[extropy-chat] Smalley, Drexler and the monster in Lake Michigan

[Eugen Leitl]

On Fri, Dec 05, 2003 at 09:43:52PM -0800, Hal Finney wrote:

> I had a similar reaction.  Here is a posting I was working on about the
> debate at http://pubs.acs.org/cen/coverstory/8148/8148counterpoint.html:

I have an unfinished post sitting in my postponed queue, which 
(quite rudely) interleaves in the Drexler/Smalley exchange.
 
> One of the problems I often see in debates about nanotechnology is an
> attempt to shift the burden of proof onto the other side.  (This is
> actually a widespread problem in debates on all topics, but I think
> it is particularly inappropriate with nanotech.)  In the online debate
> between Drexler and Smalley, it seems sometimes that the issue is, "Are
> the fat-finger and/or sticky-finger problems inherent to all possible
> nanotech assembler designs?"  This puts the burden on the critic to show
> that there are no possible designs which could evade a particular problem,
> a virtually impossible task.

The problem with Smalley that his critique is limited to a few off the
cuff statements he's unwilling or unable to expand. We've had heard
these same points before (in fact, I used very similiar points once,
from a pure synthetic chemist's point of view), whereas the machine-phase
approach has meanwhile resulted in considerable number of publications. 
 
> But that's not fair.  The big issue here is obvious: will Drexlerian
> nanotech work?  Will we have the kind of revolutionary developments
> described in Engines of Creation and Drexler's other books?  Will we
> have self-replicating machines which can replace most of the world's
> industrial capacity in a manner of, what, a few years? months? days?
> 
> These are extraordinary claims, and many experts in physics and chemistry
> say that they are not credible.  The burden of proof here is obvious.
> It is on the nanotech supporter, not the critic.  He is the one making
> amazing predictions.  He is the one who must support his claims by
> providing evidence in the form of technological plans and designs
> sufficient to make a strong case that this will all be possible.

The Bush administration has just approved a major spending package
for R&D in nanotechnology, machine-phase being notably exempt from
the plan. Now this is not very nice.

What would be sufficient as a burden of proof? We already have
evidence of machine-phase in manipulative proximal probe. There are
no simulators large and precise enough to contain a full design.

In a sense, the only hard proof will be a working assembler itself,
or something with a self-rep closure so close to unity it doesn't
require a leap of faith to go over unity, and to scale down the
design.

Where will money for this come from? Smalley is blocking R&D
in that development quite efficiently by using essentially arguments
from authority. This is not a laid-back ivory tower discussion.
This is political science, with R&D budget and long-term policy
being at stake.
 
> He can't just wave his hands and say, if one thing doesn't work, we'll
> try something else.  He can't point to living things as an existence

Actually, yes, because there are several approaches in design space.
Just because one aspect has been invalidated (I personally expect
crosslinked polymer as structure bulk, not diamond nor graphenes;
and possibly sorting of precursors from a stochastical synthetic batch,
along with microfluidics functionalization, self-assembly *and* machine
phase as the most viable approach) it doesn't mean the whole thing
is a no-go.

> proof (because Drexlerian nanotech's revolutionary properties go far
> beyond anything possible with biology).  He needs to come up with

Not really, enzymes (and enzyme assemblies) plus active transport
within the cell do fall within mechanosynthesis domain. The reactive
site in an enzyme does resemble processes occuring at the tip, albeit
minus some aspects (substrate recognition and envelopment,
bond weakening before breakage). 

> enough specifics to make his case.  The burden of proof is on him.
> Supporters of Drexlerian nanotech must take on this burden squarely and
> refrain from demands that critics prove that the technology is impossible.

It's still perfectly valid to call bogus arguments that, I hope.
 
> Along these lines, let me ask a question.  In his open letter, Drexler
> complains about Smalley's statement that assemblers will suffer from
> the "fat finger" and "sticky finger" problems.  He writes, "I find this

Fat fingers is a perfectly valid point, though stated in dumbed-down
language. There's steric hindrance from the side of plane substrate,
and tool tips coming from other space quadrants (to be stiff, they need to
be bulky). How many tool tips need to be simultaneously present, though?
It's not obvious we need more than one, or two. I don't like cycles,
even if it's driven at resonance, but you don't really need cycles. A SWNT
can continuously extrude an allene strand, and polymerize this linear
monomer with a minimal-amplitude tooltip oscillation. This is a continuous, 
one-tip deposition. Where are the fat fingers here? This is the same
thing as bucky-tipped AFM head, and you know the resolution of that.
You can buy these tools, and they resolve deep crevices at atomic resolution.

Now sticky fingers are a red herring, because no one is proposing
to handle naked reactive monoatomics. It's perfectly possible to drag
and position e.g. Xe on Ni at cryogenic UHV conditions precisely because
Xe sticks to the STM tip. Cycled deposition doesn't deposit atoms
directly, only formally so. You don't need a methyl radical to methylate
Hg with Acetyl CoA. It's a cycle activate-deposit-regenerate, and it's
all perfectly vanilla chemistry in-between, except it's all dry and
extremely controlled.

> puzzling because, like enzymes and ribosomes, proposed assemblers neither
> have nor need these 'Smalley fingers'."
> 
> So I will ask, what "proposed assemblers"?  What is Drexler referring
> to, a proposal for an assembler that doesn't have these problems?

I presume he refers to mechanosynthetic reaction set, and these reactions
indeed do not require Smalley fingers. Apart from Merkle's stuff, you'll 
see interesting hits for mechanosynthesis on Google.

Now these are theoretic cycles, but they do use classic chemistry
assumptions and calculations to back them up. I agree fundamentally
that the set of these reactions the critical part of the proposed
classical (you don't need a true assembler for a number of applications) 
assembler.

> My understanding is that we lack any designs for self-replicating
> assemblers that would be sufficiently detailed to know that they will
> work and not need "fingers".  If Drexler has an assembler proposal that

Self-assembly doesn't need any fingers, and Drexler/Merkle stuff is limited
to a few tooltips, so I wouldn't get too worked up about what Smalley
says.

> answers this question, I'd appreciate a pointer to it.

Ultimatively, the best proof is to try designing your own. For the
most part, it's currently all about modelling and software engineering.

The bootstrap issue is not orthogonal to the design, but fund allocation
is critically dependant on a killer demo, and if it's only in the virtual
dry dock.

-- Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBM: 48.07078, 11.61144            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
http://moleculardevices.org         http://nanomachines.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20031206/4dca73b8/attachment.bin>