Copyright 1992 by Mary Eisenhart. All rights reserved.

Nanotechnology
Separating Myth From Reality
Preparing For The Future
An Interview With Eric Drexler
By Mary Eisenhart
Photos: Glenn Matsumura

I expect in the ultimate destiny of this that there's going to be an
appliance that sits on your desktop that can make anything. I don't
expect that to happen within the lifetime of AutoCAD Release 13 or 14.
But that's where it's all going.
	This week we're shipping HyperChem, which lets you design objects at
the molecular level. The technology is going to go on scaling down
until we have the same level of fine-grained control over the
structure of matter that biology has been using for the last billion
years or so. And at that point, we're going to be able to make
*anything* from a digital model.
	That's going to change the world, beyond comprehension, I think.
		--John Walker, MicroTimes #95

Imagine that many of the key developments of twentieth-century
technology were consequences of something that could be figured out in
the nineteenth century, but still would take many years to implement.
So in the nineteenth century, someone was able, not to understand the
whole of the twentieth century, but to understand that it would be
possible to develop antibiotics, nuclear weapons, and farm
technologies that made it possible to feed the population with only a
few percent of the population working in agriculture. Let's say that
this was determined at a time when most people were farmers, when
infectious disease was a dominant cause of death, and when no one had
any concept that it was possible to destroy the planet.
	One could imagine a range of scenarios where the ability to have
food without people working on the farm produces unimagined wealth and
leisure and is portrayed as Utopia. One can imagine the existence of
nuclear weapons as being portrayed as a sure recipe for our immediate
obliteration, and the idea of penicillin eliminating a wide range of
infectious diseases as a fantasy--"a pill that cures all these
things?"
	The reality has been much more complex than any of those views, and
much of what has happened would have been rejected as a fantasy at the
time. Indeed, many correct predictions at various times were rejected
in just that manner. I think that's a useful comparison to judge
what's being said now, and how it may be compared to the eventual
reality.
	--Eric Drexler

 Back in the 1970s, while an undergraduate at MIT, Eric Drexler read
about the advancing state of research in molecular biology, where, he
says, "researchers kept discovering new molecular machines within
organisms and studying how they worked. I began to wonder what kinds
of things we'll be able to build when we have tools we don't have yet,
tools able to build molecular machines."
	Not without a fair amount of struggle for acceptance in academic
circles, Drexler ultimately received a Ph.D. in Molecular
Nanotechnology from MIT, and is currently a Research Fellow of the
Institute for Molecular Manufacturing. He is also the president of the
Foresight Institute, a Palo Alto-based nonprofit educational
organization founded, as he explains it, "to gather, digest and
distribute information on molecular nanotechnology and related
advanced technologies, discussing both what's happening in the
laboratory today, what the pathways are, what the long-term
technological capabilities are, and the consequences for policy and
decisions that need to be made."
	Since Drexler's first book, Engines of Creation (Doubleday),  first
appeared in 1986, nanotechnology has garnered enthusiastic interest
(and equally vigorous nay-saying) in the technical and scientific
communities. It has also brought on a bizarre assortment of half-baked
speculation, snake oil, and pie-in-the-sky schemes.
	This, Drexler points out, is only natural for discussions of
products likely to be decades away. And while the inevitable onslaught of
flakes and charlatans is a known road hazard on the technological
highways--particularly when sober scientists discuss nanomachines
dedicated to killing cancer cells or reversing the aging process, or
food production reminiscent of the have-it-your-way gizmos on Star
Trek--it need not detract from useful study.
	"It's very important to distinguish between what is near-term and
what is long-term," says Drexler, "and also between what's real and
unreal. And to not confuse 'long term' with 'unreal,'  which is a
classic American disease."
	While in the United States, there is little funding (public or
private) committed to research in the field, the situation in Japan is
quite different, with MITI announcing last year a commitment of $185
million to nanotechnology research, and the Science and Technology
Agency funding research projects aimed at building complex structures,
atom by atom. One such project, the Aono Atomcraft Project, has been
taking a hands-on approach that parallels theoretical studies by
Drexler and Xerox PARC researcher Ralph Merkle, but, says Drexler,
"Aono has funding and a laboratory. Ralph and I are both doing
theoretical studies--studying what kinds of systems can be built in
the long term, and how to get there. Which is different from taking
steps in a laboratory along one of these paths. And also much less
expensive!" he laughs.
	"I don't know of any people who have looked closely at molecular
manufacturing, and understand the physical mechanisms involved and the
analysis behind it, who are not of the opinion that, yes, this is
going to be a dominant manfacturing technology in the twenty-first
century. Certainly that it's going to be used to make computers. Quite
likely that it's going to be used to make a much broader range of
products," says Drexler. And the Japanese, historically more concerned
with the long term than the quarterly bottom line, are putting
significant resources into nanotechnology. "Rather than seeing twenty
years as forever, they see it as a fraction of a human lifetime, as
planning for one's own personal future, for that of one's friends and
relatives and children."
	So, for now, Drexler and his colleagues are engaged in research and
information dissemination, taking the small steps along the path to a
radically different future. In November, the Foresight Institute will
present the First General Conference on Nanotechnology: Development,
Applications, and Opportunities in Palo Alto. The conference lasts
three days and features discussions of issues ranging from
"Intelligent Computation in the Age of Molecular Manufacturing" by
MIT's Marvin Minsky to "Designing Molecular Components" by Ted Kaehler
of Apple Computer to "Nanotechnology R&D Sponsorship" by Neil
Jacobstein of Cimflex Teknowledge.
	For the last four and a half years, Drexler's been working on the
just-published Nanosystems: Molecular Machinery, Manufacturing, and
Computation (Wiley-Interscience), a college-textbook-level work
intended to serve as a foundation for the studies of future
researchers in the field. In 1991, he and coauthors Chris Peterson and
Gayle Pergamit published Unbounding the Future (Morrow),  which
explores, at a level accessible to a nontechnical audience, the basic
concepts and social implications of nanotechnology.
	With a view to similar exploration, we recently paid Drexler a
visit.

What do you mean by "nanotechnology"?
	Since 1986, it's become a buzzword in science and technology, but it
has been applied to anything on a nanometer scale, rather than to what
I've been terming here "molecular manufacturing."
	There are two basically different approaches to making small
structures. One is the top-down approach that's been followed in the
semiconductor industry, where you use large machines to make small
features, and you strive to make smaller and smaller features every
year, while still keeping enough control of the shape of those little
irregular structures that they do what you want. The payoff from this
has been enormous; that path has a considerable distance to go, and
the computer capacity that results will contribute greatly to the
development of molecular nanotechnology.
	But molecular nanotechnology itself moves in the opposite direction.
It starts with small, precisely structured molecules, which people
have been making for a hundred years or more through organic
chemistry. And the challenge, rather than making things smaller, is to
make things larger, while keeping that precise control.
	So molecular nanotechnology is a bottom-up path. Conventional
microtechnology, which has recently been relabeled with the sexy term
nanotechnology, is a fundamentally different top-down path that I
don't think leads to the same destination.
	I see them as moving together on a size scale, and having some
synergy. I can imagine hybrid systems for a time. But then the
molecular approach, I believe, will displace the other, because of its
advantages of precision, range of control, and cost. Nobody's ever
suggested to me how anything else could substitute for it.
	Molecular manufacturing, which is at the heart of molecular
nanotechnology,  will be based on the use of molecular machines that
use molecules as building blocks to build complex structures.

Explain "molecular machines."
	A conventional machine is a large collection of atoms with parts
that move and do useful things. A molecular machine is one that has very
few atoms, relatively speaking--thousands or millions or billions are
typical numbers--and that has all its atoms in precisely defined
locations as an aspect of its design and manufacture. So they're
devices that have scales measured in nanometers instead of
millimeters, which makes them roughly a factor of a million smaller
than conventional mechanical devices.
	The things that molecular machines do in biology include taking
molecules and putting them together in complex patterns to make other
molecular machines. A wide range of products--everything from protein
molecules to redwood trees--is made by biological molecular machines.
	Studying what could be done with artificial molecular machines led
to the conclusion that they could be used to build better molecular
machines that could be used to build better molecular machines that
could be used to make--it's a slight exaggeration to say "anything
that's physically possible by arranging atoms," but surely a much,
much wider range of physical structures than can be made any other
way.
	In the 1970s, genetic engineering technology was just getting
underway. This made it possible to synthesize, more conveniently than
before, new protein molecules. Proteins serve as components of
molecular machines in biology, and that meant that the problem was one
of design. We had the tools to make building blocks for molecular
machines; the challenge was to design them.
	So in 1981 I published a paper in the Proceedings of the National
Academy of Sciences, arguing that protein design was in fact
feasible--many people at the time thought that it was impossible
because of the complexity of different ways that a long, floppy
molecule could fold up--and then presenting an analysis that indicated
that with protein-based molecular machines, you could start this
process of building better molecular machines, and eventually have
very sophisticated molecular manufacturing systems.
	And in fact in 1988, a protein molecule was successfully designed
from scratch, by Bill DeGrado at DuPont. A number of groups have done
likewise since.
	Unfortunately, this work has not been done by people with a systems
engineering perspective, and so little or no effort has gone into
making building blocks for larger systems. It's been done by
biologists and chemists, who are chiefly interested in studying
natural systems by imitating them, with small differences, and in
making devices that do what we see in nature, perhaps under different
conditions.
	For example, proteins have been engineered for greater stability, to
enable them to sit in bottles of enzyme detergent on shelves in
grocery stores for a long timeYx
So what do engineers contribute?
	Engineers bring an understanding of the principle that a set of
building blocks and a way of putting them together can be used to make
systems of indefinitely large size and complexity.
	We see in computers that knowing how to make a few kinds of
transistors and connections between them is sufficient to make the
most sophisticated supercomputers in the world, or at least the CPUs
for them. We see in software that a few instructions in a RISC machine
can be used to implement absolutely any computational process. And
that understanding, applied to the molecular domain, can open a new
world of engineering.
	In some ways the development of molecular manufacturing will be like
the development of the digital computer. Once upon a time, there were
many different special-purpose devices for processing
information--adding machines, and analog computers, and so on and so
forth. Then there was the development of a general-purpose device for
processing information that worked by handling information in what
could be seen as its fundamental units--bits--performing repetitive
operations in a controlled sequence at high speed, using smaller and
smaller devices.
	In conventional manufacturing today there are a wide range of
special purpose devices--milling machines and lathes and injection molding
machines and so forth--but what we can look forward to is the
development of a general-purpose device that can make virtually
anything by working with matter in terms of its smallest building
blocks, atoms and molecules. And again, working at high speeds, using
small devices performing simple repetitive operations in a controlled
sequence.

What kinds of products and technologies might come out of this?
	The easiest products to imagine just offer large performance
improvements over ones that we already have. For example, we've seen
computers for many years now become smaller and faster and more
energy-efficient. With molecular manufacturing, it will be possible to
make machines that are somewhat faster--billion-cycle-per-second clock
rates--that are smaller than a bacterium and consume so little power
that a billion times the capacity of a modern supercomputer could be
put in a desktop box and cooled with a fan. The reason that these
devices are only somewhat faster than modern electronics is that this
particular calculation is based on a simple mechanical scheme, a
hybrid between Babbage's Analytical Engine (laughs), modern VLSI
architectures, and molecular components.
	Ralph Merkle at Xerox PARC and I have been looking at electronic
schemes for computation that would likewise exploit nanometer-scale
components built with molecular precision.
	Most of the work that Ralph and I have been doing, though, uses the
standard computational modeling tools that chemists have developed to
predict the behavior of the molecules they're working with, and
applies those tools to describe large structures that are designed to
work like macroscopic machine components. Things like robot arms in
factories--for an example that's highly relevant to molecular
manufacturing--have gears and bearings and a wide range of moving
parts. So Ralph and I have designed a set of bearings with hundreds to
thousands of atoms--the principles scale up to larger structures--and
more recently have designed the first nanometer-scale planetary gear
system, a device that has a shaft coming in one side spinning it at
one speed, and a shaft coming out the other side that spins at a
different speed, and transmits power from one to the other.

What might this gizmo be used for?
	This particular device is primarily a demonstration of what can be
designed and modeled with available tools. But planetary gears are
found, in the macroscopic world, in automobile transmissions, and in
nanometer-scale mechanical systems they can serve similar functions.

What separates the designing and modeling, which you can do with
current molecular modeling technology, as I understand it, from the
ability to actually build the things?
	What we've been designing are things that can't be built with
anything like today's tools. They can be modeled, in part, because
they contain very few atoms and therefore are among the simplest
mechanical devices that are possible in the physical world.
	But to make them will require tools that can position very reactive
molecules precisely in three-dimensional space, to build up these
structures in more or less a building-block fashion. Those tools don't
exist, but we can see how to go about developing them in a
step-by-step fashion. It will take many years.

What is the current state of tools? It is now possible for people at
IBM to line up a bunch of atoms to spell IBM. But as I understand it,
this is done at very, very low temperatures, and when you get to room
temperature the atoms go about their normal business... So this is
transitory and expensive and not currently useful.
	The next major step on the pathway that currently seems most
attractive involves using a relative of the scanning tunneling
microscope (which was what Don Eigler's group  used at IBM Almaden),
the atomic force microscope, as a positioning mechanism for reactive
molecules, molecules that, when put in place, will form strong, stable
bonds to their neighbors.
	Markus Krummenacker, a researcher supported by the Institute for
Molecular Manufacturing, is presently studying the design of molecular
building blocks with those properties. And research is getting
underway here in the Bay Area aimed directly at building an instrument
able to position such building blocks, a descendant of the atomic
force microscope. This instrument, as planned, will be an adaptation
of the atomic force microscope.
	The atomic force microscope can act like a robot arm, able to
position things in three degrees of freedom, not rotating, but X-Y-Z.
What has been lacking is the functional equivalent of a hand at the
end of the arm. It's like a robot arm that ends in a rough rock,
rather than a selective gripper. A suitably attached protein molecule,
in particular an antibody fragment, can serve as a selective gripper
for other molecules.
	This family of technologies can initially be used as an improved
atomic force microscope--rather than bumping a blunt rough object
against something to probe its shape, you can bump a sharp, precisely
defined molecule against something to probe its shape, and can in fact
probe it with several different tips, one after another.
	This promises to be a powerful tool for biological studies, and for
the pharmaceutical industry, as soon as it can be developed, and
without building anything else. So that will, I believe, help to fund
the development of the technology base, but that same instrument, by
positioning reactive molecules, can build things. And the question is,
what?
	The first set of applications will be molecular instruments, made by
building a complex molecular structure that's comparable in size to a
protein molecule, or a set of protein molecules. That, for example,
reads DNA--the DNA molecule binds, and feeds through the structure,
and is destroyed one step at a time. If you can probe the DNA molecule
as it goes by, so that you are reading out the sequence, you would
have a molecular DNA reader. We know that molecules can read DNA,
because that's how DNA is copied; the challenge is to build a device
that reads DNA, and instead of producing a copy, produces, through a
suitable interface, a pattern in computer memory.
	This data could be used to learn about the molecular mechanisms of
healthy tissue and of disease processes, and would be an aid in
medical studies and pharmaceutical design, as well as in basic
biology. If you can easily map all of the genes in a normal cell, then
all of the genes in a cancer cell, you're going to more easily learn
what the genetic changes were with that particular cancer. And from
there, you'll have made at least a step in learning how to deal with
it.

One of the scenarios in Unbounding is that of taking a pill of little
nanomachines that attack only your cancer cells. How realistic is
that? How do you separate the snake oil from the reality when you're
at such an early stage? Some people would have it, for instance, that
nanotechnology will enable us to live forever...
	Well, first, it's clear that the ability to build sensors and
medical instruments at a molecular scale, and to guide them using computers
that are themselves smaller than a cell, will open a new world of
medical capabilities. It will bring surgical control to a molecular
level. What that can mean for treating disease is extremely broad.
What it can mean for reversing the molecular and cellular changes of
aging is also very broad.
	But when people speak of immortality, they're speaking of something
that can't be achieved within physical law as we understand it,
because nothing is permanent and indestructible.

This spring I heard you say that when you first got interested in
this, it seemed like a very, very long-term thing, and then relatively
recently enough progress has been made to make you think that some of
it, at least, is fairly short-term.
	Over the years we've seen both an improvement in the technology
base--the ability to make molecules and the ability to maneuver
physical objects with subatomic precision. That's come out of
laboratory work in the rest of the world.
	In my own work over the last several years I have begun focusing on
the question of the first steps. My initial studies were of long-term
capabilities, which in many ways are simpler than first steps, in much
the same way that digital logic is simpler than complex nonlinear
transistor circuitry where you're paying attention to all the
messiness of what's going on in those circuits.
	Studies of those questions, taking advantage of recent developments
in the technology base, indicate that there are impressive advances
that can be made with something on the order of one year of laboratory
effort, involving just a handful of people and relatively inexpensive
equipment. And that there are a series of steps of increasing expense
and difficulty as one builds more complex systems that lead from there
to advanced molecular manufacturing of a wide range of products. With
payoffs at every step along the way.
	It appears that a first generation of scientific instruments based
on improved atomic force microscopes can be developed in a time frame and
cost range that is standard for venture capital. A few years, and a
few million dollars.
	Those instruments would then be very useful in biomedical research
and pharmaceuticals, and would have a market that's substantial
compared to that initial investment.
	That same technology base, with moderate extensions, can be used to
build complex molecular structures by positioning reactive molecules.
Development can be rapid, because the time required to do an
experiment to find out what's wrong with it and try again is short, on
the order of a day. And the instruments required to do that are
expected to be in the cost range of a high-end workstation. And so
it's reasonable to expect that a wide range of moderately complex
molecular mechanical devices can be built in a moderate length of time
with a moderate development cost.
	That range of devices includes a yet broader class of scientific
instruments--I gave you the example earlier of a DNA reader. That's
the most obvious  one, and I suspect that as we get closer to that
time and look more closely, a wider range of molecular scale
insturments will appear attractive.
	The reason for focusing on scientific instruments is that the value
of information is independent of the size of the mechanism that
produces it. Information about molecules that are found in the human
body or that are used in industry in the million-ton-per-year range
can be very valuable, even if that information was gained by studying
just one molecule. So building a single molecular instrument can yield
information that's worth thousands, millions, in some instances
billions of dollars.
	Pursuing that development path leads to a technology base able to
build fairly complex molecular machines. Design analysis indicates
that a mechanism with roughly one hundred moving parts is complex
enough that with broadcast acoustic instructions to make it step
through a series of states, it could be used much as the AFM mechanism
could be used, as a construction mechanism, to build things. And since
it itself would have  been built by the AFM mechanism, one of the
kinds of things it could build is more devices of the same sort.
	At that point, you're in a position to make products from chemicals
that have some cost that doesn't look ridiculous when expressed in
dollars per gram, and is extremely small when expressed in dollars per
molecule, since there are many billions of billions of typical
molecules in a gram. At a thousand dollars per gram, you're talking
about much less than a nanodollar per molecule. (laughs)
	If you can produce complex structures from those building blocks for
that kind of cost per gram, the cost per device is many orders of
magnitude smaller than what is possible with semiconductor lithography
and the integrated circuit technology on which computer electronics is
based.
	At that point it's reasonable to expect that someone will design
devices that can serve as RAM storage cells, and will do the
engineering work necessary to build RAM systems in which these
molecular devices serve as storage cells. At that point a reasonable
thing to anticipate is memory devices that use a semiconductor chip as
an addressing mechanism, and a layer of repetitive molecular devices
on top of that--a slab consisting of many layers of molecular devices
on top of that, such that bits could be read from any of the molecular
blocks on the interior of the slab. We're not talking a few atoms, or
a few thousand atoms, but  perhaps a few million per block.
	Nonetheless, reasonable storage capacities for a device like that
are in excess of 1015 bits in a chip-size package.

Compare that to what's currently available, or planned.
	There is talk today of billion-bit chips. This would be a million
times that capacity. And it's possible to do much, much better than
that. What I described is a sketch of what one might see as an
early-generation hybrid system.
	The direction from there leads to building larger and more complex
structures with molecular precision, including CPUs, and denser memory
systems that are not limited by addressing through a chip, but instead
have all of the required devices implemented on the molecular scale.
	And at that point the hardware technology base for computers has been
replaced, and we're clearly speaking of a many-billion-dollar
world-class industry. In that same time frame, a wide range of other
applications can be expected, but I think that example is sufficient
to show the scale.

What time frame do you think that's likely to be? Two years? Five
years? Twenty years?
	Dates are difficult to estimate. It's much easier to calculate a good
approximation to the behavior of a molecule than it is to make even a
rough estimate of what a world full of researchers will do over a span
of years when they're developing new tools and pursuing diverse
goals.
	But keeping that in mind, it is easy to see how to get a
first-generation molecular manipulator with what a number of
laboratory researchers have independently told me looks like a few
months of work, for a crude prototype that demonstrates the
principles. Developing that into a capable system and learning to use
that well for molecular construction might take five years. At that
point it's plausible that you could make machines with a hundred
moving parts.
	With machines with a hundred moving parts, you can now build a
molecular-scale device, a hundred-nanometer scale (tenth-micron scale)
device, that is doing molecular construction, and you're in a position
to begin a multi-year development effort that leads to making
macroscopic quantities of increasingly complex products.
	To reach the threshold of advanced molecular manufacturing, able to
build a very wide range of kinds of materials and structures, will
require another layer of advances on top of that. Again with fast,
inexpensive experiments at each step along these paths, it's plausible
that another five years, with at the later dates large resources being
poured into what is already a very useful technology, could lead to
these kinds of advanced results.
	If you add up that sequence of steps, it adds up to fifteen years,
but I wouldn't place any great confidence on that number.

Where do computers as we currently know them fit in with this, as
tools or otherwise?
	Microcomputers are used to control scanning tunneling microscopes and
atomic force microscopes, to capture the data from them, to process
that data into images. Increasingly in coming years, I would expect to
see more powerful computer capabilities being applied to the
interpretation of those images.
	That will be important as molecular nanotechnology is developed,
because the cycle  of experimentation will involve trying to perform
an operation and then probing the product with the point to determine
what happened. Software for very rapidly and reliably interpreting
those images will play a major role.
	The other major application is in molecular CAD. Ralph Merkle has
been developing software that can be used together with the chemists'
standard modeling packages to help design the structures that the
other package models, and at this point he has quite a long wish list
of capabilities there that they're beginning to work on.

What is the distinction between designing and modeling?
	In molecular design you decide what atoms go where, what is the
pattern of atoms and bonds that makes the shape that makes the
structure of the component. In modeling you take that design, which is
basically just a drawing--and then you turn on physics, or at least you turn
on an approximation to the physics of real molecules, and as a result things
move around, it all relaxes to a minimum-energy state. In other
applications of modeling you can turn on thermal motion, and now the
atoms move around subject to the forces that approximate the ones they
would experience in a real molecule, so you can watch the dynamical
behavior of the system.

So this is in a sense mechanical engineering on a completely different
scale from what we're used to.
	This is precisely mechanical engineering, in which something like
finite-element approaches come naturally because there are a distinct
number of atoms, though it's mathematically rather different from
finite element analysis.

One thing I've picked up over the last few months is that all this
molecule-manipulating is happening at very cold temperatures, and once
you get to something like room temperature all the molecules go on
about their usual business. So how do you deal with things like
turning off heat? How do you isolate what you want to do?
	That's a set of questions that has different answers in different
contexts. The first in a long series of steps will involve having a
lot of stray molecules around that you don't want, having all the
problems that chemists have with choosing molecules that will react
and trying to surround them by others that won't react if they're not
supposed to.
	Today chemists successfully make things by what is the equivalent of
throwing parts in a box and shaking. What these AFM-based molecular
manipulators will provide is the equivalent of a hand to put parts
together. You'll still have other parts shaking around, but you'll add
to the process a hand. There are a lot of limits to what you can do
with that, but it's a big advance over where we are.
	At the end of a long development pathway, what is described in
Nanosystems is macroscopic systems that have a gas-tight wall around
them. There are no stray molecules inside; the only molecules inside
are ones that were brought in there after several steps of sorting and
testing. They're bound firmly, and they go where you want them to,
subject to only very small deviations due to thermal vibration. They
can't wander, they can only vibrate within small range, and that range
can be small enough that they end up where you want them.
	This is at the end of a long series of developments. This is where
you are many years from now with a lot of work.

In one of your papers you talk about how there is not, at this point,
a formal academic discipline called nanotechnology, and you talk about
what somebody can study at this time to acquire a good background in
the field.
	I would say that some of the more relevant standard academic
backgrounds are in physics, chemistry, mechanical engineering, and
computer science. Molecular biology has some substantial relevance,
particularly if there's been a focus on the molecular mechanisms
involved. A lot of molecular biology is at some remove of abstraction
from the nuts-and-bolts physical level.
	I just spent 4 1/2 years working on Nanosystems, and giving it the
content that it has, for the purpose of enabling people with
backgrounds in those areas to learn what is necessary to analyze and
design molecular machine systems, or to develop software to aid in
design and analysis, or to gain enough understanding of their
capabilities that it's possible to think more concretely about what
can be done with them. My hope is that that will be useful for people
who are interested in working in the field.
	Molecular nanotechnology is a strongly interdisciplinary field. The
fundamental interactions are chemical, but they need to be analyzed in
a way that has more in common with the way physicists think of things
than the way most chemists think of things. But the systems to be
built, and therefore much of the analysis of what kinds of things to
design, are primarily a matter of mechanical engineering.
	This has led to two problems in the academic world. In the United
States at least, chemistry, including synthetic organic chemistry,
which is about making new molecules, isn't taught as a kind of
manufacturing, or indeed as any branch of engineering and
construction; it's instead taught as a natural science. At the
University of Tokyo, in contrast, the Department of Synthetic
Chemistry is in the school of engineering.
	So this is an engineering discipline which is founded on the
molecular sciences, but people in the molecular sciences, by and
large, do not know how to think as engineers. Some do, but it's far
outside the mainstream of the culture. That has led to difficulties.
	And also, deeply rooted in chemistry as practiced today is the
assumption that molecules are small and that they move around freely
and run into each other in all possible positions and orientations,
and that if a reaction is possible in any of those positions and
orientations, it will happen. That tremendously limits the building
blocks you can use and what you can do with them. A physical system in
which those same molecules move in a controlled way and only encounter
each other as planned, and can then be pushed together--that seemingly
simple change upsets a tremendous number of usually unexamined
assumptions in the field, even though there is nothing fundamentally
novel about it in a scientific sense. And that's led to a whole
pattern of superficial misconceptions.
	For example, looking at a proposed structure and saying, "That
structure is unstable." What's often meant is, "If that structure were
to run into another identical structure in a particular position and
orientation, then those two copies of the same structure would react."
That doesn't arise in the context of advanced molecular machine
systems, and in that context such a structure is stable. But a quick
glance would suggest to a practicing chemist that whoever wrote down
the structure doesn't know any chemistry, and that's enough to block
the further study necessary to understand the system context in which
the proposal makes sense.
	That's one of the more striking cases in my experience. In fact, the
specific chemist I have in mind who made that observation was quoted
as saying that in an article. I subsequently talked with him and put
him on my doctoral committee.(laughs) At the end of which, I gather,
he decided that I did know some chemistry.

So do you find that as time has progressed, the academic community is
more receptive than it was initially? One of the points in Unbounding
that really struck home was the statement that any new technology is
initially going to attract as many flakes and charlatans as reputable
practitioners.
	And it will do that in particular if it has large consequences. What
you're describing is a pattern that also works with
scientists--society trusts scientists to judge new techn%
1proposals, rightly or wrongly (laughs). Many scientists, however,
judge new proposals based not on how they look when they're examined,
but on how they smell culturally. And if a set of proposals has
striking, exciting human consequences, it will stir a fuss, and the
fuss will create a stink, and the scientists will turn up their
noses.
	And therefore we have a systematic mechanism which is very good at
rejecting nonsense, that also rejects some of the more important
things that are sensible. One of the consequences of this mechanism
which is good at rejecting nonsense is that some of the most important
long-range developments will be systematically scoffed at, for a
while.
	It's in part a matter of asking the wrong experts. Asking chemists
about molecular manufacturing is like asking transistor designers
about computer architectures. Organic chemists work with structures
that typically have a few to a few hundred atoms. A simple low-end
molecular manufacturing system might have a billion atoms. The gap
there is comparable to the gap between a single transistor--an
isolated transistor and package, that level of technology in the
1950s--and the kind of VLSI that we're only getting toward today.
	So, go to the 1950s. Find someone who is working on growing germanium
crystals to make the old germanium transistors, and ask that person to
judge proposals for ten-million-transistor-on-a-chip computers, and
whether RISC or CISC architectures are preferred. You'll get a blank
expression, and the notion of having ten million devices on one piece
of semiconductor would seem to be a fantasy.
	There's also a major cultural gap between the United States and Japan
regarding setting goals, articulating goals. Many Japanese research
projects are given extremely ambitious goals. For example, the
Kunetaki Molecular Architecture Project speaks of self-organizing,
intelligent, self-repairing, etc., etc., etc. molecular systems.
	That project is over. It was a five-year project.  Did they achieve
that? No. On the other hand, what they did was, they set the best
direction they could think of, very ambitious. A direction that, if
achieved, would lead to enormous payoffs. In that project they took
steps in that direction, because that was the direction they'd
articulated.
	If you judge their achievements relative to the end point of the
direction they were pointing, they fall short, and you call it a
failure. If instead you ask, "Did they direct their resources better
because they chose those goals?" I think the answer is yes.
	In the United States we are afraid to articulate ambitious long-term
goals, for whatever reasons--in part, the tendency of people to come
back two years later and say "You haven't done it yet, and since you
couldn't possibly have been planning ten or twenty years in advance,
you must have failed."
	Because of those sorts of attitudes we don't articulate ambitious
long-term goals. As a result, I believe, we tend to put our research
efforts into directions that have, on the average, less long-term
payoff, because we're not looking.
	I'm not saying that we can predict the outcome of research programs,
or that we can predict what will pay off in the future. I'm saying
that we do better if we try than if we don't. If we try intelligently
than if we don't. Otherwise you're throwing away information, soft
information, in favor of hard laboratory data on something that may be
uninteresting in the long term.

It's safe to assume that for every glorious life-saving application of
this technology that might come along, there's almost sure to be an
equally sinister and destructive one, perhaps developed in some secret
government research lab. Is that an accurate perception, and if it is,
how can people who are interested in nanotechnology keep it from going
off to the dark side before it even gets out of the lab?
	That's been a subject of a lot of discussion among people who are
really concerned with these issues. The policy that people seem to have settled
on as making the most sense is to make it clear that yes,
among the applications of this manufacturing technology is the
manufacturing of systems that are tremendously subject to abuse,
things that are inherently tremendously dangerous.
	But it's important to more than balance that by focusing on how these
capabilities can be used for medicine, increasing wealth per capita
while cleaning up the environment, and a whole range of beneficial
applications that would naturally occur in the civilian sector, and
would naturally encourage international cooperation.
	Because at this point, with all that we know of the potential dangers
of technology and all the discussion there has been of this one, the
potential nightmare scenario is not that we plunge ahead blindly
ignoring the dangers and have them take us by surprise. The nightmare
scenario is that a focus on the destructive potential of new
technologies leads to them being classified as a military technology,
leading to international rivalry with a military flavor, and
precipitating a war.
	The optimistic scenario is one in which we keep in mind, as has
happened so far, the potential for abuse, and the things that would be
dangerous if you did them, and consider how we can avoid doing
anything like that, while focusing on goals that encourage
cooperation.
Nanotechnology Resources: Where To Learn More

First General Conference on Nanotechnology:
Development, Applications and Opportunities
November 11-14, 1992
Palo Alto, California
Sponsored by the Foresight Institute
Box 61058, Palo Alto, CA 94306,  USA
(415)324-2490    
Fax (415)324-2497
Internet: foresight@cup.portal.com

Institute for Molecular Manufacturing
A nonprofit foundation supporting nanotechnology research
555 Bryant Street Suite 253
Palo Alto, CA 94301
(415)852-1244
fax: (415)852-9098
Internet: kshatter@tmn.com

  "The Assembler Multitude"
 An open discussion of New Technical Developments in Nanotechnology
and  Social Consequences and Dangers of Nanotechnology
Meets every other Monday,7:30 PM to 9:30 PM
University Lutheran Church
1611 Stanford Avenue, Palo Alto
Organized by Ted Kaehler 
Net:  Kaehler2@AppleLink.Apple.com     
Mailing address:  
20525 Mariani Ave. MS 301-3D
Cupertino, CA 95014

American Information Exchange (AMIX)
(Extensive library of nanotechnology-related documents available for
online purchase)
1881 Landings Avenue
Mountain View, CA 94043
(415)903-1000
fax (415)903-1094
.