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

Eugen Leitl eugen at leitl.org
Tue Dec 30 10:56:38 UTC 2003


I'm flushing out a few old replies from the postponed message queue before I
leave the list for a while. I'm sorry if quality is lacking, there is not
time for more.

On Sun, Dec 07, 2003 at 07:36:56PM -0800, Hal Finney wrote:

> And I also object to some of the tactics being used in this spin
> doctoring.  Smalley is being misrepresented and taken out of context.
> From the Foresight press release: "In the current C&E News exchange,
> Smalley now agrees that assemblers (without impossible "magic fingers")
> could use something like enzymes or ribosomes as tools for doing precise
> chemistry."  This makes it sound like Smalley has conceded that Drexler's
> ideas for nanotech could work.  But in fact, Smalley was attempting
> to show that to the extent that nanotech needs to rely on things like
> enzymes and ribosomes, it will share the limitations of those systems,
> such as reliance on water and the inability to make strong objects,
> hence that nanotech could not meet its claimed goals.

The argument as presented is muddled, and as such worthless for both sides.
Biology has a makeup, and a modus operandi. Both are orthogonal. It is the latter that is
illustrative to outline the differences between classic organic synthesis and
high degree of control in biosynthesis. This is a continuous axis ranging
from low to very high degree of control in machine-phase.

I'm reasonably sure that this distinction is completely lost on the average
synthetic chemist. I don't know whether it's lost on Smalley, he never said
enough to tell, and I'm not going to start guessing. 

I'm not sure it's worth having a public discussion where issues are so
complex one's mainly operating in opinion space. Results shut up critics,
eventually, not talk.
 
> And about this water business, which has also been trumpeted as showing
> that Smalley is ignorant of chemistry (a claim which sounds absurdly
> arrogant to the unbiased spectator).  Maybe there are some enzymes
> which can work in polar solvents or even in gas or vacuum.  But ribosomes

Polar solvents yes, some of them, especially enzymes optimized for a different
solvate, not gas or vacuum, not that.

> don't, and so Smalley's overall point still holds.  The chemistry of water
> dominates virtually everything about proteins, enzymes (which are a subset
> of proteins), and ribosomes (which are built equally of RNA and protein).

Water assembles shape, it is in most cases nothing but a nuisance in the
reaction center. Machine-phase as postulated doesn't autoassemble bottom-up,
so it doesn't need solvate (at least during operation) to maintain the
operation geometry.
 
> So what does Drexler really mean when he says that nanotech will work
> like enzymes?  As Smalley asks, "do you think it is really possible to

Enzymes use control, solvated orgchem doesn't. The more control, the higher
the yield of a specific product. 

> do enzymelike chemistry of arbitrary complexity with only dry surfaces
> and a vacuum?"  I think the answer Drexler would give is basically yes,
> albeit not of "arbitrary" complexity (no system could achieve that goal).

If you've got a good mechanosynthetic reaction library and high-precision control the space of
structures you can build is vastly bigger than anything biology could muster.
Arguably, it could build anything stable, but I guess we can agree that a
tiny subset of all possible structures is enough if there's full self-rep
closure goodness hidden in there.

> He does believe that he can achieve and surpass the power of enzyme-like
> systems using dry surfaces and a vacuum.  He refers to this somewhat

These assumptions are reasonable.

> obliquely in his reply: "Bound groups adjacent to reactive groups can
> provide tailored environments that reproduce familiar effects of solvation
> and catalysis."

Enzymes cut and create bonds, the result is what is important. The manner by
which it is done is quite irrelevant as long as cheap high processivity
constraint is satisfied.
 
> I've read Nanosystems, and it's always bothered me that the technology
> described there is so different from what most nanotech fans are
> familiar with.  Drexler describes the famous robot arm; you can see a
> picture here, http://www.zyvex.com/nanotech/images/fig13.14left.jpg.
> He goes into a lot of detail about how it would work and how strong it
> would be for mechanosynthesis.
> 
> But he doesn't use it!  The proposed manufacturing system that he
> describes in some detail only uses giant robot arms in a final assembly
> stage, to manipulate relatively large, pre-built blocks that are a
> cubic micron in size, far larger than the arm above (which is only 0.1
> micron long).  Instead, the actual molecular manipulation and assembly
> is done by means of a "mill".
> 
> This is a system which holds molecular fragments on miniature pallets
> attached to 4 nm wide assembly belts. The belts run past each other such
> that the pallets are pressed together, and possibly rotated or manipulated
> in some way, so that the payload molecules react, transferring one or
> more atoms from one pallet's payload to the other.
> 
> These mills would gradually build up larger and larger pieces, which
> would eventually be transferred to appropriately scaled-up mill systems.
> These would use the same principles, now working with pieces with
> perhaps thousands of atoms, attaching them together in various ways.
> Eventuallly we get up to the micron size, which will have billions of
> atoms, and at that point he uses almost-macro-scale robot arms to start
> attaching these "bricks" together.

It's an illustration, no one is going to build that as is.
 
> This molecular mill system provides a possible context to understand
> Drexler's comparison of nanotech assembly to enzymes.  Enzymes bring
> reactants together in a carefully controlled environment.  The active
> site is surrounded by atoms which provide the exact pattern of positive
> and negative charge, hydrophobic and hydrophilic regions, acidic or base
> pH, necessary to promote the desired reaction.  This is what Smalley

Do you need all of this just to make proteins? Synthetic chemistry has no
trouble making pretty long peptids without needing a kidney, whether of man,
or dog. Enzymes don't zap educts with electric potential, manipulative
proximal probe does.

> means when he talks about biologically catalyzed chemistry occuring
> as the interaction of a dozen or more atoms, which would seemingly be
> impossible using robot arms.

It is obviously unnecessary to replay the exact processes within the cubic nm, or
so, in order to do chemistry. As long as the reaction outcome is the same,
no one cares how exactly you did it. Whether you synthesized your peptide
enzymatically, synthetically, or by pushing around activated monomers absorbed to a
flat substrate, who cares?
 
> But it is conceivable that a mill could provide a similar degree of
> control over the external environment as what we get in an enzymatic
> active site.  The pallets that hold the reactants can be designed
> to have desired patterns of charge, ionization and electron binding.
> When we bring the reactants together, surrounded by the two pallets that
> held them, the environment surrounding the reactants can therefore be
> controlled to a considerable degree.  The reactants can even be held
> in pouches or grooves within the pallets to provide for full 360 degree
> control over the local environment.

You're latching upon a design that's essentially a red herring.
 
> This is what I think Drexler means in the quote above, "Bound groups
> adjacent to reactive groups can provide tailored environments that
> reproduce familiar effects of solvation and catalysis."  The bound groups
> are part of the pallets that hold the reactants; solvation, that is,
> the effects of water molecules, can be mimicked by means of patterns of
> charge; and catalysis achieved by carefully adjusting the properties of
> the atoms lining the pallets so as to provide the necessary lowering of
> energy barriers.

Or you could just drag around stuff on HOPG in cryogenic conditions. Or use
SWNT ducts. Or radical moisture. Whatever.
 
> I think it is rather difficult to interpret Drexler's comment in terms
> of robot arms.  They have a limited flexibility in terms of tailoring
> the environment in the near vicinity of the reaction.  It would be
> extremely complicated to recreate the effects of solvation (the widespread
> presence of water molecules shielding and modifying electrical effects)
> and catalysis using robot arms, raising fat-finger problems.

You're recreating the vis vitalis argument here a bit. You have to abstract
the product from the reaction from the reaction conditions. The product is
what counts, not how you access it.
 
> The point of this rather lengthy digression is that Drexler and other
> nanotech proponents are not doing a good job of explaining their design
> concepts.  This is in part why they are so commonly misunderstood.

This is not something which can be unambiguously explained in plain language.
This is something you can write specialist papers no synthetic chemist or
biologist is going to understand, without learning a couple of whole new
fields.

> And it seems almost willful.  Drexler is trying to explain exactly
> how his system will work: "In machine-phase chemistry, conveyors and
> positioners (not solvents and thermal motion) bring reactants together."
> But this terminology is complete generic and unspecific: conveyors and
> positioners.  It applies to mills, but could also apply to robot arms,
> where reactants are conveyed to the tip and then positioned to be applied
> to the work piece.

Of course, because he's describing a principle, and not a particular
implementation. 
 
> Didn't anyone reading this exchange wonder about Drexler's insistence
> that "fingers" are not needed for his assembly process?  What is a robot
> arm if not a finger, in this context?  Do people realize that Drexler
> has moved away from arm-based assembly, perhaps due to some of the very
> objections that Smalley has (re-) discovered?

The finger issue is what happens when experts start dumbing down the
discussion that everybody can understand it. The issue is that no two
tooltips can occupy the same space simultaneously, and I fail to see how this
is a major constraint on anything. (Unless you start counting every atom
within the reaction site as finger, which is a vis vitalis/holistic
argument).
 
> Of course, I'm not even 100% sure that my interpretation is right, either.
> I've explained above how I think his comments relate to the designs
> in Nanosystems.  But maybe I'm wrong, maybe he does still plan to use
> robot arms.
> 
> Why should I have to guess?  And why should Smalley?  This continued
> evasiveness and refusal to plainly specify a design strategy
> forces nanotech critics to extrapolate their own understanding and
> interpretations.  And once this happens, the nanotech proponents sit back
> and smugly call "strawman".  Drexler and Smalley are talking past each
> other, because Drexler refuses to plainly state how his manufacturing
> system will work, contenting himself with telling Smalley that all his
> guesses are wrong.

The issue boils down to modelling and validating a set of mechanosynthetic
reactions (according to Smalley, these don't exist, however, a 30 sec session
with Google will show everybody that they do exist) experimentally.
 
> This isn't a game of 20 Questions.  If nanotech were the dominant
> paradigm, this lack of specificity might be acceptable.  But when you
> are on the outside looking in, it will not succeed.  All you're going
> to do is make people confused and angry.
> 
> I share Smalley's frustration when he writes, "it would be helpful to
> all of us who take the nanobot assembler idea of 'Engines of Creation'
> seriously if you would tell us more about this nonaqueous enzymelike
> chemistry."  Rather than lobbying and spinning the debate, I'd suggest
> that nanotech proponents work harder at fleshing out and clearly
> describing their proposals.  Give your critics something to criticize, and
> at least the debates won't be as empty as the Smalley-Drexler exchange.

Let's see whether http://foresight.org/stage2/project1A.html
will produce results, so we can stop handwaving, and just tell people to
RTFM.

-- 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/20031230/b3efdffe/attachment.bin>


More information about the extropy-chat mailing list