Re: Emulation vs. Simulation

From: Robert J. Bradbury (
Date: Sat Mar 24 2001 - 23:39:00 MST

On Sat, 24 Mar 2001, Lee Corbin wrote:

> I think that you are visualizing a zombie as a self-contained
> robot of some kind. Dennett, Hal Finney, me, and many others
> maintain that any self-contained unit that behaves at a level
> that is indistinguishable from human, MUST have consciousness,
> feelings, etc. (or more).

Ah, well, as much as I elevate Lee's (and Hal's) discussion points
because of their depth and knowledge, I must simply state in
this case, they are wrong. (Note that I do not state they
are less than wrong).

I'm unclear what a 'self-contained' unit has to do with it.
Clearly this requires significant expansion of definitions.
A human that is can be verified to be human in every respect?
Or a human that can be verified to be human to the degree
we interact with most humans but they are really nanotech
enhanced humans. I.e. they can interact with us but can
dissipate ~10^7-10^9 J of 'waste heat' (by dumping it
into a large quantity of frozen CO2 that is slowly outgassed
for example). Their internal dry-ice sink is reintialized
each evening during 'regeneration'. They aren't 'human' but
they look and walk and talk like 'humans' except their
computational capacity is way beyond ours.

Also, are you imposing an EM 'sterile' environment on your
'self-contained' unit (i.e. they cannot be tele-operated)?

The reasons for the above clarifications points out that if
you cannot guarantee that your 'zombie' is operating with the
computational hardware that normal humans are based on,
I believe the argument completely fails. At some level
of technological sophistication, I can statistically 'emulate'
human behaviors (i.e. given these inputs produce thse outputs)
and the humans would not have a clue that they were dealing
with a zombie vs. a real person. You *DO NOT* need
consciousness or feelings. You simply need to replicate
the most common responses in those situations. That is
something that is entirely programmable.

> In other words, in the classic definition, there cannot be any
> such thing as a zombie. So no observer can be a zombie, because
> they cannot exist. It's either an emulation --- in which case
> it has everything that we do, or it's a simulation --- in which
> case it's merely a portrayal, a moving picture, as it were.

You will have to split hairs here -- is a statistical
response behavioral unit an 'emulation'? If you say yes,
then I would argue that it does not have everything we do.

I would argue a statistical response behavioral unit is
much closer to a 'portrayal'. I would also argue that
few humans have the necessary tools to distinguish between
an emulation and a portrayal.


This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:59:43 MDT