From: gts (gts@optexinc.com)
Date: Thu Sep 12 2002 - 09:43:36 MDT
Eugene, continuing...
> You're *still* confusing entropy and randomness.
As an example of a "Raw RNG" (as opposed to a PRNG) I offer (again) John
Walker's hotbits. (Perhaps you really meant to refer to a "raw PRNG" in
your last message rather than a "raw RNG"? That might explain our
disagreement).
In any case, these numbers from the hotbits device are derived from a
quantum process that is entirely undetermined (that is, if we believe
the common interpretations of QM). They have high entropy. You can
download a program to test the entropy of hotbits here:
http://www.fourmilab.ch/random/
About six years ago, when I first heard of these genuine random numbers,
I downloaded a large sample of them and used ent.exe to test their
entropy. I was quite satisfied. A person concerned about security leaks
could build his own hotbits device.
> Raw RNG output almost never passes tests for
> randomness. You have to postprocess it.
But genuine random numbers from hotbits-like raw RNG's require no
"postprocessing."
Furthermore, any such postprocessing of non-genuine random numbers that
did not entail introduction of quantum indeterminism would result in
only another sequence of non-genuine random numbers. For example there
is no form of deterministic processing capable of removing the
pseudorandom property from the decimal sequence in pi.
-gts
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:16:59 MST