X-Message-Number: 26139
Date: Fri, 06 May 2005 00:36:55 -0700
From: Mike Perry <>
Subject: Information Paradigm and Such
References: <>

John de Rivaz writes:

>But David Deutsch also wrote that it is not possible to simulate zero-G
>flying in an electromechanical flight simulator, even though other
>sensations of flight can be simulated.

Deutsch's proposed reality generator would overcome the limitations you are 
referring to.


>The simulation of sleep poses a philosophical dilemma. If the subject
>actually goes to sleep in the simulated world, surely he goes to sleep in
>the real world as well, therefore it is not a simulation.

It appears you are saying that the "simulation" is somehow invalidated if 
what it does is "real"--I don't see it that way. "Simulation" to me is 
inclusive of the possibility of it being the real thing, the main point 
being that you can in fact achieve certain effects by a different, maybe 
unexpected route.


>Death can be simulated by a "game over - you lose" notice, and the program
>can even refuse to accept the same player with a different identity, but
>annihilation is just that, there is no way to simulate it. It is real or it
>doesn't happen.

Again, if the simulation is actually the real thing, no biggie.

Thomas Donaldson writes:

>1. Quantum computers are no longer Turing computers by any
>    reasonable interpretation. I hope we can build useful ones,
>    but that also looks like a hard problem.

Actually, strictly speaking there aren't any Turing (classical) computers 
at all--it's a quantum world we live in--though our present computers are 
good approximations of truly classical devices.

>2. I haven't read Deutsch on this question, but I would certainly
>    agree that in some sense (very loosely speaking) we are machines,
>    but not Turing machines. We're even highly parallel.

It is worth pointing out that parallelism, per se, does not convey power to 
compute any function that cannot be computed on a sequential machine, 
though in practical terms it can make a big difference.

>It does
>    not take quantum mechanics to simulate a human brain.

I think so too, but we haven't simulated a human brain yet, so we'll find out.

>It WILL
>    take processors which can do things no present processors
>    can do, like change their connections (remember those connections
>    are our memories), grow new ones, and get more processors
>    when needed.

It is also not ruled out that certain technological innovations yet to be 
developed will prove superior to some of those the brain uses. The brain 
has the interesting ability to grow new connections, as you've often 
pointed out. No doubt this works in its favor, but might there be something 
better that could be developed artificially, some sort of software 
simulation perhaps? Naturally I don't know the answer, but by way of a 
possibly feeble analogy I'm reminded of the early days of computing when a 
machine was reprogrammed by plugging and unplugging wires on its backside 
("growing--and breaking--connections"). Here the stored program concept 
soon proved much superior. Might something like this eventually occur in 
our efforts to develop superior AIs?

>3. We will not have truly simulated a brain until our manufactured
>    brain can actually deal not with a simulated world but with
>    the real world that surrounds us. [snip] An artificial
>    brain is useless unless it can deal with the real world. It's
>    even useless to itself: who wants to live in a doll's house?

Well, as I've often tried to emphasize before, part of the reason (maybe 
the greater part) for considering the possibility of simulations is not to 
suggest how to build practical devices but to shed light on difficult 
philosophical issues. This too has some importance even though we don't 
want to overlook the practical side either.

Best wishes and happy ever-after to all,

Mike Perry

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26139