Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!nrl-cmf!ames!ucbcad!ucbvax!sdr.slb.COM!SCHOLTES%ASC
From: SCHOLTES%ASC@sdr.slb.COM ("SCHOLTES%ASC@SDR.SLB.COM")
Newsgroups: comp.society.futures
Subject: nanotechnology
Message-ID: <8711171658.AA07280@bu-cs.BU.EDU>
Date: Tue, 17-Nov-87 10:10:00 EST
Article-I.D.: bu-cs.8711171658.AA07280
Posted: Tue Nov 17 10:10:00 1987
Date-Received: Thu, 19-Nov-87 20:58:02 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The ARPA Internet
Lines: 21


I've started reading Drexler's _Engines_Of_Creation_, and had a fairly
intriguing idea.  (Disclaimer-- I haven't finished it, so maybe he
beat me to this one).  Drexler points out that mechanically instrumented
computers on the microscopic scale will be very fast, and possibly more
natural to build than electronic ones.  So, imagine a Von Neumann computer,
with the memory cells laid out on some substrate.  The process of 
converting assembly language to machine language could be done in parallel
by special-purpose nano-agents swarming all over the substrate like
spiders; one kind to recognize ADD instructions, another for JMP, and so
forth.  (Imposing a Von Neumann architecture on an inherently parallel
technology might be specious -- I haven't thought much about it yet).
More complicated loop-optimizing beasties would need to establish 
monitors or other synchronization protocols.  All kinds of interesting
problems spring up; how do you download/upload? How do you know when
they are finished?  Can you implement traditional electronic computer
fundamentals like remote memory addressing, dereferencing, etc.?  What
about hybrid electronic/nano-mechanical systems?    Nano-nano.

---mike scholtes
   Schlumberger Well Systems, Austin Systems Center