[ExI] Semiotics and Computability

[Spencer Campbell]

Stathis Papaioannou <stathisp at gmail.com>:
> Software exists separately from hardware in the same way as an
> architectural drawing exists separately from the building it depicts.

When I read this, a billion neurons in unison screeched "NOOO!!"
inside my skull. It was an interesting experience. Now that it's over,
I find myself with a whole litany of much more verbose corrections and
objections. I feel I must share.

One: software does not exist separately from the underlying hardware.
It is epiphenomenal, which in my terminology is synonymous with
virtual and imaginary, meaning that it relies on a more real substrate
to support its existence. When you pull the plug, the hardware only
gets quieter; the software ceases to exist.

Two: a blueprint is hardware, not software, unless it is encoded on a
hard drive and displayed on a monitor. Even in this case, there is no
relationship between the blueprint and the building except in the
imagination of a mind considering the two of them. Demolishing the
building has no effect whatsoever on the blueprint, except perhaps to
increase the number of people giving it wistful looks.

Three: in stark contrast to Gordon's views, I believe that software is
capable of (and is in practice) far more than the mere depiction of
things. A blueprint depicts a building, but a structural simulation
instantiates an actual virtual building that obeys its own laws of
physics (which, we hope, are similar to our own). There is a
surprisingly subtle difference between these two things.

A virtual building has no more relationship to the building it
emulates than does a blueprint of that building. However, unlike the
blueprint, the virtual building really is a building. You could put
little virtual people in it if you wanted (and if you could figure out
how to make virtual people to begin with).

It's a difficult conclusion to convey because I didn't arrive at it
through wholly rational means. As shorthand, I might refer to the two
things as a depiction and a simulation. A depiction can only be
related to what it depicts by a third party, which might be identical
with the thing depicted (as in the case of looking at a picture of
myself). A simulation has no such inherent limitation; it can relate
itself to what it simulates, just as I can relate myself to a person
whom I am doing a rather good impression of.

This has nothing at all to do with whether or not simulator and
simulated are identical. Obviously, they are not. Even a perfect
simulation of a mind is not that mind. However, it would be *a* mind;
it would be able to do everything that a mind can be expected to do,
and be everything a mind could possibly be. Including conscious.

I was careful to say "mind" instead of "brain", here. We could make a
virtual brain that does everything a real brain does, but it would
probably be a waste of processing power: there isn't much sense in
granting the ability to be squished, unless you want specifically to
perform various questionably-ethical stress tests on your imaginary
brain construct.