Seeds

William Tanksley wtanksle at dolphin.openprojects.net
Fri May 12 14:01:56 EDT 2000


On 12 May 2000 03:52:37 GMT, Paul Jackson wrote:
>My favorite Random Number web site (being as I am from SGI) is:
>    http://lavarand.sgi.com/
>Lavarand:
>    ...harnessing the power of Lava Lite€ lamps to generate truly
>            random numbers since 1996.

That's cool, but I've never trusted it.  Lava lamps are chaotic, but
they're _slow_, and SGI actually publishes the pictures from which they
derive the random data -- I can envision a program which could take those
pictures, interpolate between them, and then extrapolate to the next
frame.  (Of course, the result with any current technology would only be a
picture which was visually identical; the hash would not possibly match,
so that's not yet a "security hole".)

Microscopic processes, such as a group of chains of NOT gates XORed
together (producing fast, chaotic oscillators), seem a lot more useful for
those purposes.  It's impossible to reproduce or even measure the
conditions bringing about the randomness. (Plus, it's a lot cheaper and
less breakable.) Slap a DES encryption or SHA-1 hash over the result, and
you've got a FIPS certifiable RNG.

-- 
-William "Billy" Tanksley



More information about the Python-list mailing list