From: Dan Clemmensen (dgc@shirenet.com)
Date: Sat Sep 28 1996 - 10:33:47 MDT
Damien Broderick wrote:
>
> Hi folks
>
> ...hitting both lists here, hope that's not impolite... (adrift in the
> datasea...) (aargh)
>
> The debate between Lyle and other more optimistic nanoists seems to me to
> depend on the question of `tacit knowledge', evolutionarily-acquired tricks
> for dealing with the world - life regarded as GAs at the end of a three
> billion year planet-sized run. How are these little buggers supposed to be
> programmed to do their job of building a starship in the back-yard pool?
> It's no good pointing to mitochondria and ribosomes and crying `existence
> proof!', unless you can specify or emulate all the subtle algorithms they
> encode. And even assuming a Very Big Smart AI Thing which isn't yet
> autonomous to number-crunch the specs for yr starship, how is this template
> vectored out to the trillions of nano fabricators doing the back-stroke in
> the milky pool of goop? Radio? Infra-red? Batons (or quipu, as the Incas
> would have done) handed physically back and forth? Is there a precise
> cascade of fabricators, and chemo-gradients, and all that standard
> make-a-meat-body stuff? (Not in a goop pool, I'd have thought.)
>
> Probably all this is explained in NANOSYSTEMS, but shamefully my university
> library doesn't have a copy... nor does my local academic book store, yet.
>
> Best, Damien
My discussions with Robin and Lyle were not intended to cover Starships,
or
even pipelines. My point about economic meltdown remains valid if we
consider
only the replacment of production of consumer goods with MNT
manufacture.
Yes, MNT designs for this class of goods are not yet available. (We can
safely
call this an understatement...) Yes, design of an MNT-produced product
must
include the design of its associated manufacturing process.
NANOSYSTEMS does in fact address the problem of communicating the
process
instructions to the assemblers. Your joke about batons is not far off.
I got my copy of the book at the local Barnes & Noble store, but you can
probably get it at any large bookstore. The book takes an extremely
conservative approach to MNT engineering at the detail level. It's
intended
to present a convincing argument that MNT violates no physical laws and
that
a particular class of structures is feasible. The only part of the
argument
that I didn't find convincing was the thermodynamics chapter. It's
probably
also completely valid, but I'm not competent to judge. The rest of the
physics, reliability theory, statistics, computer science, and chemistry
is deliberately kept simple enough to allow an intelligent layman with
some
background in science to form a valid opinion. This simplicity is
possible
because of extremely conservative assumptions and error bounds.
IMO, the major problem will be in the design of the products and
systems,
just as Lyle claims. We differ in our respective perceptions of the
degree of
difficulty, and the analogies from which we reach these perspectives. I
feel
that the design process for the class of consumer goods in question will
look a lot like the design of a computer system's hardware and software.
Current computer hardware design is done using CAD tools, and software
design
is done with softwar development tools. These tool suites will evolve
into
tool suites for MNT design. Most exsting "durable goods" can be designed
hierarchically, with the individual pieces designed independently and a
MNT "gluing" strategy used to build the final product. Individual pieces
such as sheet-metal parts can be designed very rapidly from prototype
"library subroutines". Larger subassemblies (pump, motors, etc.) can be
designed using specialized software packages, or an off-the-shelf design
can be used. However, I suspect that most people will simply use
freely-available complete designs for most goods. The designs will be
available because most designers are creative people who enjoy design
and
who like to see their designs used.
Lyle believes that it takes a car company to build a car, and that, by
analogy
it will take the equivalent of the human talents of the employees of
that
company, as they apply to the building of cars, for an MNT "factory" to
build the same car. I concur, but I differ wil Lyle in that I don't
think that
very much of the human activity in a car company actually goes into
building a
car. The vast majority of the activity goes into material acquisition,
accounting,
management, sales and marketing, plant maintanence, and other peripheral
activities. the only piece of the human activity within a car company
that I
think we need but don't have with MNT, is the design, which I discussed
above.
Lyle's other analogy is with a biologist attempting to re-engineer a
tree.
IMO this analogy is inappropriate because a biologist does not have the
level of uderstanding of a tree that we have of MNT. Once we have the
first MNT assembler, we know how to proceed. There is no equivalent for
trees.
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:35:46 MST