The Proceedings of the 1989 NanoCon Northwest regional nanotechnology conference, with K. Eric Drexler as Guest of Honor.

Go to NanoCon Foreword
Go to NanoCon Proceedings Table of Contents
Return to Jim's Molecular Nanotechology Home Page


Van der Waals cylinder-and-sleeve bearing

Seattle, Washington
February 17-19, 1989

Guest of Honor

K. Eric Drexler, Visiting Scholar, Stanford University, and author of Engines of Creation

Guests (in alphabetical order)
Greg Bear, author of BLOOD MUSIC
Dr. Gregory Benford, University of California
Dr. John Cramer, University of Washington
G. Louis Roberts, Boeing Computer Services
Dr. Bruce Robinson, University of Washington
Dr. Nadrian Seeman, New York University
Marc Stiegler, Xanadu
Mike Thomas, Boeing Computer Services
NANOCON Chair: John L. Quel
Vice-Chair: Dr. Jim Lewis

This document was transcribed and edited by Jim Lewis and John L. Quel between March and May of 1989. This WWW version was created by Jim Lewis between March and June of 1996.

© Copyright 1989, by NANOCON
Published by NANOCON. All rights reserved under international and Pan-American conventions. However, we permit copying for educational purposes only provided that the source of the document, NANOCON, is acknowledged at all times.
Cover, diamond cube, and rod logic graphics © copyright 1989, by K. Eric Drexler


The theme of NANOCON was nanotechnology, the predicted development of the technological capability to manipulate matter at the atomic scale and to build complex devices to precisely atomic specification. Current technologies that are leading in this direction include protein engineering and scanning tunneling microscopy. However, this conference made no pretensions to covering either of these areas, each of which has been the focus of several recent conferences. Instead it featured brief glimpses of three very disparate topics: The conference focussed on a diverse set of ideas advanced by K. Eric Drexler in the book Engines of Creation, published by Anchor Press/Doubleday in 1986.

The idea of atomic-scale manufacturing is an outgrowth of earlier suggestions [K. Eric Drexler. 1981. "Molecular engineering: An approach to the development of general capabilities for molecular manipulation" Proc. Natl. Acad. Sci. USA 78:5275-5278]. Drexler discussed his recent work on design considerations for molecular-scale mechanical computers, which are expected to produce the equivalent of a current main-frame computer the size of a bacterial cell.

Other work on molecular engineering by Bruce Robinson and Nadrian Seeman, researchers at the UW and at NYU, used a different approach: base pairing of specially-designed DNA segments to provide a self-assembling 3-dimensional matrix to which conducting organic polymers could be attached. Such constructions could provide nanomanipulators or an electronic computer, somewhat larger than Drexler's proposed mechanical computer, but still molecular in size and faster because it would be electronic rather than mechanical in operation. [Bruce H. Robinson & Nadrian C. Seeman. 1987. The design of a biochip: a self-assembling molecular-scale memory device. Protein Engineering 1:295-300.]

The second major theme of NanoCon was the use of hypermedia to manage information. G. Louis Roberts of Boeing Computer Services discussed a hypertext system developed at Boeing to access a huge library of visual information. Marc Stiegler described the Xanadu Hypertext Project based on the proposal of Theodor Nelson in Literary Machines. Xanadu will constitute a vast library of published works, interconnected by two-way links.

The third major theme encompasses the effects of such technological revolutions. Long term effects upon society were discussed by a panel that included two prominent science fiction authors, Greg Bear and Gregory Benford. An additional panel considered possible developmental paths toward these technologies.

James B. Lewis, PhD

Table of Contents



I. K. ERIC DREXLER: An Introduction by Dr. John Cramer

B. Discipline in an Interdisciplinary Field
C. Nanotechnology is Engineering, Not Science
D. Building with Atoms
E. The Paths to Nanotechnology
F. Building with Assemblers
G. Nano-Scale Mechanical Computers
H. Cell Repair Machines
I. "Mega-Brain" Computers
J. Assemblers and Industrial Production
K. Nanotechnology in Space
L. Potential for Abuse
M. The Emergence of Nanotechnology
N. Questions and Answers

III. MOLECULAR MANIPULATION and MOLECULAR COMPUTATION A. Eric Drexler: Rod Logic for Molecular Computing 1. Constraints on Molecular Design
2. Molecular Mechanics
3. Transmitting Signals in a Nanocomputer
4. Computing with Sliding Rods
5. Thermal Noise
6. Programmable Logic Arrays
7. Thermodynamic Reversibility
8. Data Registers
9. Mechanics
10. Computers from Molecular Mechanical Components
Questions and Answers
B. Ned Seeman: Nucleic Acid Structural Engineering 1. DNA as a Polymer
2. Branches in DNA
3. Making and Analyzing DNA
4. Building in 3D with DNA
5. A DNA Computer
6. A DNA Nano-Manipulator
7. Engineering with Weak Bonds
C. Bruce Robinson: Conducting Polymers for Electronic Switches 1. Designing Polymers
2. Conducting Polymers
3. Chemical Switches
4. Conducting Polymers & DNA Scaffolds
D. General Discussion of Molecular Computers

IV. HYPERTEXT PUBLISHING A. Eric Drexler: Hypertext and Nanotechnology
B. Marc Stiegler: The Xanadu Project 1. The Hypermedia Shopping Plaza
2. The Xanadu Information Server
3. Links in Xanadu
4. First Xanadu Configuration
5. Applications of Xanadu
C. Louis Roberts: A Hypermedia Image Access System 1. Overview of the System
2. How the System is Used
3. Other applications
4. Lessons Learned
5. Discussion
D. Hypermedia Panel Discussion 1. Eric Drexler:
2. Louis Roberts:
3. Greg Bear:
4. Mark Stiegler:
5. Mike Thomas:
6. Discussion

B. John Cramer
C. Greg Benford
D. Eric Drexler
E. Greg Bear
F. Discussion

B. John Cramer
C. Vonda McIntyre
D. Mike Thomas
E. Eric Drexler
F. Jim Lewis
G. Bruce Robinson
H. Discussion

VII. APPENDICES Appendix A: A View of NanoCon by Dr. John Cramer
Appendix B: An Additional Perspective on NanoCon by Dr. Jim Lewis
Appendix C: The Cosmic Pie by Tracy Harms
Appendix D: Branched DNA and Nanotechnology by Dr. Ned Seeman Abstract
Nanotechnology as an outgrowth of biotechnology
Nucleic acids as components of nanotechnological systems
Construction with nucleic acid branched junctions
Restriction-growth cycles
DNA networks as molecular scaffolding
Appendix E: Conductive Polymer Structures by Dr. Bruce Robinson
Appendix F: Hypermedia Photo Retrieval System by G. Louis Roberts


I wish to give special thanks to my partner, Dr. Jim Lewis, and the four people who worked with me from the beginning to make NANOCON a success: Dr. John Cramer, Mike Thomas, Rick Burton and Steve Salkovics. They were the primary contacts with the business and academic communities, giving it the credibility it needed to succeed. Without their efforts, organizing the conference would have been out of the question.

Special thanks also needs to be given the guests, especially K. Eric Drexler. Without their support, this conference would not have been possible.

Jane Hawkins, Kathleen Critchett, Marcie Malinowycz, and Eileen Gunn provided invaluable assistance with the registration process.

Finally, Grant Fjermedal and Steve Salkovics are to be commended for their work in recording the entire conference, and David Gagliano and Rick Burton for help with printing and copying this document, making the proceedings available to all who are interested in these extraordinary ideas.

A Note on the Proceedings

These proceedings have been edited for conciseness and clarity. People tend to be more repetitious when speaking than when writing. In editing this document, we saw little reason to transcribe verbatim what was said in all cases. We have, however, struggled to keep the meaning and emphasis of what the speakers said intact. Possibly offensive comments have been edited.
The speakers used slides or other visual materials to illustrate their talks. Only a very few of these could be included either with the talk or in the Appendices. We have tried to explain the references to the slides whenever possible, but this compromise will inevitably prove inadequate. In the case of Eric Drexler's rod-logic talk, additional material is available from the Foresight Institute, as noted below.


NANOCON was a hybrid and an experiment. Neither a fan gathering (as the use of the suffix "con" would imply), nor a true scientific conference, it shamelessly borrowed elements of both. I doubt that anything quite like it had been tried before, at least with any regularity. NANOCON sought to bring together people who rarely interact: scientists, technicians (by that term, I mean those who apply science to business), and writers, to see what would happen. NANOCON attempted to please everyone, and I think did not do such a bad job of it.

The three day conference was a forced march through a lot of material. The goal was to achieve by the close of the conference a state in which the seeds of future thought would be planted. It is our hope and intent that in the coming months and years those seeds will grow.

I initiated the events that led to this conference because I believe the history of human progress is the history of evolving ideas and institutions. In addition, I believe that while society cannot be reduced to economics, there is an undeniable under-pinning of economics to social action. The effect of "nanotechnologies" (I prefer the plural form because the singular, which I reserve for the final achievable goal, has already become overladen with emotionalism and utopianism) on both will be profound and lasting; indeed, in the most fundamental social sense of all -- upon our whole understanding of what it means to be human.

Can such all encompassing changes be understood? Can we, as individuals, prepare for them? Can the ideas be explained and communicated responsibly, or are they doomed to remain the staple of bad science fiction stories until it is much too late?

I had been concerned about these questions ever since I read Engines of Creation on Christmas Eve, 1986. NANOCON grew out of my frustrations that little was being made in enlarging upon the vision described in that book. NANOCON grew out of my despair over the enormous difficulties involved in communicating that vision, even as we understood it today; difficulties that threatened to drown the ideas under waves of foolishness and fear. I felt that something different had to be tried, something that would make these extraordinary possibilities more accessible to a broader audience than the traditional "futurephiles." Some way had to be found to get beyond the "Giggle" stage -- and that would require more than the usual intellectual tilt-a-whirl "cons" traditionally supply. The problem as I saw it was not one of intelligence or knowledge, but of thought. And by thought, in this context, I mean the willingness to engage in conceptual exploration.

Since that Christmas Eve, I have had many talks regarding nanotechnology with people differing widely in career, income, religion, age, and education. While I expected few to embrace the ideas, it was disheartening that the vast majority of responses could be classified into but three simple groupings:
1. Nanotechnology will never happen (or so far in the future it might as well be "never").
2. Nanotechnology is wrong, if "it" does happen -- that was a very common response.
3. Nanotechnology is nothing special. I attributed that response to "end of the world" burn out, an attitude for which I have a lot of sympathy.

These depressingly unimaginative attitudes seemed to me to all sum to the same underlying feeling -- "Don't bother me: I don't want to think about it."

Why were all these good, honest, intelligent people, so loath to distance themselves from rash moral assessments where understanding was so obviously lacking? Why so little motivation to examine the assumptions which underlie their feelings? Why the eagerness to make all-encompassing pronouncements where not a single argument or fact could be given to sustain their statements? It seemed to me that they could understood the words, but the logic or syntax, if you wish, of their thinking did not allow for consideration of anything beyond their own highly restricted domains of comfort.

I concluded that understanding the implications of nanotechnology is most emphatically not a matter of education or intelligence. It is a matter of how we use our minds and our emotions, and most people clearly use them badly, whatever their innate mental capabilities.

In structuring this conference, we tried to give the participants as much of the fundamentals of these profound future changes as time would permit and put off as long as possible normative considerations. We also attempted to make these prospects as concrete as possible. That was why economic considerations were so vital. Economics does bring a refreshing earthiness to speculation. By requiring us to ask detailed questions flavored with strong dosages of reality and by inserting the "I" into the equations of social action, it can enable us to avoid the pitfalls of seeing ourselves as "cosmic spokespersons" for the whole of existence. Economic motives can be an excellent incentive for focusing thinking, something that sermonizing rarely achieves.

By exploring alternatives and possibilities, one thinks not to be "right", either in the sense of being correct or moral, but instead engages in a process that cannot be measured against arbitrary abscissas of "right" and "wrong." Such thought exploration simply "is." It is unquestionably difficult by being both a rational function and a creative process, and it is certainly uncomfortable, as judgments and proofs while not being excluded, are postponed (that is a crucial point, because I am not arguing for moral nihilism). It is tragically unfortunate that such thinking is seldom attempted.
These notions of exploration are difficult to communicate. The tendency is always to assume one's prejudices are the laws of existence, and to denounce and insult anyone who dares disagree. It seems shallow and somehow lacking to look upon thinking as a skill. That would cause us to humble ourselves before the future; to admit we are struggling to manage these awesome speculations with very limited tools; to realize that our vision of the future will always be a vague simulation, until that future is upon us. But, I am convinced such unavoidable vagueness may be sufficient and, in any event, is certainly better than nothing.

Consider the following: if an asteroid were heading towards earth, we would have at our disposal Newtonian mechanics and a good understanding of the effects of high energy/momentum impacts to argue effectively for the diversion of all necessary resources to avert global catastrophe. With little imagination, detailed scenarios could be constructed on the basis of the time left -- if the impact were a year, a decade, or a century away. We would feel confident that our audience would be extremely attentive, as our language would possess graphic precision. For once, nonsense would be drained from our discussions. For once, our thinking would be clear and unencumbered. But this speculative future, not the happiest, is a poor guide when attempting to understand the massive social changes resulting from the nanotechnologies.

One obvious difficulty is that we have no social science corresponding to Newtonian physics to communicate effectively what is to come, nor is such a "science of the future" terribly likely. Since there is no analogous asteroid to point to, we are stuck with "vague simulations", where equations are few and untrustworthy. Abstractions, limited knowledge, and extreme uncertainty are implicit in every statement we make. And that makes the process of communication, at any level, very difficult indeed -- unless, and this is a crucial distinction, our intent is to instill fear, our goal power over the minds of others. The cautions of Dr. Gregory Benford during the social issues panel need to be pondered in depth. Let no one doubt for a moment that the ideas implied by nanotechnology promise a global field day for the ignorant, the incompetent, and the irresponsible, but the option of closing all discussion on the matter is much worse, even if it were possible to do so.

Despite the problems and the misgivings, NANOCON courageously aimed to put those future thoughts and communications on as firm a basis as possible, given our present knowledge. Now that NANOCON is over, it is our hope that those who met during the conference will remain in contact, and that the cross fertilization of ideas concerning these technologies will continue and spread. It is our hope that the legacy of Engines of Creation will be built upon, so that when the future arrives, not so many decades from now, we will be in some measure prepared for it. That would be a first in human history, and it is certainly doubtful, but that is no reason not to try, and try we did.

In retrospect, I believe NANOCON will be looked upon as only an early, crude attempt at understanding and communicating what is to come. And judging from the interest that grows more evident every day, it will certainly, and quite properly, not be the last such conference.

On behalf of the people who made it possible, I thank you for your support.

John L. Quel


An Introduction by Dr. John Cramer

We can think of ourselves as standing in the trough just before the tidal wave hits; the only question is just how far away that tidal wave is. It's perhaps a unique circumstance in human history: a revolution that is going to have a profound effect on our society, and the way we do things, and the way we build things, has been anticipated in the way that it has in this particular circumstance. I can't think of another example of an instance in which a monstrous societal impact of a technology was seen coming far enough in advance that one could do advance thinking and planning. To some extent, one could say that people like Norbert Wiener and John von Neumann thought about computers before they were upon us, but while there was plenty of time, there was very little advance planning as a result of their visions. With Eric Drexler, however, I believe the situation is different.

When I read Engines of Creation, my reaction was "Of course"! It's ideas were something I had been thinking about for a long time -- in a rather vague way. Suddenly, they came into focus. The focus is that, as Scientific American said, there is an "air of inevitability" about these ideas. They are coming. The best we can do is try to understand and digest them. And towards that end, I hope that we can accomplish something at this conference other than deciding "how many nanotechnologists it takes to screw in a light bulb."

(Answer: None -- that's a problem for conventional technology. With nanotechnology, you can make a light bulb that can fix itself!)

Eric was an undergraduate at M.I.T. studying interdisciplinary science. He was all over the map studying physics, mathematics, and various aspects of engineering. He wandered into the engineering program and received a master's degree, but before he received a PhD., he wandered out again, because his interest in nanotechnology took him in a direction for which there is no PhD.
I think Eric's experience is in a way a comment upon our education system. Nanotechnology is an interdisciplinary field where so many different elements are being brought into play that no one department is willing to grant degrees in such a subject. It's a comment upon the specialization of our educational system -- "you can't do things like that!"

But their loss is our gain. It's a real pleasure to hear Eric tell us about the shape of the future.


K. Eric Drexler

I might add to the previous remarks that nanotechnology fit in especially poorly in an aeronautics and astronautics graduate program. "You want to take a course in molecular morphogenesis?" they said. "That's not part of your field. You can't."

A. The Four Challenges

Having given my introductory nontechnical talk this evening the label "The Challenge of Nanotechnology", I found myself thinking: what is the challenge of nanotechnology? I decided that there are at least four categories the challenge could be divided into:

1) The Challenge of Technological Development

The challenge of going from the technology base we have today towards greater and greater control over the structure of matter to the point where one is able to build complex (and increasingly complex) things atom by atom, including molecular machines and assemblers, which will then enable a very general control over the structure of matter.

2) The Challenge of Technological Foresight

Trying to understand what lies, not necessarily at the end, but well along this path of technological development; trying to get some sense of the lower bounds of the future possibilities. Not exploring all the things that would be possible, since that would be foolish to undertake, but trying to get a sense of a few of the key capabilities we will be able to equal or surpass.

3) The Challenge of Credibility and Understanding

Imagine that we had gone on a safari into the conceptual world of the future and bagged some big, strange looking technological "animals" and dragged them back. How would one package this information? How does one present it? How does one make things that are true sound credible and distinguish them from things that are not true (indeed nonsensical, yet sound superficially similar), and thus give people a clearer understanding of what these technological possibilities are? That understanding is the necessary foundation for dealing with the fourth challenge.

4) The Challenge of Formulating Public Policy

How do we formulate public policy based on that understanding, so that when the "tidal wave" hits, we are, in fact, as ready as we can be to deal with it?

B. Discipline in an Interdisciplinary Field

I usually start off my talks, various technical colloquia and such, with questions for the audience. These questions are intended to address the point of credibility and understanding. How many people have backgrounds in physics, chemistry, biology, engineering, computer science? I always get to comment: "My there are a lot of people in computer science here." My explanation for that is that people in computer science are used to the notion that making things very small and controlled and fast can be valuable. When they hear that more is coming they say, "Oh, yes. Tell me more."

The reason I ask this question is that nanotechnology is very interdisciplinary. It cuts across all the fields mentioned and more. But, unfortunately, interdisciplinary subjects have a way of escaping from any discipline whatever. If you don't watch out, you end up with the equivalent of Velikovsky with his book "Worlds in Collision." He wrote in that book how ancient writings "explained" how the solar system formed and that Venus was a comet coughed up out of Jupiter. And he made substantial headway in the scientific community with these theories of the past and the solar system. The historians thought his astronomy was quite interesting, but, of course, his history was bunk. The astronomers thought that the history was fascinating though, of course, the astronomy was bunk. Similarly, experts in one of the above fields tend in general to be harshly critical of any ideas that fall into their own fields, but less critical about ideas in other fields. I believe that it is extremely important for meaningful discussion of nanotechnology that ideas be subject to demanding criticism. Accordingly I strongly encourage this audience to be harshly critical of any ideas labeled "nanotechnology", starting with mine - with what I say now, and extending to anything you may hear in the future.

I will briefly outline some of the content of nanotechnology and how it relates to where we are in technology today. I will discuss paths towards nanotechnology -- the challenge of technological development. I will show some pictures illustrating things found in exploring where this technology will be in the long term. I will close with what I believe are some crucial points in understanding the challenge of public policy.

C. Nanotechnology is Engineering, Not Science

In thinking about nanotechnology, it is of vital importance to distinguish engineering from science. If you believe the media, you would conclude that when people are out on the launch pad working on the space shuttle main engines, that those people are "scientists." We are told they are "scientists", but I don't believe they are out there studying space shuttle main engines as natural phenomena, or taking samples of the metal to study the precipitation or hardening of these metals, or something like that. Instead, they are doubtless engineers, or perhaps technicians.

Here is the difference. If you ask a scientist to make a prediction about the future of a field: "What will you discover 10 years from now, sir?", and if that scientist responds that "In 10 years, I will be discovering X" then that is obviously bunk. If you already know it, it cannot possibly be a discovery. It is a contradiction.

If you asked an engineer, on the other hand, "what if we give you enough time and money, what will you be able to build in five or ten years?", you would expect a more reasonable answer. That question was posed to some aerospace people in the early 60's and they replied, "We think we can land a man on the moon and return him safely, Mr. President." They gave a cost figure, (and then they doubled it), and the budget was submitted to congress. In fact, it was done. The reason being that people understood the fundamental scientific principles, such as Newton's laws, they understood how to build fuel tanks and engines and so forth, and they had confidence that systems like this could be built and debugged.

I will argue today for a similar position with respect to nanotechnology -- that we understand fundamental scientific principles well enough to see much of what is possible, though a lot of work remains to be done. A lot of that work will have the flavor of science in finding out details and sorting out what works. But the lack of effort in what I call exploratory engineering has left us, as a society, with a huge blind spot. Scientists don't look too far ahead, because you can't in science. Engineers don't look too far ahead, because they are not paid to. If you were an engineer and went to your boss and said,"Just give me a year to think about what we will be able to build in another 20 years." The response would be: "No.", at least in this country. However, I sometimes get the impression you would get a different answer in Japan, or at least a somewhat greater time horizon.

D. Building with Atoms

I have been trying to fill a little bit of that blind spot in the area of nanotechnology by trying to get some understanding of what could be built with tools we don't have as yet. If you look at present day technology, a modern research and development laboratory would look something like this: people manipulating atoms -- huge, thundering herds of atoms, statistical populations of them, stirring them around, heating them, reshaping them by pounding, whirling, and so on. By these techniques, we make all sorts of impressive things. We make sophisticated devices like transistors. We are advancing in semiconductor technology and are getting very good at miniaturization. Nowadays, we make very fine features on chips by processes that include, for example, oven baking, and from these techniques we get some impressive, extraordinarily useful devices, such as microprocessors.

I will argue that it is possible to put entire mainframes, with memory and disk drives, in a cubic micron. Yet, the smallest features on a contemporary chip are several microns across.

Here is an example of an intermediate stage on the conventional path of miniaturization, that is, trying to use large technologies to "build down." This example is near the limits of scale for that process that has been achieved in recent years; while not quite at the limit, it is close. It is the surface of a salt crystal that has had lines drawn on it by a tightly focused electron beam. This was done by the Naval Research Laboratories.

In the upper left-hand corner, you see an 18 nanometer scale bar. In an etymological sense, you could call this "nanotechnology" because it is on a nanometer scale. The next figure shows the contrast between that and the results of the kind of "bottom-up" nanotechnology that I discuss in Engines of Creation:: the technology that I think is the cause of the kind of excitement that has resulted in this conference, as opposed to the kind of excitement that motivates further miniaturization in the computer industry. Both of which are important but are on very different time scales and scales of consequence.

Consider one cubic nanometer of diamond and imagine what it looks like to a "nanotechnologist." If the planes in the figure below are cutting cleanly between the planes of carbon, it is slightly less than one cubic nanometer. Diamond has an extraordinarily high number density of atoms, but most materials have on the order of 100 atoms/cubic nanometer. If each of those atoms is something you can think of as a building block, then it becomes clear that one can build relatively complex things in a single cubic nanometer. Now a cubic micron, which is currently considered fairly small in microtechnology, is a billion cubic nanometers. To a nanotechnologist, therefore, a cubic micron is a vast amount of space to work in.

One cubic nanometer of diamond, containing 176 atoms.
A cube 100 nm on a side would contain 176 million atoms

To give a sense of the kind of structures one thinks about in looking at advanced nanotechnology, things based on assemblers able to build general structures atom by atom, one might build a path for transmitting force, or part of a lever to transmit torque, using a structure composed of carbon atoms. In the example, hydrogen atoms are left out for simplicity. The structure is a nanometer long and contains a countable number of atoms, each one of which is in a precise place. If you remove one of them, it would no longer work.

Another example of an atomically precise structure, again not showing all the atoms, is a roller bearing -- nothing a chemist would think about making today, but something one can think about making with assemblers [see cover]. It illustrates the principle that one can get smooth rotary motion despite atomic "bumpiness", if you have the atomic "bumps" on the two surfaces mesh in gear-like fashion. This answers one concern that one might have about friction in molecular mechanical devices of this second generation.

E. The Paths to Nanotechnology

Where we are today on the road to nanotechnology? DNA can be considered as an "engineering material" and certainly has a possible role in the construction of molecular objects. However, the interest most people have had in it is not in its engineering properties, but in its informational properties.

The overwhelming reason people have done genetic engineering is that you can take a piece of DNA and insert it into bacteria, where the DNA is transcribed to make RNA molecules that contain the same information. The RNA molecules in turn bind to ribosomes and the ribosomes read the RNA. The RNA base turns out to have two bits of information. RNA reads three bases at a time, thus six bits at a time, that is, RNA reads a series of six-bit words. One word might say: "start" (... making a protein with a particular amino acid); another might say: "add" (... another particular amino acid); and so on. The result is a growing polypeptide chain. Finally, the six-bit word "stop" (... release the chain and, perhaps, start over again) is reached.

The reason people are interested in programming ribosomes to produce proteins is that proteins can serve as components of molecular machines. A protein doesn't remain as just a loose, floppy chain, but instead folds up into a three dimensional configuration, in which the interior is a closely packed arrangement of side chain atoms. Proteins look like random, haphazard things, but every protein of that sort will roll up in the same way to make an object which, at least in the case of some proteins that serve structural roles in bacteria, has about the stiffness of a piece of wood, or a piece of epoxy engineering resin. Proteins are molecular objects. They are pieces that fold up and then go together to make more complex objects. An example is a collection of two protein molecules and a strand of DNA. They are like this because of Brownian motion. The random motion of things suspended in solution under thermal agitation banged these molecules together in all possible positions and orientations. Eventually they bumped together in the right orientation, because these molecules had complementary surfaces at that point: bumps matching hollows, patterns of electric charge matching, and so forth. This resulted in selective stickiness and self-assembly.

A very powerful principle that will be used, I believe, in developing nanotechnology, is the principle that if complex molecules are made with complementary surfaces, they will self-assemble to make complex structures.

We see that in nature: here is an example of a complex assemblage of protein molecules. It looks like something out of a Grade B science fiction movie, or an industrial small parts catalog (an analogy more promising from a nanotechnology perspective). It is a collection of protein molecules stuck together with some DNA in the head. In certain conditions it falls to pieces, pieces which can be made to reassemble by putting them in the right conditions of temperature and solution composition.

The assemblage is a T4 bacteria phage, a bacterial virus. You can take it apart to individual protein molecules which have been assigned numbers by the people working in the field and again they will assemble in the right conditions to form a complete device. I call it a device because it will selectively stick to the wall of the bacterium and act like a spring-loaded hypodermic syringe: the base plate helps to make a hole in the bacterium cell wall, the sheaf contracts and drives the core down and injects the DNA. The DNA is then copied to make more DNA and is transcribed to make RNA. The RNA re-programs the ribosomes of the bacterium to make more of the T4 proteins. The proteins and the DNA spontaneously assemble inside the bacterium to make more viruses. This process finally leads to the production of proteins that break up the bacterial cell and completely destroy it, releasing the viruses.

The above example is depressing if one is bothered by the presence of parasitism in the world on all known size scales. It is also an example of a molecular machine. If you look at nature you find a variety of molecules, you find a variety of components that are very tempting to think about from a mechanical engineering point of view.

There are bearings: a molecule that is held together by a single sigma bond and does not have any interference between the two parts of the molecule will allow one part to rotate freely with respect to the other. That bond can serve as a bearing with a load strength of a number of nanonewtons. That is a healthy load for things on this scale, though perhaps not all that one would want, which is why other bearings are being looked at.

There are rotary motors: Some bacteria can swim, and while advanced cells can swim using flagella, bacteria have simple helical rods of protein, rods that rotate: where the rod attaches to the cell wall there is a variable speed reversible motor. This motor has been described in the literature as a "proton" turbine.

There are also linear motors, like the molecular fibers that drive muscle.

We already have an example of more complex machines as well, a numerically controlled machine tool called the ribosome.

What this suggests is that there are paths leading from engineering folding polymers, such as proteins or things like proteins that it might be easier to design the folding of, such as perhaps properly configured DNA molecules. These paths lead from those sorts of systems, as we improve our ability to design them, to building molecular machines.

Now, a lot of people have said to design a protein from scratch is an extraordinarily difficult problem -- yet it has been done in the last year ["Characterization of a helical protein designed from first principles" L. Regan & W.F. DeGrado. 1988. Science 241:976-978]. In Engines of Creation, I waffled on how long that might take. Now the milestone has been passed.

Along that path there is still a lot of improvement to be made in design techniques. When they are improved one could build machines, not just things that fold, but things that fold to form objects that do something, and use those machines to build better machines. We know by looking at nature that molecular machines can, by holding reactive molecules at particular positions and orientations, perform chemical operations to build up complex structures in specific ways. That is the function of enzymes at one end of a spectrum of machines. If you have more flexible, programmable machines, they start to look more and more like general purpose assemblers. One can use low end machines to build better machines, and better machines until one has reached the kind of assemblers that form the bulk of the subject matter of Engines of Creation.

There are other paths. One could work in non-biological chemistry, such as supramolecular chemistry, which is the chemistry of the assemblage of molecules. Three people shared a Nobel prize recently for their work in that field. Again, since this slide was prepared.

And one can, perhaps, extend the technology of the scanning tunneling microscope (STM) or its relative, the Atomic Force Microscope (AFM), for molecular manipulation. The STM is a device that can position a tip to atomic precision near a surface and can move it around. Since this slide was done, people have demonstrated ["Atomic-scale surface modifications using a tunnelling microscope" R.S. Becker, J.A. Golovchenko, and B.S. Swartzentruber. 29 January, 1987. Nature 325:419-421] the ability to get atoms on a tip by touching it near a surface at one place and evaporate them off the tip at another and create a new "glob" on the surface that seems to be a single atom. Unfortunately, the last I heard it only worked on germanium and the Bell Laboratory workers were unable to "call their shot", i.e., they were not able to see where it goes. The STM is not something at this time that can build nanomechanisms.

Also, since this slide was done, at IBM Almaden ["Molecular manipulation using a tunnelling microscope" J.S. Foster, J.E. Frommer, P.C. Arnett. 1988. Nature 331:324-327] people have scanned a surface in an organic liquid with an STM tip, then placed a voltage pulse on the tip and apparently electrically excited these molecules, made them reactive, and bonded them to the surface. This resulted in nanometer scale "blobs" that were visible when the surface was later scanned. This may be very useful for building computer memory. Again, however, they were not able to make what a chemist would consider a specific modification. To do that, one may need to create a hybrid technology: develop molecular tools through one of the previous methods, and bind them to the tip, for example, of an Atomic Force Microscope, to give it greater specificity of action then these metallic or ceramic tips do today.

F. Building with Assemblers

A key point in thinking about these enabling technologies is that from a longer perspective, from the point of view not of the challenge of technological development, but from the points of view of technological foresight and public policy, it doesn't matter what path is followed. All paths lead to the same place.

When you have assemblers, you build assemblers. And what kind of assemblers you build does not depend upon the things that led up to the assemblers. My standard metaphor for this is, look at aircraft today, such as a Boeing 747. We note a certain shape to the wings, and a certain composition to the metal, which has no necessary connection to the shape of the wings and the composition of the cloth in the Wright brothers' original aircraft. The Wright brothers brought us into the domain of a new technology, but what we do there depends upon the tools that we have today and the design ability that we have today, not how it all started. It will be the same with full-blown nanotechnology.
As I discussed in Engines of Creation, there are very strong reasons for thinking that assemblers can be made to work based on pointing to things a lot like them that already do work. Chemistry shows us a wide range of reactions that can be made to occur when molecules come together in the right positions and orientations. Enzymes show that if you hold reactive molecules together, in a particular position and orientation, you can get a particular reaction to occur. What is needed to build complex structures is systematic positioning of molecules to make reactions occur in very specific and very complicated patterns. That is the core of nanotechnology. That is what assemblers will accomplish by using the kinds of tools we are already familiar with. The important addition is that, instead of being a specific jig that can only catalyze one reaction, as an enzyme is, we are talking about things that can do programmable positioning; something that is a general purpose, flexible tool for construction. And, as icing on the cake, it will then be possible to drive a lot of these reactions using external sources of energy, such as voltage, or even mechanical force by means of the molecular machines involved.

And that leads to this slide, which is really intended to summarize assemblers and what is important about the case for them. Here is the summary of what it means. What assemblers will give us is thorough and, as I will argue, inexpensive control over the structure of matter. That means that in contrast to today, where technology is very strongly limited by fabrication, which conditions everything that we do, there are a tremendous range of things that one can design that one would not normally think of trying to design, because it would be ridiculous to think of being able to build it. A large part of what I have done is simply to ask that if we can build almost anything that makes physical sense, what then becomes possible? And then explore the very elementary possibilities that are opened up by that fabrication capability.

It appears that assemblers can build anything that makes chemical sense, and at that point the main limits will be physical law: what does natural law actually predict to exist and function and what will be the design capabilities -- what are we clever enough to design? I have been trying to stay well within the limits of physical law, indeed well away from those limits, so as to have things that are easily defensible and will clearly work. That involves things that are very far from the limits when you have advanced design capabilities, more people working in the field, the ability to test questionable ideas against nature to see whether they work or not, and so forth. All these things will enable people to push much closer to the frontiers of the possible than one can do with any safety in exploratory engineering today with limited resources and without the possibility of experimental feedback. Therefore, the things that I design are "stupid." They are clunky. In a moment, I will discuss one of the stupidest and clunkiest devices of all -- a mechanical nanocomputer.

G. Nano-Scale Mechanical Computers

Ordinarily, in thinking about making small things, such as small computers, one says "Well, if computers are electronic devices, and if we are going to build molecular computers, that means molecular electronic devices." So, we have conferences on molecular electronic devices, of which there have been a number. I very strongly suspect that some of the designs that have already been presented at these conferences or will be presented in coming years, will work. They will be vastly superior to the kinds of computers that I am designing. I would guess, off hand, that these molecular electronic computers will be three orders of magnitude faster than my molecular mechanical device.

The problem with designing molecular electronic systems, however, is that one must deal with the quantum mechanical properties of electrons and very small, irregular structures. If you look at the current work in trying to understand the structure of high-temperature superconductors, where we know where all the atoms are, all the relevant fundamental laws of physics, there is still a Nobel prize for any theoretician who can figure out how they work. Despite all those favorable factors, no one has done so and defended his theories in a fully credible fashion. Here we are talking about systems that are, again, complex and electronic, and while you may have something that will work, as the superconductors work, but you may not be able to argue that it will. Therefore, I don't try to argue that, for any given design, they will. I simply say "probably one of them", and I wander off and instead design "stupid" things like a mechanical nano-scale computer.

Computers did not start with electronics. They started with machines, though they didn't quite get off the ground with that technological medium. The illustration is a picture of part of a machine designed by Charles Babbage back in the middle of the 19th century, the analytical engine. If Babbage had had more time, more money, and perhaps better machinists, he would have built the world's first programmable computer back around 1860.

And then in subsequent years as systems were refined and the Swiss got into the business and displaced the English and there was a national hue and cry about the loss of the computer industry to Switzerland (where they are better at miniaturization of mechanical devices), we would eventually have had computer science departments emerging out of mechanical engineering departments. Everyone would then have thought that software was fundamentally a branch of mechanical engineering, instead of being confused, as we are today, that it is a branch of electrical engineering.
The Babbage machine would have been quite slow. However, it turns out that if you scale a mechanical system down by a factor of 10, it becomes 10 times faster in the frequency of operation. If you scale a mechanical computer down by a factor of a million, it becomes a million times faster. If you then make it out of stronger, lighter, stiffer materials, that helps as well. Another nice side effect is that you reduce the volume by a factor of 1018.

One would transmit signals in such a device by moving rods: the rods would have knobs on them that mechanically interact, blocking and unblocking each other. This turns out to give one the ability to build things that are analogous to transistors. An example would be two rods, one of which would move if, and only if, the other rod was out of the way. This is like a transistor in which current will flow if, and only if, the right voltage is on another conduction path. One can look at complex systems built out of these. You can analyze them using Newtonian, instead of quantum, mechanics. It may be a "stupid" design, but undoubtedly it could be made to work, which is my ambition. Not to make it work as such, but to give convincing arguments that it could work and, therefore, one could do at least this well or better. Now you have a conceptual building block that can be used for thinking about what nanotechnology can do.

H. Cell Repair Machines

In Engines of Creation, there is a discussion of cellular repair systems: medicine based on extremely small computational devices hooked up to molecular scale sensors and devices that can do operations on molecules -- taking them apart, synthesizing them, and so on. To do that, one is interested in how small one can make a general purpose computer, given that one could put all the atoms where wanted. To estimate that, take the scale of rod-logic devices, and look at the number of devices and conducting paths and so on, in a simple, ancient, bottom-of-the-line Intel 4004 4-bit microprocessor, the first processor that saw any substantial commercial use. If one goes through that exercise and asks how large a block is required to hold the equivalent of an Intel 4004, something on the order of a few tens of nanometers will about do it. If you estimate the volume of memory devices, one would conclude that a roughly comparable volume will hold one kilobyte of RAM. Comparable volumes will hold roughly a hundred kilobytes of tape memory, and a lot of molecular sensors and molecular machinery suitable to characterize and manufacture a wide variety of macromolecules.

That whole package of "stuff" combined with unspecified software, which may be a greater challenge than the hardware in the long run, is something that might be described as a "repair device." If you take that collection of objects, it will look very small in comparison to the diameter of a 20 micron size cell. Yet, a cubic-micron computer is on the order of a contemporary main-frame: tens of megabytes of random access memory and some hundreds of megabytes of fast tape memory. It turns out that this gives one a database with more information than was used to construct the cell in the first place, along with a number of mainframe computers, and a (very) Local Area Network, sufficient to connect to 100,000 similar repair devices. If you can come up with the software, which is another question I am not addressing, all this would function in less than one percent of the volume of the cell. If the software problems can be handled and if one can use this same technology to figure out what needs to be done by doing a very thorough job of characterizing biological materials, then one should be able to bring surgical control to the molecular level and begin to repair tissue at a level that medicine cannot begin to deal with today.

Today, one mode of therapy is to throw drug molecules into the body: they diffuse around and selectively stick to things and perturb the behavior of the biological structures. The other major mode of therapy is to take an enormous piece of metal and hack through tissue, ignoring entirely where the cells are. The result is that the body abandons its dead and self-heals -- if things go well. Technology like this, however, would bring surgical control to the molecular level, which means tissue could be either healed or reconstructed -- again, if you have the software to handle the task, and there are arguments that such is achievable, though the arguments are in the software domain.

These ideas led to a column in Scientific American in January 1988, which had an illustration of a possible repair device. These repair "submarines" were drawn rather larger than I told them to, so the devices pictured could easily hold a gigabyte of memory, and thus think rather deep thoughts about the fat they were chewing.

I. "Mega-Brain" Computers

Now, what happens when you have a lot of computers? If you can make a mainframe in a micron, and I might add that the clock rate estimate for these mechanical computers is moderately faster than a contemporary CRAY (about a gigahertz clock rate), though technological progress will shortly give us faster computers. Take a cubic centimeter volume, allocate half of it to mainframe-cubic micron devices and the other half to cooling and communication channels, you find you have room for about 0.5 trillion computers, possessing far more computational capacity than has been built in the world to date.

Cooling is a problem, but it is possible. If you have these devices executing one "gram-mole" gate operations/second, which is a chemist's idea of a round number (6x1023), the result is about four kilowatts of power which can be dissipated by pouring in a couple of liters of cool water per minute, and getting a couple of liters of warm water out.

Another result from this exercise is that one can make a very crude estimate of the computational capacity of the human brain. This crude estimate, which is argued to be grossly generous to the brain, results from the idea of considering a synapse operating for a millisecond to be equivalent to a gate operation during a clock cycle. This says nothing about software, however. A computer with all the computational power one could want can do nothing without the proper software. Comparing only raw computational capacity, however, shows this device to be in the mega-brain range. There are all sorts of conceptual problems involved in artificial intelligence, artificial neural networks, and trying to make machines think. There are obvious software difficulties involved, but I would like to point out that almost all the work done to date has been on machines that are not even in the monobrain range, but instead in the microbrain range. It is remarkable that anything has been accomplished at all. It is entirely possible that sliding up several orders of magnitude in raw computational power might make the task of AI easier.

J. Assemblers and Industrial Production

This whole cubic centimeter is something "big" in the field of nanotechnology, and this raises the question of making large things. An example of a lot of small things is a "paste" of E. Coli bacteria, a paste that started out as a single genetically engineered E. Coli some modest number of days previously. Exponential growth can take you from the scale of a single nanomachine (a bacterium) to planetary masses (if you could supply the necessary raw material and energy and get rid of the waste heat) in a matter of days. This is in terms of raw reproductive capacity. If you have devices that can reproduce themselves, you have a very powerful industrial technology base for making things. In fact, if you look at assemblers, they are very well positioned to do just that.

Previously we have shown bearings and the like that would make up the parts of an assembler, and discussed some general aspects of the software problem. Let us consider assemblers proper. The kinds of advanced nanotechnology assemblers I talk about will be very much like industrial robots: they will be special purpose machines. How will they be built? Think of a rigid, jointed, programmable "thing", like an industrial robot, but with parts roughly a million-fold smaller, thus roughly a million fold faster in characteristic frequencies, able to do a million operations per second. Now, take that technology which is 1018 times more compact and a million times faster, and let it work not with prefabricated pieces, but with the building blocks of matter, atoms. One can build anything that can be built with atoms, if one is careful about unit operations and chemical reactions.

The devices would operate in a world that we often forget is rich in identical, prefabricated parts. In Japan today, there is at least one factory where robots assemble parts into more robots of the same type. This is not a process that gives a tremendous economic advantage because one still has to make the parts and most of the expense is in making the parts rather than in the assembly. But instead of having to have a world of factories and mines and so on to make the parts, what if the parts are abundant molecules or can be had at the price of industrial chemicals? Under those conditions it is much easier to imagine a device making a copy of itself, as we already seen the robot factory doing as a proof of concept. Rough calculations that I went over at my class in Stanford (Spring 88), indicate that a device like this can be made that can build a copy of itself in something like a thousand seconds. That is about the time it takes a bacterium to replicate. Actually, the original figure was 100 seconds, but some conservative factors were added to the calculation.

If you can take raw materials, and an assembler, and end up with a lot of assemblers, and have the assemblers work in parallel, then nanotechnology should be able to make big things.

The Eiffel tower was once the tallest structure that human beings had built on the face of the planet. In later years, this steel structure had to be retrofitted with warning lights for aircraft. If you were to build an analogous structure, not out of steel, but out of well bonded carbon structures like diamond, and ask how tall you could make that tower, the result is impressive. Aircraft warning lights would again be needed, but they would be around the base. Around the sides and top, you would need a traffic control system for dealing with possible satellite collisions. Such a tower would extend well beyond the atmosphere.

This structure is rather larger than a redwood tree, which is already built by molecular machines, but on a log scale, it is not that much larger. That is probably not the best way to get into space, however. It gets you out of the air where you can see a lot, and there are advantages to that, but it doesn't get you going anywhere. To do that, spacecraft such as the shuttle have been used, with mixed success.

K. Nanotechnology in Space

What are the implications of nanotechnology for things like spaceflight? Today, spacecraft are a fairly marginal technology. We are pushing the limits of the strength of the materials that we can fabricate reliably into structures of this sort. We are pushing the limits of the reliability of operations, because we require vast numbers of people making things with small margins of safety, so that a small flaw can destroy the entire space-craft. The amount of labor required is incredible. By comparison, the input of raw materials is trivial. The energy required, by the present standards of launch cost, is essentially negligible.

If you are able to make complex structures, atom by atom, you are not going to be sticking human hands into the process. There is no point in sticking your hands into a bunch of assemblers. Therefore, there is not much role for human labor, so the labor cost is very small. An analogy is the production of wood. In wood production, one takes solar energy (and nanomechanisms can certainly build effective solar collectors), and abundant raw materials. One can build things out of carbon (that is already too abundant in the atmosphere) possessing something like 50 times the strength to weight ratio of what the space shuttle was built from, and produce those things in intricate shapes for a cost per pound on the order of, perhaps, cordwood. If that can be done, then spacecraft can be built that can fly much higher, faster, and further than anything that can be built today. In addition, costs will be vastly lower, margins of safety will be substantially higher, and reliability much greater. I emphasize such things because there is nothing "small" about nanotechnology. Some of its consequences will be far removed from the domain of small things.

One of the consequences will be that the space frontier will be opened. If there is routine, inexpensive access to space, materials among the asteroids can be used. Out there are enough raw materials to bury all of earth's continents kilometers deep. That means that what is out there is an awful lot compared to what we are using down here. Space is rich in raw materials even if you just use the rubble left over from the formation of the planets.

In space, there is also the sun -- our very own nuclear furnace. The sun puts out every second a substantial fraction of a kilogram of energy per capita for everyone in the human race. That means there is a lot of energy out there, most of which plunges past the planets into interstellar space. If there is access to materials in space, and you have already amortized your R&D costs, then you can cheaply produce hardware that produces more hardware at a very high rate.

Today, NASA's idea of an ambitious thing to build in space is a few tin cans in orbit. A more ambitious idea, that was discussed in the 70's, and in the light of this production capability becomes modest, is that of building very large, inhabitable structures in space: cylinders kilometers across, with sun-light brought in by mirrors through large windows, air, and the feel of gravity underfoot, resulting in a pleasant environment inside.

There is a book coming out in a few months, part of the Time-Life series on computers, that is going to have a picture essay on nanotechnology. It pictures a space settlement being constructed by assemblers from asteroidal materials. The size of the settlement is a thousand kilometers in diameter. That is the sort of thing one can do with superior structural materials.

The marginal cost of building such structures using nanotechnology, with respect to human labor and terrestrial resources, will be essentially zero. The greatest issue will be R&D cost. Further, if you have a general way of applying assemblers to making things, then you can specify the size and shape of those things, and thus even that cost may not be so very high. That makes the "world" look like a very different place, if suddenly the "world" is larger than the earth, because most of the "world" isn't the earth, by many orders of magnitude.

L. Potential for Abuse

There is another side to these technologies, also discussed in Engines of Creation. In the illustration, you can see trees silhouetted against the early stages of the expansion of a nuclear fireball. Nanotechnology has nothing to do with nuclear technology. There is no transmuting of nuclei as the alchemists tried to do, and as is done by nuclear technologists. Nanotechnology only does what chemists do: rearrange molecules. Nonetheless, it is a technology where the principle of exponentiation can be brought to bear: nuclear explosions come from an exponential proliferation of neutrons in a critical mass of fissile material. Here, we are talking not about an exponential growth of destroying things and releasing energy, but instead a potential exponential growth of constructing complex artifacts. In its way, that is a far more powerful capability. Powerful not only for medicine, not only in a higher standard of living for everyone on the planet than we have in this country today, but it can also be used to produce very high performance weapon systems in vast quantities and virtually overnight -- once you have worked out the prototype and know how to build one. It could also be used to make computers that are smaller than bacteria, and thus make programmable germs for germ warfare. That is an ugly possibility.

The reasons for looking forward in time to this technology are not just for the "Gee whiz, it will be wonderful" aspects but, also, that there are uses we would like to prevent and/or control. That is the core of the challenge to public policy.

Returning to the example of the sun, again, a nuclear fireball. In the foreground of this picture, we have self-replicating, solar-powered, molecular machinery systems -- plants. It is clear these things can come together to make a pleasant world. What is at stake is this: a very large collection of atoms in space known as Earth. It is the difference between having a world that is polluted and having a world that is clean; the difference between dying of cancer and having a healthy body; the difference between having a biosphere and not having a biosphere. Nanotechnology will enable us to achieve the cleanup of toxic wastes by taking molecules apart and doing things with them. It will also enable us, if the wrong kind of replicator or weapon system is built, to destroy the biosphere. It is something that can be used to extend human life or destroy it.

M. The Emergence of Nanotechnology

Why should one expect something as outrageous as nanotechnology to emerge in the real world and expect people to have to deal with it, and possibly not in a time frame that is generations away, as I think some people would very much like to believe?

It is clear that the basic principles of nanotechnology work because they are demonstrated by biology and we are alive. We know that if there are paths to "there"; that there is a "there" there -- the possibility of nanotechnology.

There are many paths to nanotechnology. There is no one problem that can block progress in this direction because there are so many ways of making that progress. I have outlined several; there are hybrids among them; there are many variations on those themes.

The goal of nanotechnology does not require the perception of a vast payoff in the future to entice people to pour research and development money into some new direction to make this happen. Payoffs along the way, things like better pharmaceuticals, scientific understanding, enzymes for industrial processes, and so forth are already leading people to learn how to build complex molecular structures and build proteins and so on. Today, researchers in those fields are increasingly seeing that what they are doing is leading towards nanotechnology. The next time they pick a research direction, it is likely to biased towards research that leads in this direction. Even without that, one would still get there, though perhaps with less warning and understanding.

Towards the end of the development paths, there are potentials for tremendous medical, commercial, and military applications. When you think about the decision makers in the technological nations of this world, it is very hard to conceive of even a single decision maker, let alone a majority, that is not motivated by one or more of the goals of greater wealth, longer, healthier lives, and either defensive or offensive capability. In a world that holds many competing companies and governments, it is very hard to imagine anything short of a global catastrophe that would stop people from continuing along one or more of these many paths with the short term payoffs, to finally lead to the kinds of capabilities that have been described.

Today, we are trying to learn how to design improvements of molecules that we already know how to build. Several of these paths begin with designing polymer molecules that fold up as proteins do. This process is underway. I believe that is an area that will see increasing commercial activity.
Today, there is a need for software tools of greater capability and/or lower cost for doing modeling of complex molecular structures in a way that is useful for computer aided design in a molecular world. I think that the combination of improved design software, along with improved methodologies for designing, improving, and characterizing these molecules will lead to an increasing range of short term applications. Enzymes, pharmaceuticals, and molecules that do interesting things from the point of view of getting information about what is happening in the molecular world, will be some of the products.

Activity will increase over the years and will blend bit by bit into programs to build complex molecular machines that can built better molecular machines. At some point the result will be nanotechnology. And I hope we are ready.

N. Questions and Answers

AUDIENCE: You spoke of the benefits of nanotechnology, the development of materials, and examples of nanomachines that occur in nature: it would appear that energy conversion is one of the most critical technological challenges posed by nanotechnology, along with the software problem. Do you plan to address those problems in detail in your next book?

E. DREXLER:. Certainly in more detail than I did in Engines of Creation. In my own thinking on these matters, in an exploratory engineering vein, where the goal is to come up with relatively simple, understandable things that are still general enough that they support arguments for a wide variety of capabilities, I found myself thinking in terms of DC power, as the basic "energy currency" for running these systems. You can get DC power, literally, by plugging into a wall circuit, and from there you can run a motor. The electrostatic motor that I showed was designed to run on five volts. You can also get DC power by converting chemical energy into electrical energy, as is done in fuel cells -- which on a very small scale would have a very high power density, since this conversion is a surface effect and nanostructures would have a high surface to volume ratio. Finally, one can imitate plants. Plants convert sunlight into chemical energy, they could do it equally well into electrical energy with some modest modifications of the kind of processes that go on in these molecular structures. They do that with an efficiency that is now around a couple of percent, typically. We already know how to do that with 30% efficiency, even without nanotechnology, in artificial structures. We should be able to do even better with molecular devices. Nanotechnology will also make energy cheap.

AUDIENCE: Your book stated that you expected the breakthrough to take place within 10 to 50 years, approximately. Near the end of your lecture you stated that short of a global catastrophe, you expected this breakthrough to happen. It looks like there has to be an enormous amount of work done before we cross this threshold, is it conceivable to you that the knowledge and motivation will continue if a number of world governments became so powerful and oppressive that they were able to halt this research.

E. DREXLER: I said that one of the things that could block nanotechnology was a global catastrophe. That was one of the catastrophes I had in mind. I think, however, that if you look at the time frame we are talking about, which is measured in small numbers of decades, a number which I can argue for as being reasonable, if one's sense of this is at all correct, and if you look over the history of the past few decades, what has happened in this century is that people have figured out how to do organized research and development, and it has spread. Now we have more and more countries that have R&D labs. Korea has a goal of becoming a power in biotechnology in the 21st century and they are actively working on it. Their educational system is superior to ours. If you ask what governments are going to be the dominant world powers a few decades out, I think it will be the governments, and supporting culture and ideology, that are effective in developing technology. Any group, or country, that says, in effect, "we are going to pull back", unless everybody else by some miracle did it simultaneously, simply pulls themselves out of the race. Soon, they would not be effective anymore, and one would shift one's attention to those that are ahead in the race.

AUDIENCE: I was not thinking that governments would intentionally pull back from this. Suppose we envision the governments controlling and consuming more and more resources of their respective economies, creeping up on the goose that is laying the golden egg. It might not require a huge catastrophic event. It might gradually happen.

E. DREXLER: This gets into speculations about future social systems and clearly a wide range of things are possible, but currently trends actually seem to be away from that. I personally hope those trends continue.

AUDIENCE: In response to the previous question, if we look at history in this century, research and development have been enormously accelerated by the competition between nations. If you have nations rising to achieve military power, the urge to create weapons to stop them and technologies to beat them becomes irresistible. Every nation is running scared of every other that it thinks is going to beat them. My analysis is that research and development are accelerating, not declining.

E. DREXLER: That matches my evaluation.

AUDIENCE: You said you expect the progression to be from protein machines to non-protein machines. Do you think it possible to shorten or skip that intermediary step? Could you design machines directly from say m-RNA to fold together to make something useful? Instead of having the t-RNA attach to an amino acid, modify it to attach to a different set of molecules to be directly used as a part in building a machine?

E. DREXLER: What you are talking about is essentially re-engineered ribosomes and t-RNA so that you could use this programmable machine tool that we already have to make a different kind of polymer. While that is a fundamentally sound notion, in practice it would be enormously difficult, because the existing systems are adapted so well to doing just what they do. It is plausible, though, that one might build systems that do the sort of thing that you are talking about, but probably designing them from scratch, by "looking over the shoulder" at the way ribosomes do it. That kind of intermediate technology, where you have molecular machines that are producing not generalized structures, but superior polymers for making molecular machines, I think is a very important intermediate step. One way of getting the information to those machines, instead of programming them by "tape", which requires some fairly complex mechanisms to read, might be to have a molecule that might be stepped through a series of operations by changing the composition of the chemical bath. Biology can't do that; chemistry does do that a lot. Think of it as a halfway house between solid phase synthesis, which is how proteins and nucleic acids are made, in a more or less automatic way, and assemblers. It would be a simple molecular machine for making the chain.

AUDIENCE: I am interested in what might be called "analyzers", machines able to look at a material and tell what the composition is and make a tape for the assembler to replicate. One could in principle create the "tape" and read it at another location to regenerate the organism or material or person.

E. DREXLER: That has also been suggested as a means of transportation. Those things are not discussed in Engines of Creation, partly because discussion of them in my experience generates more heat than light. However, the general class of capabilities that you are pointing to is going to be an important one for people to be concerned with.

AUDIENCE: You discussed in your book the "germ" theory of information, memes. "Nanotechnology" is essentially a meme. One of the things I have noticed is that when I mention your book and the concepts in it to rather intelligent people, the first approach is one of fear, very definitely: won't this make human beings obsolete?

E. DREXLER: If human being are in some fashion in charge and don't consider themselves to be obsolete, then the answer is 'no'. If that condition is not met, then the answer is 'yes', and things are either very awful or very strange, depending on whether its involuntary or voluntary.
In considering the implications of nanotechnology, I would like to distinguish two phases in the development of nanotechnology. Phase I involves the ability to make very small computers and assemblers, things that are no more complex then we already know how to make on a macroscopic level but are simply implemented on a molecular scale. Phase II involves design and software capabilities far beyond what we can do today.

Nanocomputers, assemblers, and even replicators are in the first phase. I believe a replicator is about as complex as a modern automated factory even though it has the advantage of working in an environment rich in pre-fabricated parts. Relatively simple cellular repair machines are also part of the first phase.

Things like very ambitious cell repair, AI (which is what I think of when you speak of making people obsolete), and very ambitious re-working of the human body are part of the second phase. The things you are pointing to are part of Phase II Nanotechnology -- nanotechnology combined with very powerful design capabilities, probably in a world that has real artificial intelligence.

AUDIENCE: How quickly would this happen? Phase I could very rapidly move to Phase II.

E. DREXLER: As I discuss in Engines of Creation, if you can build genuine AI, there are reasons to believe that you can build things like neurons that are a million times faster. That leads to the conclusion that you can make systems that think a million times faster than a person. With AI, these systems could do engineering design. Combining this with the capability of a system to build something that is better than it, you have the possibility for a very abrupt transition. This situation may be more difficult to deal with even than nanotechnology, but it is much more difficult to think about it constructively at this point. Thus, it hasn't been the focus of things that I discuss, although I periodically point to it and say: 'That's important too.'

AUDIENCE: One of my big concerns is not that human beings will become obsolete, but that a lot of human institutions will become obsolete. I have tried to conceive of major social institutions that could deal with full-blown nanotechnology. I don't see any.

E. DREXLER: The Foresight Institute is intended to encourage people to think about these matters. I expect that most of the high quality debate will eventually be in media with fast publication of little bits of ideas that can be tied together and criticized, i.e., hypertext publishing.

I think that there are some basic principles of checks and balances that work fairly well in some of the democracies. I think that something in the direction of these principles can be applied to the very important problems that you have been thinking about, and I encourage people to keep on staring at these problems to see what can be done. The way that huge problems can be made manageable is to try to whittle them down by finding partial solutions here and there.

AUDIENCE: A lot of what you have been discussing is based on very small computers to be made possible by nanotechnology. How are you going to transfer this information to and from these very small computers?

E. DREXLER: You can take the "I/O problem" and separate it into the "I problem" and the "O problem."

If you can build a nanocomputer, you can certainly build a wire that is thin at one end to bond to the nanocomputer and fat at the other to bond to a microchip. If a wire ends in a plate that is 10 nm across and separated from another plate by a few nm, and the other plate is kept at ground potential, and we move the voltage on the first plate from ground up to a few volts and back down again, it will carry an electrostatic field between the plates of this little capacitor. The numbers suggest that you should be able to generate the force necessary to yank a rod in the nanocomputer in a way mechanically compatible with the operation of the system. Thus you can go from a standard 5 volt electronic signal to a logic state on a nanocomputer.

For output, you can have two parallel plates and look at the current flowing between them, which varies as a function of distance. We know from the scanning tunneling microscope that conventional electronics can detect changes in the conductivity of a circuit that result from the interaction between a surface and the single atom at the end of the STM needle. Two plates 10 atoms on a side give you a factor of 100 more detectable than what the STM can detect now. The plate separation can be changed by moving a rod, and now we have an output channel.

AUDIENCE: What about organic luminescence?

E. DREXLER: You can also use light for both input and output. A limitation is that the focal spot of a light beam is large compared to a single device. You could still get multiple channels by going to multiple frequencies. I haven't looked at this approach quantitatively.

AUDIENCE: What about the accidentally destructive aspects of nanotechnology as well as the beneficent and the malicious applications of nanotechnology? As a software writer, I've noticed that I rarely write a program that doesn't have bugs the first time.

E. DREXLER: The problem you are raising is of accidents in design that lead to devices that run amuck in a destructive way. I feel that I didn't address this problem as well as I might have in Engines of Creation. Since writing that, I've come to the conclusion that if people are really concerned about such a thing happening, they will try to avoid a situation in which a small accident can produce a run-away self-replicating machine that gobbles up the world.

Here's an example of how a very little bit of care could eliminate that problem. Never build a replicator that's anything like a replicator that could survive in nature. In biotechnology, people are tinkering with cells that have evolved to live in nature. That has the flavor of a dangerous thing to do. The danger, nevertheless, was very over-rated in the early days due to a lack of understanding of bacterial ecology. If instead you are working with devices that are no more like something that could live freely in nature than is a piece of machinery, the danger of run-away growth is non-existent. Here is a metaphor: Imagine that you design a replicator that works in a vat of industrial chemicals, that requires for its oxygen source hydrogen peroxide, and for its carbon source, some petroleum derivative. Such a thing would have an obligatory requirement for those things in the same way that an automobile has a requirement for gasoline and transmission fluid. To have something like that accidentally be able to live in nature would be like having your mechanic slip up when working on your car with the result that the car could go into the woods and suck sap from trees.

This realization has made me feel much better about accidents, but very scared about abuse.

AUDIENCE: One thing that might help is if for every team working to build something, you had another team working to figure out every way in which it could get loose.

E. DREXLER: That might be useful in some cases; in others it shouldn't be necessary.

AUDIENCE: There is a problem with your last argument. The nature of the human spirit is to create organisms that can go and live by themselves. They're normally called 'children' but also are called 'computer viruses'.

E. DREXLER: Yes. Again, I believe that people deliberately doing these sorts of things is what we have to watch out for.

AUDIENCE: I feel that the competitive spirit tends to drive a lot of things. I was wondering what your estimate is of the competition among companies and among countries for nanotechnology in the near term.

E. DREXLER: At present, the state of competition with respect to nanotechnology per se is essentially nonexistent. There are a few companies that have expressed an interest in putting nanotechnology on their research agendas. I have heard essentially nothing from the government. Nanotechnology will come out of other areas that are the focus of intense competition because of short term pay-offs. I expect that in coming years, as we see the transition from people reacting to nanotechnology as a wild, unworkable idea, to people saying (and this is already starting to happen) that it's obvious, not worth talking about, that companies and countries will recognize nanotechnology as one of a very few key research priorities. You may well see something like the Manhattan project.
If history is any guide, it is likely that such programs will be competitive programs. I would like to see them be cooperative programs across as wide a range as possible of decent governments, which hopefully will embrace all governments.

Next page of the NanoCon Proceedings
Return to NanoCon Proceedings Table of Contents
Return to Jim's Molecular Nanotechology Home Page
Return to Web Site Design and Authoring Services

Comments and correspondence? Send email to Jim Lewis:
NanoCon Proceedings © Copyright 1989, by NANOCON
This page is part of Jim's Molecular Nanotechnology Web, copyright ©1996 James B. Lewis Enterprises. All rights reserved.
Last updated 24June96.
The URL of this document is: