Re: CA rules and external random oracles

From: Hal <HalRuhl.domain.name.hidden>
Date: Sun, 26 Jan 2003 14:25:40 -0500

In Stephen Wolfram's book we find examples of rules that start out with a
complex behavior but then settle down in various ways to something more
routine. At an extreme they land on an attractor.

If one was to randomly input to such systems information from an external
random oracle [true noise] so as to locally reestablish the general
character of the initial conditions [flipping cells here and there] then it
seems to me [I currently have no modeling tools set up] that the system
would maintain a higher level of complex behavior - such as avoiding the
attractor - but not depart from the nature of the rule in question - few
exploding cows if the rule in question makes exploding cows
scarce. [Something of this sort may be in the book but I have not yet been
able to read it all.]

If one then allows the externally sourced noise to accumulate by adding it
as groups of new cells between existing cells [an accelerating expansion of
the "space" of the universe in question if the noise density per unit of
the "space" does not decrease] then I believe one has most of the basis for
a model of universes like the one we seem to be in. To finish the basis
the noise must also be able to target the rules.

I prefer that the global system we are attempting to invoke be one that has
no meta information. If one sort of universe should have a frequency of
occurrence or quantity of example different from that of another sort then
that would be meta information. I see no reason or need for this and in my
view the concept of true noise defeats such lopsidedness directly.

In order to maintain a given lopsidedness all universes would have to be
free of true noise [meta information]. If not then the noise if it can
target the data and the rules [why down select from this?] would it seems
continually reconfigure the lopsidedness by transforming noise susceptible
universes into all types.

On the other hand allowing all to be true noise susceptible [data and
rules], each by at least one of all possible ports requires no sustaining
meta information - no exclusion or other selection is needed. Actually
there would be a flow of universes between noise port types since the right
dose of true noise would change the character of the rules of a universe.

Hal
   
Received on Sun Jan 26 2003 - 16:39:52 PST