[extropy-chat] Re: fwd -- Robert Bradbury post

[Chris Phoenix]

Damien Broderick forwarded me a post that
Robert Bradbury wrote:
> I think people are missing a couple of important issues here.  Right now
> a Drexler type assembler arm probably is *effectively impossible*.
> [And one can comfortably ignore nanorobots.]  The reasons are poorly
> developed methods and very high costs.

Robert, I don't think anyone is saying that we could build such a thing
today.  That's not the important issue, and using words like
"impossible" and "ignore" obscures the issues that are important.  Try
these questions instead:

Actually doing it:

Would general-purpose diamondoid manufacturing be worth trillions in
2015?  (My answer: Yes.  It would be decades ahead of other technology
road maps, and could produce a wide range of products.)

Could we build it in ten years at a cost of multiple billions?  (My
answer: Yes, given ten years of targeted funding, we could probably find
and refine some technology to do mechanochemistry.)

Learning about it:

How much would it cost today to get more information about timelines and
capabilities?  (Not much.  A few theoretical physicists and chemists to
review Nanosystems.  A few more chemists and computers to do preliminary
mechanochemical simulation.  Some experimental physicists, mechanical
engineers, and a polymath or two to look for cheap ways to bootstrap.)

How important are these questions?  (With our current information, they
look very important.)


You mentioned several expensive bootstrap pathways, but you did not
consider several others.

Dip-pen Nanolithography:  Has been done with multiple "inks", multiple
tips, arrays of tips, re-registration, 2.5-nm precision.  BTW, parallel
AFMs from the Millipede project have already been applied to DPN. 
"We're up to a 10,000 pen array now, where you have 10,000 individual
pens that can grab 10 000 different chemical agents from ink wells." 
http://www.materialstoday.com/pdfs_6_5/Gould.pdf

Atom holography:  I haven't heard anything about it recently, but a few
years ago they were able to deposit complex patterns of atoms by
shooting them through a programmable grid.  Apparently this was not
merely masking, but interference.  

Bose-Einstein Condensates: One more way of handling very small groups of
atoms with high precision.  Last I heard, they could make them travel
along switchable wires.  I don't know if they've got single-atom control
yet, but it wouldn't surprise me.

Sub-wavelength optics: At last count, I knew at least four ways of
breaking the lambda/2 barrier.  Of course most of them haven't been
applied to fabrication.  I mention them as an indication that nm-scale
fabrication has not been fully explored and might turn out to be
significantly easier than the methods you considered.

You mention the cost of designing the parts to be fabricated.  If
diamondoid mechanochemistry works as Drexler, Merkle, and Freitas
predict, then the design cost should be very low compared with today's
synthetic chemistry designs.  The set of reactions to choose from would
be far smaller, and the results effectively digital.  You can't assume
that the design cost per atom in a nanodevice will be any higher than
the design cost per transistor in a CPU.  

Also note that for most products of interest, a few good designs could
be re-used and re-combined to make a wide range of products.  Levels of
abstraction have been extremely useful in computers (both hardware and
software), and if chemistry can be made digital, levels of abstraction
will work just as well in designing MNT-built products.  

Chris

--
Chris Phoenix                                  cphoenix at CRNano.org
Director of Research
Center for Responsible Nanotechnology          http://CRNano.org