From: Billy Brown (bbrown@conemsco.com)
Date: Mon Jan 11 1999 - 07:18:43 MST
Entropyfoe@aol.com wrote:
> Billy Brown wrote,
> >....I agree. However, you can generate a scenario where
> >that would happen if
> >you mix the right assumptions together. What we need is:
>
> >1) automated engineering is very easy
> >2) nanotech design is manageable
> >3) computers are very fast
> >4) nanotech is very hard to actually implement for some reason
>
> We have 1) and 3), at least to some degree now. At least progress has
> been dramatic in recent years, but of course will increase
> exponentially more in eht future. Design tools that allow design of chips
> with 0.18-0.25 micron features is massively complex, and by necessity
> highly automated. Massive computers are being used to make more powerful
> computers. This is happening
> now. I just bought a P II with 7.5 million transistors. The design
> complexity is massive, geometric data, logic, simulations,
> electromagnetic models.
By 'automated engineering' I mean an AI that can turn "build be a widget
that does thus-and-such' into an actual design. Silicon compilers and
similar systems took decades to evolve - what we need for extreme nanotech
design-ahead is better systems that are developed much faster.
By 'fast computers' I mean machines capable of doing fast MD sims on systems
of millions of atoms - obviously well beyond current capabilities. However,
I do agree that we'll meet this condition pretty soon.
> 2) we are short on this one and 4) may be actually true. So
> we are getting close to the described scenario.
IMO, #2 is too close to call so far - no one has had a chance to try
anything more complex than basic mechanical components. I don't think #4 is
especially likely - pushing atoms is complex, but no more so than any other
emerging technology. There is no obvious, huge barrier to overcome that
could hold up implementation for decades.
> >That creates a situation where you have designs for very
> >advanced devices that no one can build.
>
> At any stage of technololgy you can design beyony your
> production technology. Leonardo drew tanks. I am biased, I am a HW guy,
> design is SW, you have good ideas, you can make something. In
fabrication,
> you are up against physics, entropy, mother nature, and her dog Murphy.
Could Leonardo have designed a space shuttle? Nanotech implies a major
complexity barrier - advanced nanotech routinely uses systems far more
complex than anything we have build in the real world. That's why we need
some kind of automated engineering to get beyond baby steps with it.
> >The first group to get an assembler can use it to
> >implement those designs, giving them a huge instant jump in power..
> >Ordinarily we would expect other groups to duplicate the feat
> >and catch up, but nanotech also lets you make faster computers. So, the
> >leading power whips up a huge mob of supercomputers and sets them to work
> >designing even better hardware. Their computers will be on a faster
> >improvement curve than anyone else's, so no one can catch up until they
> >hit the limits of what is possible..
>
> This exact scenario took place in the world
> geopolitical/technological sphere
> 1940-1990. The US dominated technology, computers, and with government
> investment, technology was applied in the most advanced ways
> to military hardware..
The 'automated engineering' component is crucial here, because it would mean
that a sudden increase in computer power would give you a sudden increase in
design ability. That in turn would let you implement more complex nanotech
faster, including more complex computers that run even faster.
Billy Brown, MCSE+I
bbrown@conemsco.com
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:02:46 MST