X-Message-Number: 26039
Date: Sun, 17 Apr 2005 07:02:08 -0400
From: Thomas Donaldson <>
Subject: More on time-sharing, for Crevier

To Daniel Crevier:

I have problems with your counterexample. Severe ones. As you surely
know yourself, weather prediction at best can give a general idea
of future weather, and frequently fails (even now, with high speed
computers etc). In terms of a computer system which can make a 
prediction before the event happens, our computers running weather
prediction software just don't work anywhere close to reality.

Unless you live in some area with very little weather, the weather
predictions you get will probably frequently miss.

I will add that my understanding of the equations used to predict
weather says that those equations don't behave very well at all
with time. Some equations have solutions which converge to 0 with
time, and so do their errors. The equations used to predict the
weather DIVERGE with time. The way this is normally handled is
to produce solutions which are as accurate as physically possible, 
which puts off the time until they've clearly diverged. We produce
a prediction to the trillionth place in the hope that it will still
follow the weather to decimal (0.01) accuracy tomorrow.

I also said that rapid time-sharing would not lead to catastrophe
always. There might be a delay (of time unknown) until the time
when predictions with rapid time-sharing run into a situation 
which changes faster than they do. But yes, I said that such 
a time would always eventually come. You don't produce an argument
that time-sharing could work when we build a computer version of
a brain just by giving an example in which it sometimes works.

           Best wishes and long long life for all,

             Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26039