Re: ESSAY: Program length, Omega and Friendliness

From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Wed Feb 22 2006 - 09:39:36 MST


Ben Goertzel wrote:
>
> (The algorithmic information of X is defined, loosely, as the length
> of the shortest program for computing X)
>
> On the other hand, if the starting program is allowed to increase its
> algorithmic information over time via adding new physical compute
> resources, then things get more complicated -- but this doesn't get
> around the basic problem and obvious I've cited above.

*blink blink*

Ben, adding new physical compute resources, without absorbing new bits
from the environment, doesn't change the algorithmic information at all.
  A webcam increases your algorithmic complexity. Adding 10^36 CPUs
doesn't change it at all. The formalism assumes a Turing machine with
unbounded tape.

So does this change, at all, the long argument you delivered about
recursive self-improvement being impossible?

-- 
Eliezer S. Yudkowsky                          http://intelligence.org/
Research Fellow, Singularity Institute for Artificial Intelligence


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT