Len:

10mhz?
My gawd man, I can walk that fast!

I stand behind what I said, no perfect random number generators exist, if you
need a really good one--it can't be done with computer algorithms (but, the one
doing our lottery is a really **** poor one--probably worse than your "no
repeats for 913 year one @ 10mhz" even... grin

It most EXACTLY becomes a question of, "How good of a random number generator
do you need?"

I grant you, most apps do not need that good of one, games of chance in
reno/las vegas are ran off damn poor ones...

Also, if you read about egg at princeton.edu, you will see that computer random
number generators are really a bad idea, the human mind can influence
results... on some quantum-metaphysical level it seems the mind has powers
which we have only guessed about in fairy tales... I leave that to your
further research however...

Remember Len, we really do agree on most though... or, close enough...

John

wrote in message
ups.com...
From: John Smith on Jul 29, 1:12 pm

an_old_friend:


Perhaps spread spectrum is used by spies, who knows, or sat links--we will
always catch only the poorest of spies--undoubtedly they DO NOT represent the
"good ones."

But, most radio is a poor vector for spies, the internet is a much more
viable
medium...

Ahem...those "number stations" on HF aren't for sports scores or
lottery numbers... :-)

A true "random number generator" has escaped being ever realized in a
practical
form.

Untrue, John. PRSG (Pseudo-Random Sequence Generators) have been
in common use in both communications and instrumentation for about
three decades now. Using just 9 standard digital logic packages
with a 10 MHz clock, the PRSG I built for instrumentation would
not repeat until 913 years had passed. Reference: Electronics
Designer's Casebook Number 3, a collection of Electronics
magazine articles published between February '78 and January '79.
I was the author of that. "Electronics" magazine was a bi-weekly
industry/subscription periodical published by McGraw-Hill; McGraw
Hill morphed it into four separate monthlies.

The "random noise" from the background radiation of the universe comes
very, very close.

No, does NOT "come close," that IS the definition of random.

In computing, if a very high quality "random number generator" is needed, it
will always be outboard (white noise generator.) No computer algorithm ever
developed is able to generate REAL random numbers. Success is only measured
in
how close they can come to the ideal...

Sigh. PERIODICITY is at question? I would say that a repeat
period of a century or more is pretty dang good. Periodicity
greater than that is found in the DES and some of the algorithms
at the NSA.

SOME finite periodicity is NECESSARY to set code keys and thus
enabling a decode to start to prepare to get ready to begin.
That just cannot be done with true random noise thingies.

In some of the Swiss (formerly Swedish) Crypto AG products, a
true natural random noise source generates the random key
patterns for both encryption and decryption sequences. A
problem there is that the decrypt sequence MUST be identical
to the encrypt sequence and that decrypt sequence transported
to a recipient. With the DES and similar PRSG-driven
sequences, the decrypt key is NOT required to be sent
separately...all that is required is to set the sequence at
some pre-determined state (the "code key" enters that) and
this aligns the sequence with the received sequence...a sync
is possible and decrypt can proceed.

As to "simple algorithm" periodicity, a 913 year pattern
repeat at a 10 MHz clock is quite long. That was achieved
with a 2^33-1 sequence bit pattern Exclusive-ORed with a
2^25-1 bit pattern. Each of the individual PRSGs had NO
common factors in periodicity so they Ex-ORed to a pattern
of 2.88 x 10^17 clock periods. That could be implemented
on any PC (I did that just for funzies) with a 2 GHz clock
and sequence faster than the hardware version clock at
10 MHz. So that one repeated about every 100 years...

... the "noise" from a large number of typists keyboards might be close
enough,
although not perfectly random, for some applications...

The humans-replacing-monkeys (at the KGB) was taken from
David Kahn's "The Codebreakers" and was a quote from
someone else.

Today, at this "late" hour of 5:15 PM PDT, thousands of
WLANs are very busy within RF range of one another, NOT
interfering with each other thanks to some clever and
longer PRSGs. All those WLANs can also synchronize with
one another should they have to start from power-down
condition.

Add to that the garage door openers which have a very short
data burst on an RF carrier...add millions more in remote
keyless entry automobile "watch fob" transmitters. Their
pattern security is so good that they CAN and HAVE replaced
mechanical counterparts in any environment. [yes, the auto
fob transmitters have keys attached as a security in case
the car battery goes kafooey...and for the mental/emotional
security of the numbnuts conservatives who don't trust those
new-fangled digital gizmos]

For pseudo-random number generation in theory, along with
tests on things therein being random, there are 155 pages
worth of good stuff in Donald E. Knuth's "The Art of
Computer Programming," Chapter 3, Volume 2. I have the
three-volume set and will entertain any offers of
purchase (plus shipping costs)...provided I can trust
the buyer (this group doesn't ensure my trust much...).

bit bit