the economics of transition to nanotech

From: Robert Bradbury (bradbury@genebee.msu.su)
Date: Thu Jan 13 2000 - 18:54:37 MST


I wrote most of the following as a response directly to Damien,
but I realized that many people on the list might be interested
in it. So for what its worth...

> Subject: Commentary on THE SPIKE edits.

> ... To put the matter cudely: putting your nest egg into nanotechnology,
> ... counterfeit money, or turn lead into gold... imagine the inflationary
> consequences...

I think this is very thin. You have to consider that "counterfeiting"
is exchanging something of little value (worthless paper) for something
of great value (hard goods). On the other hand, a nanotech economy
is going to be exchanging something of "presumed value" (monetary
equivalents) for something of "equal" value (nanostuff). The relative
values will be determined by standard economic laws of supply and
demand just as Robin has suggested. Note that Eric has never said
things will be free, just that complex "stuff" should decrease to
the cost of "bulk" stuff. Even things like sand and salt for the
streets of Moscow are not "free". You just get very interesting
changes in the cost equation that have to be subject to ROI analysis.
For the price of one-time costs, you can engineer a layer of energy-storing
and releasing surface coatings, paint them onto the street and
forget about them for 20 years. But these will only be used if the
costs of design, test, production and application are lower than
the costs of "sand" and "salt".

The trends towards hyper-low cost technologies are present now
(i.e. no "minting" required) in microchips. When chips with unique
IDs, GPS locators, out-of-the-store suppresion range self-position
auto-broadcasting capabilities are available and relatively cheap
($.50-$.01) each, they will be applied to everything in a store and
shoplifting will cease. Micro-electronics & Micro-Mems are on
similar curves (the latter following the former), as the volumes go up,
the costs decrease and the applications explode. Its worth noting
that the obsolete Micro-electronics foundary becomes a state-of-the-art
Mems foundary, so those considerations (Whitesides?) of $10-billion
FAB costs ignore the possible reuses of such fabs in alternate arenas.
For example, the typical analysis of the fab costs ignores the residual
value of a fab making state-of-the-art microprocessors for 2-4 years
followed by a 2-4 year lifetime making lower ROI ICs (in larger volumes
due to their decreased size), followed by an unknown lifetime producing
alternate applications (MEMS, etc.) we can't even imagine yet.
Due to the extraordinary difficulty predicting market size and
sales of products in the 2-5 year future time frame, it is *very*
difficult to say whether $10B fabs are lemons or cash cows.

> Re: Barry Jones & "exponential rise in output together with an
> exponential fall in total inputs"

Its worth noting that this *only* works until the market is saturated.
I'd agree with Robin, that all of economic history has been about
exponential growth and decreasing the total cost of inputs.

There are two forces at work here (1) Growth in Market size causing
decreases in costs, causing expansion in market size. Seen most
clearly in the early days of the automobile (Ford, I think is
attributed as saying, for every doubling in market size, there
was a halving of costs.) This is currently a major part of the
driving force in the computer industry. (2) Miniaturization, as
you mention, allowing you to produce "more" with the same raw materials.

Its interesting to compare the auto & microelectronics industries.
While the former is halving car mass perhaps every 20-40 years,
the later is effectively halving mass (chip area) every couple
of years. The difference is simple in that you couldn't reduce
car size beyond that of "people", while in microchips you can
go down to atomic sizes before things get really hard.
Because the microelectronics industry has much farther to go
before hitting bottom, it is more likely that the costs will
allow everyone to afford computers, long before everyone
can afford a car (until minting steps in).

I don't know if you separate things out, but there are 3 aspects
of minting -- self-replication, self-assembly and molecular assembly.
We will borrow from nature methods for doing the first two before we
master the third. That means the really hard part ("diamondoid" molecular
nanoassembly) is divorced from much of the economic consequences!!!

If we can get "programmable" self-replication we get the costs of
most "parts" down to the costs of wood, agricultural products, etc.
or even lower because we can produce them off of waste materials.
If we can get "programmable" self-assembly, we eliminate much
of the human "labor" costs in putting things together (this
is a logical continuation of Apple's "automated" Mac plants,
Japanese Car factories, etc. with the difference being that
you build the "intelligence" of how to assemble into the parts
themselves. So the really hard part of MNTing (the construction
of a diamondoid biochemistry system) is *only* required if you
want to build stronger & lighter structures than required for
most current activities. So, dMNTing isn't required for food
production or most energy production (from solar, wind, waves)
or most housing. Building high-rise buildings is limited to
the strength of materials like bone, hydroxyapatite (teeth),
sea-shells, bamboo, etc. These are only "somewhat" lower than
things like steel & concrete. But I think you could see how
semi-intelligent pieces of steel or concrete could self-assemble
or how bamboo-beamed houses could "grow" themselves.

If you want to throw down the gauntlet with respect to MNTing,
you need to break it up into separate areas, each of which provides
distinct economic benefits. That makes it harder for critics
to attack since they have to attack all of the legs it stands on.
It makes it more viable from an economic investment standpoint
because different groups of investors will invest in different
aspects. Biotech literate people will invest in self-replication.
Robotics/MEMS/Construction literate people will invest in self-assembly.
Materials engineers and the people who need the absolute limits of
performance (military, rocket builders, etc.) will invest in dMNTing.

The three areas will develop at different rates but once mature
will allow a large amount of borrowing and cross-pollenization
that merges the advantages of each field.

Getting back to Barry Jones and the "exponential fall" of total
inputs. I question this. Presumably what happens is the
recognition of the most "expensive" parts of the production cost
(labor, capital, energy, space, etc.) and a gradual substituation
of some other "cheaper" component. If labor is expensive, you
invest in labor saving automation. If energy is expensive, you
borrow capital to design less energy intensive ways of performing
the production. If space is expensive you move to a country
with lots of land. The key factors are knowing how large your
market is or will grow and the relative growth rates. It would
be interesting for Robin or someone else to comment on market
"saturation" economics. I think you have to deal with the
declining returns on substitutions. Exponential sales growth
(to previous non-owners) is relatively simple compared with
exponential decreases in input costs. One involves "sales"
while the other involves increasingly difficult cost analysis
and engineering. The curves may be exponential, but they
have different exponents and different limits.

The major factor, as implied by Anders, and perhaps others is the
rapidity with which markets "saturate" in the SPIKE era. I may own
2 or 3 or 4 copies of "typical" consumer goods if I'm maintaining
apartments or houses in that many cities but once I've got 1 of each
in each location, sales of the "new" and "improved" models are likely
to be very difficult. This becomes even worse if most of what you are
selling has to compete with open source "software" that is downloadable.
"Reconfigurable" matter is going to significantly impact the development
and sales paradigms. After my general needs are met, you have to
move innovation from "more" (of the same) to *much* better, or
*very* different to make a sale. I can see this today in an average
store in Moscow. A typical electronic coffee grinder may cost
~$30-40, while an old, "unique, elaborate" hand-grind coffee grinder
may cost $120+. You pay a premium for the low-volume, "unique"
sales items. The sales paradigms we have today don't change,
they just get shifted onto different technology.

> Re: "bloatware, sloppy or massively redundant code requiring
> colossal extra amounts of computing grunt to do pretty much the
> same jobs about as well."

This isn't really what bloatware is all about. Bloatware is about
adding functionality for a decreasing fraction of the user base.
For example, I really *care* if my OCR program recognizes superscripts,
subscripts and Greek letters. The average person cares much more
if it can accurately scan a fax or enter a "form". Many people "caring"
results in a bigger program to handle all of the special cases. *If* the
market is big enough, there is an incentive for developers to handle those
special cases (just as we look for "employees" who understand the
terminology or protocols in specific areas of human activity).
While the programs may get bigger, this is easily fixed with "download
on demand" software.

Now, I will *freely* admit that major software developers are "stupid".
For example Microsoft, "re-invented" their spell checker in FrontPage
so it doesn't take advantage of all of the previous development in
the spell checker in Microsoft Word. But given my experience in major
software organizations, after a couple of iterations, they will usually
get the re-usable code paradigm correct.

Net result, software will grow until it has the capacity sufficient
to handle *most* cases found in human experience. Case in point --
we now have ISO fonts that handle most major languages as well as
a bigger 16-bit "universal font". Language "specificities" is a
relatively "solved" problem, so those fonts will grow marginally,
if at all in the future. The computer world is still playing catch-up
to the variety in the real world, once the knowledge bases are relatively
equalized, the "bloat" growth will match the true creativity growth.
Another example would be cross-language dictionaries. Once all of
the human languages have cross-language translations that can be
stored in an inexpensive amount of memory, this "bloat" growth will
cease.

> Re: free assemblers
> (from ... Meanwhile, the cost of refitting fabrication plants to create
> chips at smaller and denser ..... assembler machines that literally compile
> material objects, including more of themselve? Where's the profit in that?)

Well, this kind of assumes there is a single race that can be "won".
As I've pointed out previously, with self-replication, self-assembly
and dMNTing, there are a number of races. Furthermore, if you really
understand dMNTing, you understand that it *IS NOT* a one-size fits
all solution. For starters, you've got dMNTing, SaphireMNTing and
TungstenCarbideMNTing. Each of which operates as successively higher
temperatures and requires a totally different biochemistry (feedstock,
tool-tips, energy sources, etc.) Whether you can use one level of
MNTing to enable the next higher level remains a very open question.
Now on a completely orthogonal axis is presumably energy efficiency
with regard to MNTing (something Eric is aware of but marginalizes).
Energy will never be "free". The less energy you devote to MNT assembly,
the more you can devote to "computing/thinking". So there will be
an ongoing incentive to reduce the costs of MNT assembly. If you look
at the biochemistry of current MNT (bacteria, eukaryotic cells, etc.)
there is a very significant emphasis placed on energy efficiency.
If you can grow more with less energy, you get "ahead" in the game
of life. That implies that even if you can do dMNT, doing dMNT
or sMNT or tcMNT "cheaper" than the next fellow will give you a
big marketing advantage. So, the corporate CEO stands up and says
"We can do dMNT!!!". I'll be in the back of the crowd raising my
hand and asking "... and what are your energy losses....". For
every "success" there is an opportunity for the competition.
If I can do it "cheaper" than he can, I can eat his lunch.

Robert



This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:26:14 MST