Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site dciem.UUCP Path: utzoo!dciem!mmt From: mmt@dciem.UUCP (Martin Taylor) Newsgroups: net.crypt Subject: Re: Why no hardware random numbers? Message-ID: <1486@dciem.UUCP> Date: Tue, 26-Mar-85 20:00:54 EST Article-I.D.: dciem.1486 Posted: Tue Mar 26 20:00:54 1985 Date-Received: Tue, 26-Mar-85 22:06:26 EST References: <868@utcsri.UUCP> Reply-To: mmt@dciem.UUCP (Martin Taylor) Organization: D.C.I.E.M., Toronto, Canada Lines: 28 Summary: >> ... Why no hardware random numbers? > >Because it's hard to build a hardware random-number generator that >will be unbiased and STAY unbiased. There are any number of things >you can use as a source of random analog garbage -- I seem to recall >that thermal noise in a suitable semiconductor is a reasonable choice -- >but converting them into bits that are reasonably evenly distributed >(and will stay that way without constant readjustment) is harder. >Plus, there is minimal demand. The reproducibility of pseudorandom- >number generators is a feature, not a bug, for most purposes. >-- > Henry Spencer @ U of Toronto Zoology A long time ago, when Flip-Chips (DEC trademark) were a nice new toy, we built a hardware random-number generator to run psychophysical experiments. We even had a knob to control the bit probability, because we figured that would be a nice way to change the probability of targets for people to look or listen for; and we had a big old meter that told us the current probability (averaged over several thousand bits, by mechanical inertia). It worked pretty well, but we never made another or moved it to a different experiment. Pseudo-random numbers are easier to deal with, as Henry says. -- Martin Taylor {allegra,linus,ihnp4,floyd,ubc-vision}!utzoo!dciem!mmt {uw-beaver,qucis,watmath}!utcsri!dciem!mmt