MNT development (was Re: [extropy-chat] Nanotech educations)

[Chris Phoenix]

Eugen Leitl <eugen at leitl.org> wrote:
> CAD is the wrong-ended approach. 3d structure editors exist, and by now scale
> into some 10^5 atoms. With current hardware there is no problem to render
> some 10^9 atoms in realtime.

I'm not suggesting that CAD should be used to develop molecular 
machines.  But many products of MNT will have molar numbers of atoms in 
multi-scale heterogeneous structure implementing complicated multi-scale 
function, and there's no bookkeeping system of any kind today that can 
deal with that.

A few years ago, I wasn't bemoaning the lack of such software.  But 
today, I think that software may well require the most calendar time to 
prepare.

> I've found the machine-phase people completely unresponsive when asked to
> contribute to a CML specification suitable for representing
> Drexler/Merkle/Freitas type of devices. There's an emerging standard, and our
> needs are not accounted for.
> Is this lame or what?
 > I haven't come across any single so-called nanotechnologist in all my 
years
 > on CCL, Gamess list & Co. Ditto biostructure.

Serious charges.  I'll investigate.

> If you want to know where the bottleneck is: it's where it's been all along.
> 
> We need a validated library of mechanosynthesis reactions.

This is one of the things we need.  But there's no reason other than 
political that requires us to develop all the technologies one by one in 
sequence.

Would a validated library of reactions be useful in convincing people...
that eutactic mechanosynthetic exponential manufacturing...
would result in system performance as calculated in Nanosystems?

If not, then we need something else.  Because without that 
understanding, there's not much motivation to develop the reaction library.

> After we have that, you can use these atomic anabolic/catabolic steps +
> deposition constraints to create the MNT equivalent of a silicon compiler.

Yes.  But even before this is done, we can be working on design methods 
for molecules, machines, systems, and products.  We may have to change 
the designs once we see what chemistry we have to work with.  It may not 
be vacuum scanning-probe deposition at all.  And it may not make 
diamond.  But many people today still think you can't do engineering at 
that scale at all.  I don't think MNT can be developed efficiently under 
those conditions.

Anyone who wants MNT developed ASAP should be supporting parallel work 
on: 1) the chemical library; 2) the design software (a HUGE task); 3) 
bootstrapping/enabling technologies (largely underway, could benefit 
from more targeted purpose); 4) high-level architectural work 
(fundamental, and almost completely unaddressed).

> I will people here would talk Jaguar, Gaussiand and Gamess (and that package
> Zyvex is using) instead of AutoCAD.

I want to see some software that integrates chemical packages, or at 
least chemical results, with CAD software that can handle quadrillions 
of components.

> Anyone now go read http://discuss.foresight.org/~pcm/nanocad/index.html

There are hundreds of posts there.  I usually make a point of reading 
URLs, but you'll have to narrow it down some. :-)

Chris

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