[extropy-chat] Re: Nano-assembler feasibility

Chris Phoenix cphoenix at CRNano.org
Tue Apr 6 20:28:11 UTC 2004


Robert Bradbury wrote:
 > On Mon, 5 Apr 2004, Chris Phoenix wrote:
 >
 >>> Biology uses active transport at many scales.
 >
 > Hmmmm... I would not say that is strictly true Chris.
 > Much of biology is based on diffusion of one kind or another.

I didn't say biology uses *only* active transport!  G's, you think I 
don't know about diffusion?

Even within cells, biology uses lots of active transport.  For example, 
dragging neurotransmitters along axons.  At intermediate scales, lymph 
is pumped with local muscle contractions.

 > Yes there are molecular targeting tags in several cases
 > such as targeting to the mitochondria, targeting to the
 > nucleus, etc.  But much of the time biology is using the
 > latent heat in the environment to drive transport mechanisms.

Have some respect for thermodynamics... The transport is lubricated or 
facilitated by latent heat.  It is *driven* by concentration gradients.

 > If you extend this you get into the entire question of
 > reversible computing.  You start to get into the things
 > that Landauer, Bennett and others worked on.  One
 > of their points was that if you wait long enough you
 > may be able to compute for "free".  I've never seen
 > an analysis of it but one might also be able to "assemble"
 > for "free".

I understand ATP synthase is reversible--effectively, it's 100% 
efficient.  (Which is an impressive feat, and may account for why it's 
so highly conserved.)  And Drexler sketches an analysis of reversible 
chemistry for mechanosynthesis and power conversion in Nanosystems.

In theory, one ought to be able to "assemble" for "free" simply by 
putting your reactions on a "frictionless" conveyor belt that takes them 
through the reaction space smoothly over a distance (e.g. by gradually 
bringing them closer to a reactant).  The force produced or required by 
the conveyor belt can be converted directly to/from electrical energy. 
If there's no abrupt state transitions (a function of the slope of the 
PES of the reaction vs. the stiffness of the system), then the whole 
thing is reversible, and lossless except for dynamic friction.

 > So you have an interesting issue with regard to making
 > nanotech faster than biotech.  Why bother?

Because otherwise Eugen will criticize it for being slow.  That's the 
origin of this thread of the conversation.

Of course, Eugen also criticized it for being inefficient.  And it's 
true that the slower you run, the more efficient you get.  But running 
at biological speeds (doubling mass in ~15 minutes), our best 
calculations indicate that a system should be within an order of 
magnitude as efficient as biology, and may be much better than biology. 
  And if it dissipates heat with irreversible state transitions (like 
forgetting bits, or doing chemistry with abrupt energy level changes), 
that heat can't be recovered by going more slowly.

Bottom line: as far as we can calculate, MNT is easily good enough for 
human purposes.  Anyone claiming problems with efficiency or scalability 
had better show an error in Nanosystems or in my Nanofactory paper.

Chris

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




More information about the extropy-chat mailing list