Tuesday, February 25, 2003, 10:12:07 AM, Ben Goertzel wrote:

BG> I view complexity as part of a web of concepts that also, centrally,
BG> includes "pattern"

BG> Roughly, an entity is complex if there are a whole lot of patterns in it
BG> (statically or dynamically)

BG> On the other hand, what is a pattern?  A pattern in X is a program P that
BG> produces X, but has significantly less "basic complexity" than X.

BG> What is this "basic complexity"?  One approach is to define it as program
BG> length: that's algorithmic information theory, in essence...

BG> This mathematical approach is part of the conceptual foundations of
BG> Novamente.  It highlights the very real subjectivity of the notion of
BG> "complexity."  Note that I defined complexity in terms of pattern and then
BG> defined pattern in terms of "basic complexity" ;-)  You can define "basic
BG> complexity" as "complexity relative to a certain observer" or "the minimum
BG> length program that computes P assuming a given knowledge base K as
BG> background knowledge."  Then you have defined complexity in a way that is
BG> specifically relative to an observer or a set of knowledge.

As I'm sure you're aware, but others may not be, some interesting
things have been proved about K and P.

The main result is in a sense analogous to G�del's proof, but far
further-reaching.  That is that you cannot prove you have found the
smallest program that computes P unless your axioms (knowledge base,
K) is of complexity P itself (for sufficiently large P).

See e.g.:
  http://www.cs.auckland.ac.nz/CDMTCS/chaitin/ieee74.html
and other works by Chaitin at
  http://www.cs.auckland.ac.nz/CDMTCS/chaitin/
  
At some point I suspect that his proofs will be used to prove P!=NP but
that may be a ways off.

--
Cliff

-------
To unsubscribe, change your address, or temporarily deactivate your subscription, 
please go to http://v2.listbox.com/member/[EMAIL PROTECTED]

Reply via email to