Prof. Smalley at Rice U thinks...
John J Lee
On 3 Feb 1997 Norman....@vixen.cso.uiuc.edu wrote:
> I just got an email from prof. Smalley at Rice, the guy who discovered
> bucky tubes and spheres, and he says he doesn't think the concept of a
> universal assembler is possible in our universe. As for myself, the idea
[...]
I would be very interested to read anyone's criticisms of this idea
(of the universal assembler), as I have so far been unable to locate any
on the Web. Of course, Richard Smalley's criticisms would be particularly
interesting considering his work in the field.
Thanks for any pointers/views.
John
Advice: the smallest current coin.
Ambrose Bierce, The Devil's Dictionary
Dan Clemmensen
David Stoner wrote:
>
> Norman....@vixen.cso.uiuc.edu wrote:
> : I wondering, just how important is the concept of a universal assembler?
>
> : Would Drexlerian Nanotechnology be severly handicapped if such an
> : assembler could never be built?
>
> In my view, the concept is crucial. The anticipated cost reductions
> don't materialize without huge numbers of assemblers that are essentially
> free because they can replicate themselves. It is not essential for
> all purposes that they be free-roaming, or even able to move about
> in a special environment, but I think it would be unwise to assume
> that free-roaming assemblers are impossible.
>
In my view (and, based on reading Nanosystems, in Drexler's
view) the way to build things using nanotech is with a
nanotech factory. assemblers are too slow, comparatively
speaking. However, the way to build the nanotech factory is
to use assemblers: the factory itself is custom designed to
generate whatever product you need. Most likely, the factory
is built up from assembly-line modules which feed
progressively-larger subassemblies toward toward the root
of an assembly tree. As I understand the concept, an efficient
design for the utlimate product will likely include components
that have a large number if identical sub-components, and which
are therefore amenable to assembly-line production. The final
assembly, and probably several intermediates, are best done
by "universal assemblers" or general-purpose robots of the
appropriate scale. Note that this "factory" can be reconfigured
reranging the subassembly lines. An individual new subassembly
line may be needed to build part of a new product. Nano-scale
universal assemblers are likely to be the easiest way to build
the new subassembly lines.
If we step back a level, the resulting self-reconfiguring
"factory" is a macro-scale universal assembler.
My conclusion is that the universal assembler is crucial to
building other special-purpose nanites. The ability to produce
arbitrary special-purpose nanites, which I believe is the
fundamental enabling technology for fully-distributed independent
production of goods.
James Logajan
Norman....@vixen.cso.uiuc.edu wrote:
: I just got an email from prof. Smalley at Rice, the guy who discovered
: universal assembler is possible in our universe.
It should be noted that Smalley made this point quite publicly at the last
Foresight technical conference. I recall that Drexler agreed, but tried to
clarify that he didn't believe anyone was proposing such a thing. (I think
Smalley is using the phrase "universal assembler" to mean a single "hand" or
"tip" that could perform _any_ localized chemical transformation.)
The phrase "universal assembler" is fraught with semantic and philosophical
difficulties. Consider the human hand & arm: is it a "universal assembler?"
Now consider the video monitor you are reading this on (excepting the paper
it may have been printed on.) One could make a plausible sounding argument
why no human hand could _directly_ construct something as sophisticated as
a video monitor. Hands are simply too clumsy and too large. If we didn't
already know better, one could erroneously conclude that all manner of
modern (and not so modern!) contrivances were impossible.
Most everyone agrees that a major distinguishing feature of humans is that
they are tool users and makers. The hand & arm isn't a universal assembler,
but it does appear that we've made it a nearly universal _tool holder_.
In the narrow technical sense that Smalley may be using the phrase, he may
be correct, but it almost certainly is not a relevant objection.
: I wondering, just how important is the concept of a universal assembler?
Not nearly as important as the concept of a universal tool holder.
: Would Drexlerian Nanotechnology be severly handicapped if such an
: assembler could never be built?
Not as I understand the term to be used in this context.
: I don't think this universal assembler concept is the way to go --
: certainly not in the near term. What nanotech needs now is factory type
: robotic equipment inside SFMs and STMs that can work at the nanoscopic
: level. Nanofactories, not nanoassemblers.
For enabling technologies, I agree somewhat. But as has been pointed out by
others, the numbers involved in going from nano to macro pretty much demand
self-replicating machines. SPMs only get you past the first step.
Julian
> > Would Drexlerian Nanotechnology be severly handicapped if such an
> > assembler could never be built?
>
> Yes. :)
>
> > I don't think this universal assembler concept is the way to go --
> > certainly not in the near term. What nanotech needs now is factory type
> > robotic equipment inside SFMs and STMs that can work at the nanoscopic
> > level. Nanofactories, not nanoassemblers.
>
> I disagree--the only way of making nanotechnology feasible is to reduce
> manufacturing to the nanoscale level--attempts to create nanoscale
> structures using macroscale tools are basically a waste of time and
> energy. Nanoassemblers (universal assemblers) are necessary because they
> will be able to cheaply, quickly, and easily assemble nanostructures;
> macroscale tools will probably never reach the same level of efficiency
> and cost that assemblers could reach.
I think it works this way (speaking as a wholly uninformed layman). Suppose
you want to make something useful like a car using nano-fabrication. By
definition, nano-fabrication means constructing the car atom by atom (which
gives amazing strength etc). The problem is that something as big as car
contains a lot of atoms. No doubt it would be theoretically possible to
construct it atom by atom using macroscopic techniques (in the same way as
IBM wrote IBM atom by atom). However, since a car has (say) 10 to the 20
atoms in it, even if your fabrication machine was very very quick, it would
take longer than the life of the universe to make your car. Even Mercedes
can deliver quicker than that.
This means that the only way to make complicated macroscopic scale objects
using nano-fabrication is to construct a huge number of nanoscale
factories, each of which can build a tiny bit of the car.
However, you would still need (say) 10 to the 15 nano-factories. In order
to get that many, they (or their precursers) will have to be reproduced
exponentially for a larger number of generations.
That (I think) is why nano-assemblers are essential. Without them, I don't
see how we will ever be able to make complex people-sized objects atom by
atom.
Just my humble contribution.
David Goldstone
ps - I'm obviously excluding "simple" macroscopic scale objects from the
above which can be made by direct chemical means (e.g.buckey tubes).
pps - when I say "reproduce" I don't mean biologically. Just that each
factory must be able to make at least 2 factories,each of which make at
least 2 etc etc so that eventually you have enough of them to make a car.
ppps - I guess that you might need to do it in stages (eg Grandparent
assemblers which make parent assemblers which make car- assemblers)
now my brain hurts .....
John Sidles
>On 3 Feb 1997 Norman....@vixen.cso.uiuc.edu wrote:
>
>> I just got an email from prof. Smalley at Rice, the guy who discovered
>> bucky tubes and spheres, and he says he doesn't think the concept of a
>[...]
>
Prof. Smalley expressed the same opinion at the 1996 Welch
Foundation Conference: Chemistry at the Nanoscale. But he
didn't assert that *all* assemblers were infeasible ... only
Drexlerian assemblers. Smalley's objection was cogent: there is
a big gap between Drexlerian theory and the experimental
reality, such that that the assembler theory community was not
governed by normal checks and balances of the scientific
community. The result: the literature is full of designs which
most likely are infeasible, but there is no good experimental
way to check, hence no easy way to tell good assemblers from bad,
without a lot of unproductive argumentation.
(Some people say that string theory is in the same boat ... but
the string people at least deliver mathematical theorems whose
proofs can be checked, and which are thus are interesting to
mathematicians, even if they have no connection to experiment).
Many scientists subscribe to the view that nanotechnology is
(1) already feasible, (2) already in existence as a working
technology, (3) already the foundation of a multibillion-dollar
industry, which is (4) already changing people's lives, and
(5) already creating jobs for young people, in a process that
is (6) already rapidly accelerating, and (7) already posing
moral dilemmas. In other words, the nanotechnology vision
is *already* a reality! Of course, the nanotechnology involved
is based on recombinant DNA technology.
I myself would not assert that the Drexlerian vision is infeasible.
But the diamondoid people are kinda in the same position as the
magnetic bubble memory people in the 1970's ... if you can't bring a
product to market, you'll risk being steamrollered by an alternative
and more viable vision of nanotechnology.
Robert Freitas
> I was wondering, just how important is the concept of a
> universal assembler?
> Would Drexlerian Nanotechnology be severly handicapped if such
> an assembler could never be built?
The answer is NO.
David Stoner wrote:
> In my view, the concept is crucial. The anticipated cost
> reductions don't materialize without huge numbers of
> assemblers that are essentially free because they can
> replicate themselves. It is not essential for all purposes
> that they be free-roaming, or even able to move about in a
> special environment, but I think it would be unwise to assume
> that free-roaming assemblers are impossible.
I basically agree with most of Stoner's comments, but the
concept of the "self-replicating system" is subtly distinct from
the concept of the "universal assembler." The former is
probably necessary for practical Drexlerian Nanotechnology; the
latter is not.
All that is required for general-purpose nanotech manufacturing
of mole-quantities of useful nanoproduct is a well-ordered input
feedstock and a factory system which displays full process
closure. Such a general-purpose nanomanufacturing system may
consist of hundreds of different kinds of nanomachine subsystems
-- the nanoscale conceptual equivalents of the drills, lathes,
mills, band saws, conveyors, parts bins, etc. that populate the
typical macroscale workshop. No one of these nanomachine
subsystems need come even CLOSE to having a general self-
replicative capability. Moved outside of the factory
environment, such a subsystem in isolation would quickly fail.
But by working together cooperatively as a SYSTEM, these
special-purpose nanomachine subsystems can perform all necessary
fabrication and assembly tasks necessary to replicate all of the
component subsystems comprising the factory.
The process closure specification thus need only extend to
factory subsystems AND desired nanoproducts. This is likely to
be a vastly smaller set than the set of all naturally occurring
materials that a hypothetical free-range replicator might be
able to use as its input substrate, combined with the set of
every combinatorially possible molecular arrangement of all
possible atoms. In other words, the replicative complexity of a
nanofactory capable of manufacturing diverse useful nanoproducts
is probably vastly smaller than the replicative complexity
required of a free-range truly universal replicator.
Thus, for example, a nanosystem able to replicate itself (and
anything else) from materials found in grass, or air, or even a
randomized dissociated atomic plasma is simply unnecessary if
the chemists can readily supply our nanofactories with pre-made
mole-quantities of all requisite nanoparts (e.g. rods, gears,
wheels, ratchets, buckytubes, etc.) produced by conventional
supramolecular chemical techniques.
Robert A. Freitas Jr.
Dan Clemmensen
John Sidles wrote:
>
> I myself would not assert that the Drexlerian vision is infeasible.
> But the diamondoid people are kinda in the same position as the
> magnetic bubble memory people in the 1970's ... if you can't bring a
> product to market, you'll risk being steamrollered by an alternative
> and more viable vision of nanotechnology.
I'm not sure, but I don't think Drexler is precisely a "diamondiod
person." My understanding of "Nanosystems" is that the
feasibility of diamondoid manufacture is used as a simple proof of
principle, and not as a prediction of the most likely path to the
achievement of MNT. Diamondoid was chosen because it is simple. All of
the computations in the first half of the book are therefore fairly
straightfowrard to understand, and it's clear that any competent
physical chemist could identify any serious error that Drexler may have
made. Thus the first half of the book makes it quit clear that
diamondoid mamufacture is feasible. The second half of the book then
investigates some of the things that could be made with this
material.
I view the book as a formal existance proof of a method of achieving
MNT. Drexler is saying "look: here's a way to do this that I can
prove will work." He's not saying that this the the best way, or the
easiest way, or the way that is most likely to occur. This is equivalent
to demonstrating that electronic computers can be made with mechanical
relays.Relays are simpler to analyze from first principles than are
semiconductors, so you don'tm have to spend any energy arguing about
dopants, band gaps, bias voltage, etc.
Drexler and others have of course continued to analyze diamondiod as
if it will actually be the most feasible path to MNT, but to my mind
it's more important as an existance proof.
Will Ware
John Sidles (sid...@u.washington.edu) wrote:
: >On 3 Feb 1997 Norman....@vixen.cso.uiuc.edu wrote:
: >> I just got an email from prof. Smalley at Rice, the guy who discovered
: >> universal assembler is possible in our universe. As for myself, the idea
: ... the diamondoid people are kinda in the same position as the
: magnetic bubble memory people in the 1970's ... if you can't bring a
: product to market, you'll risk being steamrollered by an alternative
: and more viable vision of nanotechnology.
Remember the AI flop of the eighties. Premature efforts to bring a future
technology to market can do horrible things to it. People are left with the
impression that it's a failure, instead of the understanding that it will
take a long time no matter how much money is thrown at it by clueless MBAs.
Nobody promised diamondoid products any time in the next couple decades.
Also, nobody promised that it _had_ to be diamondoid, or that it _had_ to
look just like what Drexler described. He was painting broad strokes and
offering feasibility arguments. He would probably welcome alternative,
equally well-thought-out proposals, just to fill out the space of scenarios.
Indeed, anybody offering such proposals would be doing a valuable public
service.
As has already been pointed out, the interesting things described in his
writings are primarily tied not to the assumption that an assembler (or a
set of assembly tools) would be "universal", but to their ability to replicate
without individual supervision. That's what gives the exponential growth in
assembler populations that produces the economic upheavals.
It would be nice if the first-generation assembly tools could build all
products of human interest, but maybe that won't happen until the fourth or
the tenth generation of assembly tools. But there won't be any second, third,
or fourth generations of tools if the first generation can't self-replicate.
--
-------------------------------------------------------------
Will Ware <ww...@world.std.com> web <http://world.std.com/~wware/>
PGP fingerprint 45A8 722C D149 10CC F0CF 48FB 93BF 7289