From: Charles Hixson (charleshixsn@earthlink.net)
Date: Thu Oct 10 2002 - 09:40:12 MDT
gts wrote:
>...
>
>If I'm sitting here drinking coffee and writing to the extropian list,
>while my so-called emulation is watching Oprah, then that so-called
>emulation of me is not actually a true emulation of me. It's one thing
>to look like a duck and quack like a duck; it's quite another thing to
>cognize like one.
>
>-gts
>
But the duplicated personality will be essentially the same as you, up
to the last few minutes. And guess what? When you go under a general
anesthetic you loose around the last 15-20 minutes from your memories
anyway. So the differences will all be in what has occurred during the
period since the split. This is about the same as a process forking
(except that the forked process doesn't necessarily take all of the
memories with it, which is assumed here, and a forked process doesn't
necessarily execute the same code, which is assumed here). To make a
simple process fork, you take a snapshot of ram, copy it to a new area,
and then start it running independant of the original process. This can
get messy, so most operating systems cause child processes to have
parents, and when the parent terminates, then so do all of it's
children, but this is a control issue, not an inherrent feature.
To the extent that you are a program running within wetware, that
program would be identical if it were translated into a mode suitable
for alternate hardware, and executed there... if the translation were
good enough. That's the real sticking point. Even translating a
program from Fortran to C is tricky. Now try translating a program from
Python to Ada95. Tricky! The basic assumptions of the languages are
different. Ada is a basically static language. It's Object oriented,
but it's idea of a Object is quite different from the idea that Python
has. Different things are easy. Different things are hard. And that's
just with two languages running on the same hardware. Wetware is a
highly parallel system that appears to operate basically via a
dynamically variable neural-net based pattern-recognition system.
(Logic is strictly an afterthought.) Even clustered hardware, however,
is basically a linear stream of computation (with thread switching, for
various weights of thread), and logic is at the core of the system, with
pattern recognition an afterthought, that still hasn't been very well
thought out. Translation is going to be quite difficult. There are
thought experiments that are "proof of principle" kinds of things, but
we pretty surely won't be doing it in any way that has been widely
discussed so far.
-- -- Charles Hixson Gnu software that is free, The best is yet to be.
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:29 MST