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

[Eugen Leitl]

On Mon, Dec 08, 2003 at 08:34:37PM +1100, Brett Paatsch wrote:
> Eugen Leitl wrote:
> 
> > I personally expect crosslinked polymer as structure bulk, not
> > diamond nor graphenes.
> 
> Interesting. I think I can see why crosslinked polymers might appeal
> given disulphide bonds already feature in routine protein chemistry,

No, the simple reason is processivity and energy use. Even if I can
deposit about everything chemically stable, but have to do it
at a MHz..GHz rate by a resonant tooltip I'm burning lots of juice
for a glacially slow deposition. 

With a monomer I just have to squirt it out, and to initiate it.
With a cumulene as linear monomer I can in fact make pretty good
diamond, if I have the F-tipped carbon nanotube as a tool, but
I can also change the polymer deposited (a continuum between diamond
and graphene) by controlling the tip movement, and quench radicals
with a gas jet. It might be not perfect, but it's orders of 
magnitude faster, and gets the job done.

> but I see protein design as pretty complex, requiring a solvent, and

Proteins are very good for purification, autoassembly and part ligation,
all in one step, and a vast existing infrastructure (biotechnology) to
produce them. This is sufficient for assembling cellular architectures,
anything regular, but is lousy for structural material (you won't get 
much better than spider silk, and that's not too thermally stable) and 
for nonregular complex shapes.

You can of course add artificial amino acids to the repertoire, it's
been done already. The problem is that the information required for folding
guidance dilutes your functionality concentration. It takes a lot of blahblah
to create the functional envelope for the enzyme core (the periphery
has also functionality, of course, but I'm simplifying this for the
sake of argument).

> very hard to do computationally (I presume you don't just want to stuff

PFP/iPFP is making nice progress. They've invented a brand new fold the
other day, and forecast the structure very nicely, too. You need basically 
a self-bootstrapping high-precision forcefield, and custom hardware to 
implement it (Blue Gene is pretty close to that) to do it robustly and quickly.

The PFP/iPFP is not a boolean event (now we've got it! The Holy Grail! etc.),
you just get lucky more and more often.

> a volume with any old shape). If diamonoid forms could be produced
> they would seem to have the advantage of being conceptually easier to
> design with. Is your expectation for cross-linked polymer bulk structures

I would forget machine-phase for time being. This is almost irrelevant
for bootstrap. I'm not sure machine-phase is exclusive means of fabbing
for mature nanotechnology either.

> based on  the view that diamondoid would be too difficult or rather that
> designer protein chemistry would be relatively easy?

Right now hacking biology is our strongest tool. The second closest 
being self-assembly by supramolecular chemistry. But here you have
to synthesize everything from scratch, so you're stuck with very
primitive structures, and lousy yield. Manipulative proximal probe
is even more limited, if we're trying to do single-molecule chemistry.

-- 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/20031209/7b43d33c/attachment.bin>