From: Brian Holtz [Brian.Holtz@sun.com]
Sent: Saturday, September 01, 2001 9:37 AM
To: Brian Holtz
Subject: Re: Hawking, Penrose: Our universe, highly unlikely.

"Paul Filseth" <pgxf@lsil.com> wrote :

> > > What I remember Poundstone saying was that someone had shown it's
> > > possible to build evolving patterns in Conway's Life
> >
> > By showing that Life supports Turing machines, or by some other
> > more direct method?  I've heard of no other such method.
>
>      I think it was a more direct method, but I read it many years ago.
> If the distinction is that important to you, well, it's an excellent
> book

I'll keep an eye out for it, but neither my understanding of
the current state of the Life art nor my recent search of the
web yield any support for your suspicion that it was more direct.
In the meantime, please forgive me if I remain skeptical that
such a method has been demonstrated.

> > if your [original] claim were merely about the theoretical possibility
> > of artificial life in a Turing machine, then it seems disingenuous
> > to instead invoke the simplistic physics of Life
>
> The mere
> fact that there's a distinction you think is vital that strikes me as
> trivial does not imply that I'm being disingenuous.

Since you just said that it was not the Turing machine method you were
thinking of, the "if" clause of my statement should be false, and
no charge of disingenousness stands.

> > and its menagerie of 2-D geometric critters like puffers and
> > floaters, when a Turing-on-Life organism would not be such a 2-D
> > self-contained pattern.
>
> How does that bear in any way on the fact that certain
> configurations of puffers and floaters can implement evolving patterns
> that might eventually develop intelligence and think anthropic
> thoughts?

I think somebody has demonstrated that you can build a Turing machine
using the gears and whatnot of an erector set.  But would you use
an erector set as your poster child for an alternative universe
that could support evolving patterns?  If your claimed support for
evolving patterns came from the property of Turing completeness,
then I found it odd that you didn't just using abstract Turing
machines as your example, but instead chose a Turing machine substrate
that just so happens to also support a menagerie of subsystems that
people talk about in pseudobiological terms. Hence my suspicion of
disingenuousness.

But who knows, maybe you just were assuming your
audience is too sophisticated to be confused by the two differnt
levels of patterns that you were talking about. At any rate,
I'm not interested in debating your possible disingenuousness;
I just wanted to make it clear why I suspected it.

> > Systems like Tierra or
> > Core Wars that maintain a program counter for each organism are
> > much more nurturing universes than are true cellular automata or our
> > physical universe.
>
>      It would be more accurate to say that cellular automata and our
> universe contain multiple environments [...]

"More accurate" here is highly subjective. The distinction between
physics that maintain per-organism data structures and those that
don't is pretty obvious.

> if an organism is running on a Turing machine
> that's running in the Conway universe, the universe doesn't know it.

The relevant universe in that situation is the Turing
machine, not the Life grid (or erector set) on which the Turing
machine is implemented.

--
brian@holtz.org
http://humanknowledge.net