well aware that so-called random number
algorithms only produce series of mock-random numbers. But as long as
they meet various tests, they suffice in the place of the real thing.
For example, my suggestion to plot a scattergram with random
coordinates using a spreadsheet's built-in random number generator --
you're not going to need more than a few hundred numbers at best.
Even if the N+1-th number starts to repeat the series, you don't care
as long as the ones you got meet the requirements.
The algorithms for "random" number generation are pretty good these
days, as the question has been examined in detail by some very bright
people and their conclusions are widely used.
As for my being in a state of sin, there's no need to invoke
something as esoteric as a heretical or misinformed belief in
algorithmic generation of random numbers to prove the point: I am
quite the sinner entirely aside from that particular personality
Chirp, chirp. Tweet-tweet-tweeeeeeeeeeeeeeeeeeeet.
Victoria, British Columbia, Canada