[extropy-chat] Re: Nano-assembler feasibility

[Robert J. Bradbury]

Side note to Chris -- it seems unreasonable to fault Brett for perhaps
being poorly informed with respect to nanoassemblers as he has been working
on stem cells and the politics thereof for the last few years.
And there are several paths to MNT where preserving our butts
for long enough (quite possibly perhaps making stem cell technology
a requirement) may be essential to get us to the point where we can
consider something like vasculoid technology.

The second point that I might make is that given Brett's questions
it points out how *really* poorly informed people who would be
viewed as highly educated are about nanotech.  And I would offer
the idea that we are going to need hundreds of thousands if not
millions of engineers who are well informed about nanotech to
make it work.

Brett pointed out:

> Which specification?

This is in large part in Chapter 13 of Nanosystems.  The essential
elements are worked out on pages 398-410.


> It's not *me* that talking timeframes - its me asking which
> specification is the basis for your timeframes? ;-)

Ok, this is relatively easy to speak to.  It took Ralph
and Eric a couple of months to design the Fine Motion Controller (FMC)
which could be considered the tip of an assembler arm as described
in Nanosystems.  The FMC contains 2600 atoms.  An assembler arm has
been variously estimated from 4 to 8 million atoms.  The assembler arm
is highly regular (unlike the FMC which is somewhat like a Stewart
Platform at the molecular scale).  We had software that was written
perhaps a decade or more ago (by Ralph Merkle, Will Ware and Geoff Leach in
various distinct efforts) that could help to automate the design of
"regular" nanoscale structures.

Based on the regularity of the nanoassembler arm and a scale up from
2600 atoms to 4-8 million atoms -- a team of graduate students (augmented
by some semi-intelligent software) should be able to manage a complete
atomic scale design.  (My gut level evaluation -- you have to decide
how much you can trust me.)

*Now* once one has a complete atomic scale design the difficult part
will be how to build it.

> And those who find Eldorado will be rich - there just one l-ee-ttle
> problem.  Where's Eldorado?  :-)

Oh its there -- and the interesting part is that there is not simply
a single city.  There are multiple paths to nanotech and companies
ranging from IBM to CALMEC to Zettacore are slowly starting to figure
that out.

> On the other hand those who look for Eldorado and don't
> find it will be shorter on time and money for other things.

Some will win and some will lose.  Do a google on Amdahl and Trilogy.
This may help:
  http://en.wikipedia.org/wiki/Gene_Amdahl

The idea behind Trilogy was great (wafer based computers).  The problem
is that it simply could not be executed with the technology available then
and perhaps even with the technology available now.  In the process it burned
$200 million -- perhaps one of the greatest crash and burn exercises
in the history of Silicon Valley -- and it was doing it at a time when
$200 million was real money.


> May (above) is a word that can be replaced with may-not. The
> point is why should they choose to try? What basis for confidence
> do you offer them that this is a good place to spend time, effort
> and money?

Ah... because ones personal survival is at stake.  Unless one is
a hard-core cryonics fan (and signed up and in a location where the
political winds may not shift) there is a significant advantage
to not becoming "dead".  If one is looking at the technologies
that provide an extension to lifespan one (a) has more time and
(b) has more money (wealth is a no brainer problem if one can simply
live long enough -- its called compound interest).

> So does Zyvex have a specification for an assembler at the
> (microscale even) or not?

Go look at their current product line.  It is most probably
at the microscale currently but presentations previously made
by Ralph suggest to me that it should scale down.  It becomes
a complex cost vs. benefit tradeoff of how difficult it is
to manufacture the machines relative to what the demand is
for manipulating structures at a specific size scale.

Now it may be useful to distinguish between an assembler
and a manipulator.  The products being produced by Zyvex
may tend more towards the second.  *But* they are both
important.  I can envision using chemistry to assemble
the components of the Fine Motion Controller -- but it
may be very difficult to produce one if one does not have
a means of manipulated directed assembly to put one together.
One is dealing with the problem where one does direct molecular
assembly and where one is using self-assembly -- but one problem
with self-assembly is that it may not happen in our lifetimes.

> If so, they can probably right their own check for R&D funding.

Oh no...  The people who write the checks are well aware of
Trilogy and other similar crash and burn experiences -- it
is their business to know them.  And if they good at their
jobs they are always asking -- "Is this a crash and burn
scenario?".

> For what its worth I hope someone can show that an assembler
> is possible at some scale. But wishing won't make it so. Mine
> or anyone else's.

It already is possible -- things from the various DNA polymerases
to the ribosome demonstrate that directed nanoassembly is possible.
What is more interesting is to explore the limits of general atomic
assembly.  The biological examples are cases where the machines
evolved to fit the parts (DNA bases, amino acids, etc.).  We
do not have any good ideas with respect to what the limits may be.
Are we going to have to design an assembler specific to each type
of nanoscale part we would like to put together?  (This is not
unheard of in the mechanical engineering world -- one designs jigs
that assemble one and only one kind of part.)  Or can assemblers
that can deal with a variety of parts be designed?  (The biological
example of the proteosome which can disassemble a variety of proteins
comes to mind).

Hope my comments which got a lot longer than I intended are helpful.

Robert