Re: No Singularity?

Robert J. Bradbury (bradbury@ilr.genebee.msu.su)
Tue, 16 Nov 1999 14:03 EST

"M. E. Smith" <mesmith@rocketmail.com> wrote

> Thank you, Geoff, for bringing this to fascinating
> site to my attention. Before now, I've never seen a
> *good* critique of "Engines of Creation". Now I have.

Well, I suspect "good" is in the eye of the beholder. I'd be much more comfortable with people who are actually well versed in the various technologies required saying what they view can or cannot be done. The semiconductor industry is littered with claims, by people *well versed* in the technology, about "hitting the wall". We haven't yet. If the "experts" on a technology can't do a decent analysis, I'd say non-experts are in a much worse position.

> Like you, I have reservations about a sentence here
> and there, but overall Lyle Burkhead's article
> "Nanotechnology without Genies" is well-reasoned.

I'll simply take the comments from his first page:

| 1. Nanotechnology is commensurable with other technologies.

Only in so far as other technologies are "designed". Nanotechnology is the only technology that can assemble nanotechnology factories that can assemble nanotechnology devices. Biotechnology is the only other similar technology (and it really is a form of nanotechnology). All other technologies (say microelectronics), use *other* technologies (say machining, grinding, element purification, crystal growth, etc.) to enable the production of microelectronics. For them to be "commensurable", you can only discuss technologies which are self-replicating.

| 2. Automated systems always exist in a larger context which | is not automated.

So what? Ultimately we exist in an "unautomated" universe (at least we believe that). He is right, to some degree that "they won't design themselves", they do have to be bootstrapped. But humans do *not* do molecular modeling, the computers and the software to do the modeling have been have been "enabled" by human designers. It is now the case that much of the work in microelectronics, drug design, etc. is done by the computer with the human just checking the results.

| 3. Artificial intelligence is irrelevant.

Tell that to Boeing and look at their documented reduction in engineering design time & reworks on the 777 (designed entirely on the computer). There isn't much "AI" in the software systems they are using, but what there is makes a difference. This claim has no validity unless the "I" under discussion is something radically different from what most humans do.

| 4. Making things with atomic positioners will be at least
| as expensive as making them with biotechnology or bulk
| technology.

I don't simply disagree with this statement, it is clearly wrong. The cost of something depends on the cost of the raw materials, the energy required for assembly and the speed of assembly. Nanotech & Biotech tend to be better than bulk assembly because they use less raw materials than bulk technology. Biotech has a big advantage in terms of "energy required for assembly" because nature has been making the reactions cost efficient for millions of years. Nanotech will have to work hard to catch up but should win in the end because biotech only harvests energy at the first level at a 1% efficiency, while nanotech can push that to 30%+. Nanotech wins big over biotech because you can have active material transport rather than diffusion transport so it assembles much faster providing a higher production rate.

All three technologies, nanotech, biotech & bulk tech have to factor in design cost and amortize it over a production run. Low volume, high complexity parts will be expensive. Because nanotech & biotech *may* have much greater complexity (which is one of their advantages), they may have increased design overhead. But then what you are getting had better be something that can't be produced by bulk tech (otherwise there isn't much point to the exercise). An example would be mass produced intelli-Al2O3-bricks that can assemble themselves into a building. Stronger, so you need less raw materials and since they are self-assembling (i.e. they read the plan and stack themselves) you have taken much of the labor cost out of a building. Since you can amortize the design over millions of buildings, even bulk stuff designed from nanotech seems cheap in the big picture.

> His overall point reminds me of what has happened so
> far with nuclear energy and it's original promise of
> "energy too cheap to meter". The promise was based on
> the simple physical fact that the energy levels
> producable with nuclear reactions are orders of
> magnitude greater than with chemical reactions.

So, your fuel costs are lower. These predictions were made before the design costs and safety and waste disposal overhead were taken into account. The design costs are particularly significant, since *because* of the high energies involved, they have to provide robust safety margins.

> However, when we actually went about constructing
> nuclear energy plants, we found we needed to staff
> them with so many well-trained people and include
> layer-upon-layer of safeguards (not to mention the
> costs of continually assuring the public that it was
> safe) that, ultimately, the energy produced costed
> more than that produced by coal-burning plants (which
> is a shame, since burning coal is so much more dirty,
> provided the nuclear plant is a good one that is run right.)

You should read Freeman Dyson's commentaries on the nuclear industry (in several of his older books). Part of what killed us was complex designs by multiple vendors. The French got it right by adopting a single standard design, mass producing it and having an experienced construction & operating industry that could work with that standard design.

It would be interesting to have comments by an economic libertarian or even just a die hard capitalist as to when competition fails due to the high costs of engineering complex systems for which there is only a very small market.

As far as, the nuclear industry goes, I'm glad it did/is failing. I want to burn all the coal and natural gas we possibly can. The "threat" of global warming is going to create allies between the "ecologists" and the nano-engineers when we can offer a cheap solution for getting rid of all the accumulated CO2. Its the CO2 in the atmosphere that provides the *free* construction material for my future toys.

Rock on.

Robert

His overall point ALSO reminds me of the so-called "productivity paradox" associated with
computerization, and certain total-cost-of-ownership arguments about PCs.

Plus ca change, plus c'est la meme chose???



** *** ***** ******* ***********
M. E. Smith
mesmith@rocketmail.com
http://members.home.net/mesmith/
** *** ***** ******* ***********

Do You Yahoo!?
Bid and sell for free at http://auctions.yahoo.com