Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site cadovax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!whuxl!whuxlm!akgua!sdcsvax!sdcrdcf!trwrb!trwrba!cadovax!keithd
From: keithd@cadovax.UUCP (Keith Doyle)
Newsgroups: net.origins
Subject: Re: CREATIONIST ARGUMENTS, PART II
Message-ID: <471@cadovax.UUCP>
Date: Mon, 18-Mar-85 22:33:41 EST
Article-I.D.: cadovax.471
Posted: Mon Mar 18 22:33:41 1985
Date-Received: Sat, 23-Mar-85 00:46:34 EST
References: <241@cmu-cs-gandalf.ARPA>, <237@rtech.ARPA>
Organization: Contel Cado, Torrance, CA
Lines: 36

[..........]
>Here is a thought experiment.  Imagine that you have 50
>dice, and that your goal is to get them all to show 3 on their uppermost
>faces by some random process.  You could roll all 50 dice over and over again
>until all threes show.  If you threw the dice once a second, the expected time
>until success would be (6**50)/2 seconds (for you non-programmers, this means
>6 to the 50th power divided by 2).  The reasoning is that, on the average,
>you should expect to go through half of the possible arrangements of the dice
>before you hit on the right one.

>Now suppose that take you one die and roll it until you get a three.  When
>that happens, take a second die and roll it until it shows a three.  Continue
>this until all of the dice show threes.  This will take much less time than
>the above method, about 3 times 50 seconds if you roll once each second.
>Here the resoning is that, for each die, you should expect to go through half
>of the possibilities before getting the one you want and proceeding to the
>next die.  With this method, you would spend about 150 seconds versus an
>extremely long time with the other method.

>                 ....  My thought experiment was only intended to show that
>order can come out of random processes in a relatively short length of time
>if partial results are preserved along the way.
---
>Jeff Lichtman at rtech (Relational Technology, Inc.)

Interesting.  This is very similar to a method used to generate 1/f noise
via computer.  With this method, you set up several 'die' and you roll
each of them according to a binary progression, i.e. the first time
you roll the first die, the second time the second, the third time
both the first and second, the fourth time you roll the third die, etc.
the resultant noise is obtained by summing all the dice.

Keith Doyle
#  {ucbvax,ihnp4,decvax}!trwrb!cadovax!keithd
"You'll THINK to know what you REALLY pay!"