X-Message-Number: 12987
From: 
Date: Thu, 23 Dec 1999 16:30:17 EST
Subject: simulation by clockwork

Still another attempt to illuminate some problems with computer simulation:

In our thought experiments, we need to get away from electronic computers, 
which tend to impress and bedazzle. Instead, think of strictly mechanical 
computers, like a Babbage engine or a Turing tape or a glorified 
abacus--clockwork. Its parts are cogwheels and rods and ratchets and 
driveshafts and disks and  rollers and pistons and styluses etc. It might be 
driven by steam, or by electric motors powered by sunlight, but best to think 
of it as spring driven, like an old-fashioned watch. You just set it, wind it 
up, then stand back and observe.

Physically impossible? Maybe not, if on the nano scale. In any case, we are 
only concerned with the principles involved.

Now, what does a digital computer do? It generates numbers. It starts out 
with two stores--the program and the initial data, the latter representing 
the initial state of the simulated system, such as a person and his 
environment. Over time, the computer generates successive sets of numbers 
corresponding to calculated successive states of the simulated system.

It is a serial computer. It only generates one new number at a time. Almost 
all the numbers it generates are intermediate steps in calculations or 
instructions, not new parameters of the simulated system. Even when a new 
number pertains to a new parameter of the simulated system, it usually 
pertains to only PART of that parameter, since any state or aspect of a state 
usually requires several numbers. 

Most of the time, nothing is happening--i.e., no meaningful number is 
changing; the machinery is just getting ready to shift another number. For 
example, maybe a pendulum is swinging--but that does not correspond to a 
change in the simulated system. The "simulation" changes only when one of the 
"meaningful" numbers changes, and even then only after the whole set of 
numbers for a particular parameter has changed. (If we need a specific 
example, a position vector cannot change until all of its new coordinates 
have been calculated, and these are calculated one at a time.)

It ought to give you a queer turn also to remember that the assignment of 
labels to "meaningful" numbers is arbitrary to the programmer. How does the 
clockwork gadget (or the simulated subject) "know" which symbols (numbers) 
represent "real" parameters of a system and which represent something else, 
such as intermediate quantities in a calculation or instruction? 

One could argue--yes, we have been over this before--that the logic space is 
so dense (in some sense, so to speak) that only one possibility exists. The 
analogy would be an artificial language, built up in such a way that any 
intelligent alien, no matter how different from us, could not misinterpret 
it. But the analogy seems very, very fragile and unreliable. 

Even if the relationships could somehow be shown to be unique, we would still 
be left with the claim that feelings and thoughts and actions are (or can be) 
nothing but relationships between sets of numbers. Don't stake your life on 
that.

Robert Ettinger
Cryonics Institute
Immortalist Society
http://www.cryonics.org

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=12987