RE: Implementation

From: Marchal <>
Date: Tue Aug 3 08:46:56 1999

Gilles wrote:

>James, I didn't want you to misunderstand me. I am confident that you or
>others will succeed in building conscious machines in the future, and I am
>"on your side" from this point of view. I disagree with Hans if he thinks
>it will be achieved just by doing an enormous list of possibilities.The
>"highly compressed HLUT" is not a HLUT by definition, it is a smart
>analogic device able to perform tasks that would require an enormous (and
>impossible to build in practice) digital computer, thanks to a correct
>representation of the outer world AND ITSELF (a "self-consistency"
>condition). I think that consciousness lies somewhere in this property,
>i.e. a conscious machine MUST have a vision of itself as correct as that of
>the outer world.

I agree a lot.
The only 'little big nuance' with the *digital* machines is that you
may substitute 'the outer world' with 'the most probable relative
(a continuation is just a computation relative to an 'attained'
 computationnal state).

Gilles, I am not against an external reality, although it would be
difficult for a computationalist to draw a precise line between "internal"
and "external". This comes from the impossibility to know your own level
of substitution and this will contribute to the abandon of the idea of
an absolute external reality...

I prefer to use the objective/subjective dichotomy, and I take
as objective truth a big part of mathematical truth.
What we call the "outer (physical) world" can be seen as the result
(infered by self-referentially correct machines) of the
interference of an infinity (c) of "parallel" deep computations.

I could take as objective truth anything communicated by
correct machines about themselves and their most probable histories.
Actually I take much more, I add to that set the true propositions about
such machines including the one which are never communicated.

Like James, I like to put it in the "everything is subjective" way.
But with comp and precise definition of subjectivity, this gives a way to
sort of verifiable TOE.

Received on Tue Aug 03 1999 - 08:46:56 PDT

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:06 PST