Random number generation, was RE: First computer with real-time clock?

[Dan Veeneman]

At 12:44 AM 8/2/04, you wrote:
> > My favorite on the opposite end, is a random bit stream produced by
> > using the pulses from a geiger counter (and associated radioactive
> > material) to clock a long shift register. It's well-discussed, but I'm
> > not sure anyone ever produced one.
>
>I *think* some military crypto gear did things link this, but I am not
>sure.

There's a "hassle factor" in selling radioactive devices to the military,
but it's easily done these days at home.  Here's what I do:

http://www.etoan.com/random-number-generation/index.html

>I know that AT&T once marketted a true random bitstream chip using a bunch
>of unstable oscillators.

I've seen that described, but no justification for the entropy claimed.

In response to an earlier comment, early noise generators used a 
photomultiplier
tube in a sealed box with the gain turned all the way up.  Using the noise
("snow") from an unused TV (or even radio) channel is not a good idea.


Cheers,

Dan