Re: randomness, experiments versus theory

[John Denker <jsd@MONMOUTH.COM>]

At 10:31 AM 3/30/00 -0500, Ludwik Kowalski wrote:

> I am not denying a possibility that deviations from non-randomness
> can be detected in the Geiger counts (when the contribution of the
> dead time is neglible, that is at a low counting rate). I simply do not
> believe it will happen.

I don't understand the relationship of those two sentences.  Nonidealities
such as dead time are precisely what I was talking about, causing the
Geiger counter output to not pass the test for randomness.  If we agree to
ignore nonidealities, then whatever's left is ideal.

To make my point another way:  how low does the counting rate have to be
before the dead-time correction is negligible?  I'll be generous and
suppose you've got a darn good tube, with a dead time of 100
microseconds.  Then if you average only one count per second, there is a
correction of something like one part in 10^4.  I can design a test to
detect that, a lot more easily than you can design a test to detect a
comparable nonideality in the output of a good PRNG.

And then there is the question of production rate.  The PRNG produces good
numbers much more copiously.  More bits per second.  More bits per dollar.