Truly random numbers

[Chad Netzer]

On Wed, 2003-02-12 at 07:56, news wrote:

> I have been told that turning up the gain on a microphone will produce
> thermal noise (= truly random numbers) without the necessity of buying a
> designated "random number generator".

Well, true, there will be some noise.  But there will also be the
potential for signal. (ie. someone could speak into the mike)  The trick
is in knowing how to extract the noise, and not the signal.

The most talked about approach is just sampling the least significant
bit, which is often thought of as noisy.  But even this might be quite
biased.  Better methods will take into account the physical parameters
of the D/A converter, and the transducer, to estimate the entropy of the
noise, then use cryptographic hashes to produce a (hopefully) known
number of "truly unguessable bits".

> But I haven't tried it myself.

Neither have I.  But I think most "random" physical sources needs
further processing to be useful.

-- 
Bay Area Python Interest Group - http://www.baypiggies.net/

Chad Netzer
(any opinion expressed is my own and not NASA's or my employer's)