2010/1/4 Nick Prince <m...@dtech.fsnet.co.uk>:
Thank you Stathis
This has helped move me on a bit. “The hardwareless computer” has been
giving me some real problems. Let me replay my understanding of what
you said back just to check it is on the right lines.
As a possible example of one of these “lurking computations” we could
consider the one which begins with no-thing and think of the null set
as made of it phi ={ } and then associating it with the number 0. Then
imagine the set { phi} associating it with 1, then { phi,{phi }}
associating this with 2, then { phi, { phi} , { ,{phi }} },
associating it with 3 etc. Hence we get an infinite sequence of
abstract (platonic) entities which can conjure up (compute) the
natural numbers and the implied successor function simply from the
abstract (platonic) notion of a set and an association rule (also a
platonic relation). More and more structure can be built up until - as
you say - the entire structure of the computation contained in the
mapping can be envisioned. Now although no external observers might be
able to access these computations, the computations might just create
conscious observers – bootstrapped into existence by the special class
of computations which these (internal) observers (if they believed in
comp) would naturally consider as non trivial. As you say the entire
structure of the mapping which describes the computation is a platonic
object too – hence the world comes from nothing and computation.
Have I got this roughly right? I would be grateful for any critical
comments from you, Bruno (or anyone).
Yes, but a critic could still say that no conscious observer could be
conjured up by a computation unless the computation is physically
implemented. At least at first glance that seems to be the case: the
brain is required for consciousness, since if the brain is destroyed
consciousness is destroyed. And if the mind is generated by a computer
program, it would be normal to think that if the computer is
destroyed, so is the mind, although the program in Platonia remains
unaffected even if the entire universe blows up. These are the common
sense objections. So the question is, is physical implementation
necessary for consciousness, and what does it actually mean to
physically implement a program?
Suppose we agree that it is necessary to physically implement a
program in order to get the consciousness. Physical implementation
then involves, essentially, causing a machine to go through a sequence
of causally connected configurations such that the configurations and
the state transition rules match up with the abstract program. There
is a mapping from the abstract program to the machine so that the
engineer, programmer and end user know what's going on. But "write 1
and then move the head to the left" could be represented in an
infinite number of ways. If a man walks down the street chewing gum,
that could represent "write 1 then move the head to the left", while
if he stood still humming "Jingle Bells" that would have represented
"write 0 then move the head to the right". Moreover the mapping does
not have to be consistent from moment to moment: chewing gum could
mean "0" on Fridays and "1" on other days. There is no reason why a
computer could not be designed to function in such an inconsistent
way, other than the practical necessity of keeping track of what's
going on, which is necessary if the computer is to be of any use to
anyone. But if we don't care about its usefulness to an outside
observer we could say that any abstract computation maps to any
physical process: a random physical process, a repetitive physical
process, or a single physical state. The man walking down the street
chewing gum over the course of a second could be seen as representing
the one thousand steps of a Turing machine adding two numbers
together, although of course it wouldn't be of any use to anyone
interested in the result of the calculation. You can see no doubt that
if you accept the argument so far the physical process is irrelevant,
and all of the computation, such as it is, consists in the abstract
machine and the mapping, which are timeless platonic objects. Arguable
the mapping is also irrelevant, since there are an infinite number of
possible mappings for an infinite number of possible physical
processes. The only thing that seems to make a difference is the
abstract machine or program itself. The program "runs" necessarily,
even in the absence of a physical universe, and it only need run on
physical hardware in order to interact with the environment at the
level of the hardware (and of course, this hardware may itself be part
of the virtual world generated in Platonia).
I think you give an excellent explication of the problem, Stathis.
However, one thing about it that still worries me is the role of time.
You say the mapping need not be consistent even moment to moment, and
yet the mapping is a timeless Platonic object. To be a timeless object
the the moments need some timeless representation. In Bruno's theory
time arises from the computational sequence. But in the mapping, time
is just a relation of similarity (closest continuation) of states. So
three states which when ordered by closest continuation are XYZ may
have been computed in the order XZY. So I find myself seeing the
hardwareless computer as a reductio against consciousness=computation
thesis and support for Peter's view that ur-stuff and contingency are
fundamental.
