[extropy-chat] Re: Nano-assembler feasibility

Brett Paatsch bpaatsch at bigpond.net.au
Sat Mar 27 23:42:35 UTC 2004


> Re your recent ExI posting on assemblers:
> 
> You seem to be assuming that assemblers must be
> incredibly complex. And that's true--but the complexity 
> doesn't have to be mechanical. A computer-controlled
> assembler can be physically quite simple, and have all the
> complexity downloaded.
> 
> A simple mechanism can be constructed in many different
> ways.  All you need is manipulators with a few degrees of 
> freedom, and enough precision to do the job.  10 N/m 
> stiffness should work for scanning probe surface deposition,
> and diamond should be that stiff even at the nanoscale.

I don't think I was assuming much about complexity at all. I said 
anything from 300 to 300,000 parts. 300 parts would not be 
particularly complex.

My point was there is no specification for a system to produce an 
assembler of any finite number of parts. And therefore there is no
sound basis for either (a) estimating how soon that system could
be produced such that the first assembler could be made, or 
(b) suggesting optimisations or streamlines on that design to bring
the estimated time of arrival pr assembly of that first assembler
forward. 

The assembler (however complex) has to be *physical* to make
physical objects. There are no computers completely without 
hardware. If the computer is not part of the assembler then 
its not a self-replicating system, if it is, the parts for the computing
subsystem have to be included in the parts for building the first
assembler.  

How do you know that "a computer-controlled assembler can
be physically quite simple" without have a specification for it that
includes a set of physical components? Do you have such a 
specification showing a finite set of component parts? 

If not, how is this a statement of anything more than hope and 
belief?

And if it is merely a statement of hope and belief why should 
public resources be directed towards it and away from other
projects that can show a return on investment?

Regards,
Brett Paatsch

PS: I've back-posted this to the ExI list and to Robert because
I thought it might be of interest there as well. 





More information about the extropy-chat mailing list