From: Ramez Naam (Exchange) (ramezn@EXCHANGE.MICROSOFT.com)
Date: Tue Oct 14 1997 - 15:08:54 MDT
Thanks for your extremely illuminating post, John. A brief comment
below on my use of "chaotic" as applied to design of nanosystems:
In my post I suggested that design of a nanosystem runs against issues
with computation of the final state of a chaotic system. John correctly
points out that a complex system needn't be a chaotic one, and parallels
the construction of nanosystems with biological processes that create
our bodies:
> From: John K Clark [SMTP:johnkc@well.com]
> Just because a process is complex does not prove it is chaotic.
Proteins
> start out as a linear sequence of amino acids that can't do much, but
> then,
> in a manner too complex for any computer to simulate, they fold up
into
> intricate shapes and that gives then their power. Despite their
> staggering
> complexity these shapes are repeatable. If the shape of proteins were
> exquisite sensitivity to tiny variations in the environment then they
> would
> never take the same shape twice and life would be impossible.
Yes. My concern is not the repeatability so much as our ability to
predict what the final shape will be based on only our knowledge of the
original configuration. This is an issue because as a nanosystem
designer, I /start/ with a desire for a particular output, and must work
backwards to a specific design for the nanites that will construct the
object.
In a broad sense what I'm saying is that the transformation from
genotype to phenotype is a chaotic one. Ie, a tiny change in genotype
may result in a much larger change in phenotype. To bridge back to
nanotech: the desired output of a nanotech process is the phenotype, the
design(s) of the individual assemblers is the genotype. How do I get
from A to B?
As you point out, I'm not necessarily correct in calling this
transformation "chaotic". It is indeed /complex/, in that it clearly
obeys some higher-level rules. Those rules, however, are part of a
theory that we do not yet have, and which I suspect to be much more
complex than we envision.
(Note that in biotech we've had the advantage of being able to sample
existing creatures, examining their genotype and comparing to the
creature's phenotype. In nanotech we'll be starting from scratch, and
it seems that without radical developments in complexity theory (or
quantum computation?) risky trial-and-error may be the only effective
means of design.)
mez
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:45:02 MST