Subject: [extropy-chat] Social Implications of Nanotech

[Chris Phoenix]

On Mon, 10 Nov 2003 13:12:36 -0500, Robin Hanson wrote:
> An easy simple opinion to have is that nanotech won't have much in the way
> of specific social implications.  In this view, manufacturing will slowly
> become more precise and more automated, as it has for centuries, and so the
> social implications of nanotech are subsumed by the social implications of
> generally improving tech, and any specific products that enables.

An argument against this view:  Different technologies can be improved
at different speeds.  Computers really took off when the transistor was
invented; they couldn't have done it with vacuum tubes.  So this view is
effectively arguing that nanotech will not invent any significant new
manufacturing technique that can be improved rapidly or that provides a
useful discontinuity like the difference between digital and analog
computation.

> Another opinion that I've heard has more distinct social implications,
> though I'm not sure how many people (still?) take it seriously.  It is
> described in the novel "Diamond Age" and in several books by Drexler and
> company.  

You think it's being taken *less* seriously over time?  Maybe the
Engines of Creation view is fading, but I think the Nanosystems view is
replacing it.  And the latter is much simpler, better defined, and thus
easier to believe.

> .... and most consumer goods are produced locally
> on PGMDs, via downloaded designs and a few general feedstocks.   A
> variation on this position posits that PGMDs can produce more PGMDs
> relatively quickly.  And a refinement of this position posits that such
> self-reproducing PGMDs dramatically lowers costs relative to technology
> available just prior to this point.

I'm not aware of any proposal for mechanochemistry-based PGMD's
(MBPGMD's) that does not posit the MBPGMD being able to rapidly
duplicate its structure.  Let's call this "autoproductive" since there's
no good existing word.  ("Self-replicating" is too confusing for several
reasons.)  

Your phrasing makes it sound like there is a range of positions, but can
you cite any source for them?  Or are they your invention?  If the
latter, I think they will not serve you well, because I doubt that a
non-autoproductive MBPGMD is technically  plausible, much less
economically plausible, since you probably can't build a human-scale
MBPGMD without bootstrapping it from a much smaller one.  

Now maybe this is what you want: to show that PGMD's may exist, but not
be economically significant.  If so, I think you're merely building a
strawman.

To build a sub-micron MBPGMD with pre-MBPGMD tools would be extremely
difficult.  To build two would be almost twice as difficult; to build a
kilogram manufacturing capability would be impossibly difficult.  So it
will have to bootstrap--to produce an MBPGMD twice as
large--approximately 30 times to make a cubic millimeter, 60 times to
make a cubic meter.  If it takes a month for each cycle, it will take 5
years before you obtain a possibly useful factory.  Since there's no
known reason why it should take even a day, and an hour is quite
plausible based on scaling laws, comparisons with bacteria, and
extrapolations from proposed architectures, the hypothetical month-long
autoproductive cycle will probably never be used: more refined MBPGMD's
will be developed within that five years.  

So how long will the first primitive MBPGMD's take to produce their
mass?  Once the technology works at all, it can be used to build more
elegant versions of the machinery.  This kind of engineering tweak
shouldn't take long, given the huge incentives for better performance. 
If I know my competitor has an MBPGMD design that will bootstrap itself
in two years, then I have a full year to tweak his design into one that
will bootstrap itself in one year.  But a one-year bootstrapping time
implies a ~6-day doubling time, or ~3 days to make its own weight.

If I can take my factory offline for three days and double its capacity,
it's hard to imagine factory capacity being scarce.  Since most kinds of
demand are not constant, a smart factory owner would use undercapacity
periods to prepare for future big orders.  And since MBPGMD's are
assumed to be small and distributed, it would be possible to take a
small fraction of capacity offline without doing much damage to ability
to respond suddenly.  Perhaps the factory would only grow at 10% per
week, instead of 200% per week.  This would still be sufficient to
ensure that enough capacity was available to meet any reasonable demand,
and most of the time, capacity would be sitting idle.

> 1.  It is often assumed that a world of PGMDs is one of marginal costs near
> the cost of feedstocks, with the main fixed cost being the cost of
> design.  But this depends crucially on the PGMDs being typically used well
> below capacity, as most PCs are today.

Yep.  The above shows why this is reasonable.

>  Most manufacturing plants today
> have a pretty low marginal cost, in terms of how much you save if you
> operate them below capacity.  But since the plants are used near capacity,
> this makes them little like software or other goods that really do have a
> low marginal cost of production.

I think this shows why you can't generalize from today's manufacturing
plants.

> 2.  A big question is by what factor general manufacturing devices are less
> efficient than specialized manufacturing devices, either in terms of
> production time, material waste, or final product quality.  

Don't compare apples and oranges.  That's like asking by what factor
digital computers are less efficient than analog computers.  Digital
computers can do things that analog computers simply can't.  This makes
them so valuable that they're ubiquitous.  And because of the various
high costs associated with analog computers, there's no market for them
anymore.  Sure, I could in theory do linear regression with a pencil, a
piece of graph paper, and a few thumbtacks and rubber bands.  But
there's really no point.  

The question is: If an MBPGMD were installed in your local convenience
store, would there be any point in either installing a special-purpose
widget maker or trucking in widgets that were made with centralized
special-purpose machines?  Installing a widget-maker: obviously not. 
There are too many different things sold in each store, and too low
volume on each one.  Trucking them in: probably not.  MBPGMD production
time would probably be lower than shipping time.  Material waste would
probably be lower than transportation fuel, warehouse rental and widget
depreciation, miscellaneous handling charges, and depreciation on the
widget-maker.  I'd expect an MBPGMD to have *better* final product
quality than special-purpose manufacturing.  There's no sense using
mechanochemistry for special-purpose manufacturing, so you're left with
bulk manufacturing techniques, while the MBPGMD can place every atom as
designed.

I'll note that no one even uses slide rules anymore.

> The bigger this
> factor is, the larger need to be the scale economies in the production of
> PGMDs for them to dominate.  

This is much too small a view--it considers too few factors.  The
disadvantages of special-purpose manufacturing are partially explored
above.  In addition, the products that could be made only with MBPGMD's
would probably become a large fraction of the market very rapidly--I
suspect they'd even dominate the market, thus automatically eclipsing
special-purpose technologies.  

> At the moment most manufacturing devices are
> really quite specialized.

Yes, and analog computers were used for a while before digital computers
became practical.  This is changing with rapid-prototyping systems.

> 3.  PGMDs embody almost *fully* automated manufacturing - if they need
> people to step in frequently to diagnose and fix assembly line problems,
> they become much less attractive.

I think MBPGMD's have to be completely, fully automated.  No one is
going to fix a five-micron assembly line.

>  While many manufacturing plants today
> are highly automated, it may cost quite a lot to produce designs for fully
> automated production processes.  So design costs may be a lot higher.

Or it may cost quite a bit less.  You don't have to retrain your
workers.  You don't have to pay workers to retool your machines.  And
you're assuming that the design space is the same, which it won't
be--not even close!.  MBPGMD's (especially since you automatically get
great material strength) give access to an unimaginably huge design
space, with lots of room for very simple assembly techniques that are
amenable to semi-automated design.

> 4.  The manufacturing fraction of the cost of most consumer goods today is
> rather small (15%), and only part (~1/3) of those manufacturing costs now
> are the physical capital, rather than labor and design.  So it is not clear
> how just lowering those manufacturing costs will have a huge effect on the
> economy.

Um, why are you not counting the labor cost as a saving?

And what about transportation costs, warehousing, the costs of
compensating for uncertain and delayed supply chains?  And you have to
regress (multiply) these costs for each level of part fabrication or
processing that requires storage and transportation.

And the cost of feedstock should be lower for simple organic chemicals
than for diverse materials that must be extracted in a variety of
complicated ways.  

> 5.  If the cost of designing and building an effective self-reproducing
> PGMD is much higher that of ordinary PGMDs, there might be plenty of
> ordinary ones around before any self-reproducing ones appear, minimizing
> the social impact of this transition.

Again, I think the "ordinary" PGMD is only a strawman.  Unless you
extend PGMD to bulk-process (non-nanotech) rapid prototyping systems?  I
admit they might produce *some* of the impacts of MBPGMD's.  But I
seriously doubt that even taking the MBPGMD transition in two equal
steps will reduce it enough to merit the word "minimize".

Chris

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