Nano-advantage critique (was Re: No nanotech before AI)

From: David Blenkinsop (
Date: Tue Aug 22 2000 - 22:33:11 MDT

A long time ago by mailing list standards, July 14th, 2000, to be exact,
John Clark wrote:

> Robin Hanson <> Wrote:
> > It seems far from clear to me that it is trivial to take a design for a
> > bridge, throw some nanodust at it and say "Build it for me over there."
> > Assembly and construction use intelligence, just as design does.
> About 5 years ago I listed 6 reasons why a macroscopic assembler, like a
> person on a assembly line, was doing a far more difficult task that requires much
> more brain power than a Nanotechnology assembler ever would. I still think it's true,
> if it wasn't biology wouldn't exist.

While it may seem belated, I'm inspired today to comment on John Clark's
"6 reasons for nanotech to work more easily than macrotech":

> 1) The parts a macroscopic assembler uses would be very expensive, the parts
> that Nanotechnology uses, atoms, are very cheap.

This is a really good point, and a major reason to expect good economy
from nanomanufacturing. However, one has to be just a little bit
cautious about the idea of replicating things from cheap atoms. For
instance, to get a useful feed stock for nanofactories you'd have to
purify some sort of construction material, a material such as an
alcohol, for instance. Uncontrolled impurities could "poison" your
construction machines, bonding with things in such a way as to spoil
your project. Also, people love to talk about mining incredibly diluted
elements (like gold) from sea water. However, if you have to concentrate
things too much, the energy cost may be worth more than the market value
of what you've mined, even with the help of nano-built systems for
getting the energy.

> 2) A macroscopic assembler must use many thousands or millions of different
> types of parts and it must learn how to use all of them. At the most,
> Nanotechnology uses 92 different parts (the elements) but in the real
> world almost everything we know of is made of only about 20 parts, and for
> life about 10.

Before, Dan Fabulich responded that programmed nanoconstruction might
actually involve the assembly of many kind of blocks, or "chunks" at
many different levels. Any real experts out there, game to engage in
relative parts counting exercises? Didn't someone mention that Robert
Freitas had something positive to say about this?

> 3) All the many different parts a macroscopic assembler must use are fragile,
> and fragile in different ways, the machine must learn the proper handling
> techniques for them all or it will destroy the parts before it can use
> them. There is no way you can damage the parts Nanotechnology deals with.

Better to say that atoms can't be damaged by mere handling, but that
radiation damage is a general concern. Also, see my comment on the
*wearing out* of parts, below.

> 4) None of the parts in a macroscopic factory are absolutely identical.
> . . . Atoms have no
> scratches on them to tell them apart.
> 5) Nanotechnology can manipulate matter without ever leaving the digital
> domain. You may have to deal with a rod 27 carbon atoms long, or 28 atoms
> long, but you never have to worry about a rod 27.5601334 atoms long.
> A Macro assembler wouldn't have that luxury when it tried to build
> something with an oak log.

Atoms have the premade, dependable, sizes and characteristics that
Nature provides for them. I'm not sure though, if this is an advantage,
or more of a neutral thing? Dependability is good, other times though,
you might wish you could trim to fit!

> 6) Most of the parts a macroscopic assembler use would have to be very
> complex and the ways they interact with other macroscopic parts would be
> even more complex. Think of the windshield of a car, it interacts poorly
> with the engine block, and even with the windshield frame the interaction
> must be managed with great skill or you'll have a disaster. Nanotechnology
> is like building with Lego blocks, you can build structures of arbitrary
> complexity, yet there are only a few different types of blocks and they
> interact with other blocks (bounds) in only a few different ways.
> John K Clark

This could imply some complications in designing and testing the exact
interactions that you need for effective *self-replicating* factories,
which Drexler and others have theorized as necessary for the most
generally programmable kinds of construction. If there is no "trim to
fit" that could mean a lot of assemblies and sub-assemblies to quickly
build one another into a self replicating system over all? Or, maybe it
*wouldn't* take so much overhead for self-replication? It's hard to
judge in general, with natural living systems being the main existence
proof for this sort of thing. Also, early nanotech might not be
tremendously programmable, nor might it be much like robots building
robots, nor like a cell dividing, even. Instead, the earliest nanotech
might be more of a chemical trick of self-assembly, molecular strands
locking into one another. Living cells make use of something like this
when proteins fold (particular stretches of a protein attract to make
the protein fold a given way).

All of which brings me to the point that nanoparts shouldn't ever wear
out, just as a result of working in a well built nanomachine (this is
most closely related to John Clark's point #3). This is a great
advantage when it comes time to build large machine systems. For
instance, if you tried building a replicating factory from Legos, you
would likely run into the problem of the Legos wearing out at least a
bit, not snapping together so well, maybe wearing out more than a bit
too, as the most often used construction gadgets finally began to make
some headway on the replication problem. With molecular gadgets, not
only would there be no wear, the parts would have much less mass, and
could hence be made to move towards a useful result (like replicating)
much faster. So one thing that's motivated me to comment on this is that
John sort of missed what may be the two biggest reasons for expecting
results from nanotech, namely:

1. "No Wear", and

2. "Speed of assembly of low mass parts" (being relatively speedy for
nano-based parts, compared to virtually anything else, especially given
the "No Wear" property, which is unique to nanotech, as far as I can

Oh, and just as a final thought, or analogy of sorts, consider that
practically every farm has a fancy gadget that you could compare to an
enzyme, namely the baler for packaging grass into winter cattle feed. A
baler gobbles up a bunch of stuff, presses it together in a small space
where it can't get away and out pops a pre-determined shape -- now,
that's a lot like what a chemical enzyme does to atoms! Just a thought,
so long for now.

David Blenkinsop <>

This archive was generated by hypermail 2b29 : Mon Oct 02 2000 - 17:36:26 MDT