1993-10-15 - Re: Generating random numbers

Header Data

From: cme@ellisun.sw.stratus.com (Carl Ellison)
To: cypherpunks@toad.com
Message Hash: 052fb8a5c3ce053fe013d629b91f74eca7445dd62a6a444411769b211c6073dc
Message ID: <9310152123.AA15536@ellisun.sw.stratus.com>
Reply To: N/A
UTC Datetime: 1993-10-15 21:27:10 UTC
Raw Date: Fri, 15 Oct 93 14:27:10 PDT

Raw message

From: cme@ellisun.sw.stratus.com (Carl Ellison)
Date: Fri, 15 Oct 93 14:27:10 PDT
To: cypherpunks@toad.com
Subject: Re:  Generating random numbers
Message-ID: <9310152123.AA15536@ellisun.sw.stratus.com>
MIME-Version: 1.0
Content-Type: text/plain


>Date: Thu, 14 Oct 93 17:46:32 MDT
>From: "Mike Johnson" <exabyte!smtplink!mikej@uunet.uu.net>
>Message-Id: <9309147506.AA750645992@smtplink.exabyte.com>
>Subject: Generating random numbers


>
>If you have an audio input to your computer, try recording noise from a fan,
>traffic on a busy street, or the roar of a crowd at a football game.  Then
>compress the output to remove any obvious redundancies such as 60 Hz hum from
>a fan motor or unused dynamic range of the input digitizer.  This results in
>much better "randomness" than some keystroke timing techniques.

True.

You get even better entropy if you turn off the mic (or unplug it)
and run the sampled audio output into compress -- assuming your mic
is like mine and lets a bit of electronic noise sneak through every byte
or two.

The fan noise sounds random to us but it's relatively simple, in Fourier
space.  The same applies to fluorescent light noise.  Although it's
simple, it can defeat compress - so you end up with low entropy per byte.

 - Carl





Thread