 Overview of
the Debate |
| Round 1 | Round 2 | Round
3 | Round 4 | Epilogue |
| Round 1 -- Part 1 |
Letters | Part 2 |
| Overview of
the Debate |
| Round 1 | Round 2 | Round
3 | Round 4 | Epilogue |
| Round 1 -- Part 1 |
Letters | Part 2 |
by Ralph C. Merkle, Xerox
PARC, Palo Alto, CA.
Scientific American staff writer
Gary Stix wrote Trends in nanotechnology: waiting for
breakthroughs (Scientific
American, April 1996, pages 94-99). The text of this
news story appears below, with detailed commentary and
corrections.
Scientific American has
evidently concluded that developments in molecular nanotechnology
have become too big to ignore. In consequence, they published a
six page story on "nanotechnology."
The story conflates two very different
definitions of "nanotechnology." The consequences of
this lead to various erroneous claims and a great deal of
confusion.
Several lines of evidence and a great deal of
published work support the feasibility of molecular
nanotechnology. Three mentioned in the story are the statements
of Feynman, Nanosystems,
and a
significant and growing body of computational studies. The Scientific
American story attempted to neutralize all three using a
variety of attacks that carefully avoided the technical issues.
The absence of any effective technical criticism in the story is
quite telling.
Scientific American, by
denigrating molecular modeling and computational work, is
implicitly advocating a purely experimental approach to the
development of molecular nanotechnology. In sharp contrast, we
advocate a mixed approach which
includes the aggressive use of molecular modeling as an integral
component of a long range strategy to develop molecular
nanotechnology.
The story devoted over a
column to the only published argument against molecular
nanotechnology that they found worth citing, a book
review (not a peer-reviewed journal article) by David Jones
(a chemist and Nature columnist). Jones' criticisms
were refuted
last year shortly after they appeared, a refutation that Scientific
American conveniently omitted. The story then devoted two sentences to the misleading claim that the
only response was to read Nanosystems.
As Jones' criticisms were based on gross ignorance of the field
-- an ignorance that was made clear in our response -- it is
difficult to understand Scientific American's
uncritical embrace of his views.
There are repeated ad hominem attacks against
Eric Drexler (a major researcher
in the field). Many researchers think that nanotechnology is
feasible and should be developed; ad hominem attacks against one
researcher say more about the agenda of Scientific American
than about the technical issues.
Many of the readers of the Scientific
American story will never be made aware of the errors and
omissions it contains. If it had been published on the web, and
if backlinks and filtering
were implemented (various
groups are working on this), then anyone who read the
original story would also have rapid and easy access to the
critiques of that story. The reader is left to imagine the
consequences if debates and discussions of public and other
issues took place in a medium where inaccurate and misleading
information was rapidly squelched.
In the meantime (before we have backlinks) if after reading
this page you think the Scientific American story was unduly
biased we encourage you to pass along the URL of this page,
http://www.foresight.org/SciAmDebate/SciAmResponse.html , to
friends, relevant newsgroups, etc. and encourage discussions
about it. Some relevant questions that
could be discussed are at the end of this page.
To stay informed about developments in
nanotechnology join the Foresight
Institute (email: foresight@foresight.org)
and get the Foresight
Update.
To encourage an open dialogue, we publicly offer to Scientific
American the opportunity to rebut our statements. We will
include a link to a page provided by Scientific American.
If that should be inconvenient for them we will offer to host
their rebuttal at the Foresight web site.
The link to Scientific American's
rebuttal is: http://www.foresight.org/SciAmDebate/Round3.html.
Scientific American earlier requested that the Foresight Institute
remove the web page you are now reading from the Foresight
server.
For an overview of the debate, see http://www.foresight.org/SciAmDebate/SciAmOverview.html.
This section gives the highlights of the more detailed critique which follows.
The first and perhaps most serious error in the article is to adopt a definition of "nanotechnology" that is grossly inappropriate. The article says:
"Nanotechnology is the manufacture of materials and structures with dimensions that measure up to 100 nanometers (billionths of a meter)."
While this definition would be quite reasonable in an article
on high resolution lithography, its use in an article which is
centered on molecular manufacturing as discussed both in Nanosystems
and at the Fourth
Foresight Conference on Molecular Nanotechnology is not only
inappropriate but extremely misleading.
An appropriate definition for this article
would be: a manufacturing technology able to inexpensively
fabricate, with molecular precision, most structures consistent
with physical law.
A consequence of the appropriate definition
is that nanotechnology is today theoretical. We do not yet have
the defined ability, and it is unlikely that we will develop it
in the near term (i.e., the next 5 to 10 years).
As the distinction between nanoscale
technologies (i.e., technologies in which some critical dimension
is less than a micron) and nanotechnology was made repeatedly at
the Fourth
Foresight Conference on Molecular Nanotechnology, and as the
author of the story attended this conference, it is hard to avoid
the conclusion that either (a) the writer failed to understand a
very basic point or (b) he deliberately selected an inappropriate
definition.
The selection of an inappropriate definition
leads, as might be expected, to a series of inappropriate and
erroneous conclusions. Among these was the confusion of today's
lithographic technology with nanotechnology. As today's
lithographic technology is both unable to achieve the objectives
of nanotechnology and is generally viewed as unlikely ever to
achieve these objectives, its inclusion creates a great deal of
confusion. Arguing that molecular manufacturing is likely to be
uneconomical or infeasible because evolutionary developments in
today's lithographic technology will be unable to deliver the
capabilities that molecular manufacturing promises is to create a
most glorious muddle. (As an aside, there is also a turf-war over
the term "nanotechnology." That is,
"nanotechnology" is a term which has an aura of
excitement and great promise. Much of this aura was created by Drexler's adoption of the term and
its association with molecular manufacturing. As a consequence,
many researchers wish to adopt a definition of
"nanotechnology" which includes their own work. An
unfortunate consequence of this is that the unqualified term
"nanotechnology" has come to mean very little. In the
present critique, "nanotechnology" is defined as a
synonym for "molecular manufacturing." In contexts
where it might be unclear which usage is appropriate, the term
"molecular nanotechnology" or "molecular
manufacturing" can be used. In any event, it is essential to
establish an appropriate definition if a discussion of
"nanotechnology" is to make any sense. The usage in the
story is variable.)
One of the arguments in favor of nanotechnology is that
Richard Feynman, in a
remarkable talk given in 1959, said that "The principles
of physics, as far as I can see, do not speak against the
possibility of maneuvering things atom by atom. It is not an
attempt to violate any laws; it is something, in principle, that
can be done; but in practice, it has not been done because we are
too big."
Even more strongly, he said "The
problems of chemistry and biology can be greatly helped if our
ability to see what we are doing, and to do things on an atomic
level, is ultimately developed---a development which I think
cannot be avoided."
Feynman's
classic talk is now available on the web and is well worth
reading.
Scientific American faces a
significant challenge in dealing with Feynman's talk, as taken at
face value it supports the feasibility of nanotechnology. The
response was to misinterpret a different paper by Feynman as
being applicable to research in nanotechnology.
In "Cargo Cult Science" Feynman
said: "But even today I meet lots of people who sooner or
later get me into a conversation about UFO's, or astrology, or
some form of mysticism, expanded consciousness, new types of
awareness, ESP, and so forth. And I've concluded that it's not a
scientific world."
Pretending that Feynman's essay opposing ESP
and mysticism applies to theoretical and computational work in
nanotechnology completely avoids the awkward need to comment
directly on what Feynman said in his talk, while at the same time
creating a spurious notion that Feynman did not support the
development of nanotechnology. Nothing could be further from the
truth.
Evidently, Carl Feynman (the son of Richard
Feynman) agrees. In a
post to sci.nanotech he said: "I was dismayed to read in
your April 1996 issue ("Waiting for Breakthroughs") an
extended quotation from Richard Feynman's essay "Cargo Cult
Science" used as a critique of nanotechnology. I am sure he
would have found such misuse of his idea quite unreasonable. I
should know, because I talked with him at length about the
prospects of nanotechnology."
Scientific American faces a difficult problem
with Nanosystems.
Nanosystems: Molecular Machinery, Manufacturing and
Computation, is an extension and enlargement of Drexler's MIT Ph.D. thesis.
Published by Wiley in 1992, it provides a dense and technical
analysis of the issues involved in nanotechnology. Drexler
negotiated with Wiley to get a purchase price low enough that
students could afford it. Over 10,000 copies are now in print. No
technical errors of significance have been found, despite several
years of exposure to the entire scientific and technical
community, including both critics and supporters.
As the reader might appreciate, if there were
any serious technical objections to Nanosystems, Scientific
American would have been more than happy to point them
out. The absence of any technical criticisms of Nanosystems
is glaring.
Computational chemists generally seem able to absorb the idea
of nanotechnology quite easily. One reason for this is that they
have the training and the tools to evaluate proposals for
molecular machines. Give them a PDB file and tell them that it
describes a molecular structure that's supposed to do something
useful, and they'll happily model it. Sometimes they come back
and tell you that it works and it's a great design. Sometimes
they torture it in horrible ways and it behaves in some strange
and unexpected manner.... at which point, it's back to the
drawing board.
To select a specific example, there has been
significant modeling work done on mechanosynthetic reactions that
should be useful in the controlled synthesis of diamondoid
structures. In particular, the ability to selectively remove a
hydrogen from a hydrogenated diamond surface appears to be a
fundamental reaction in the controlled synthesis of diamond and
related structures.
The analysis of the hydrogen
abstraction tool has involved many people, including: Donald
W. Brenner, Richard J. Colton, K. Eric Drexler, William A
Goddard, III, J. A. Harrison, Jason K. Perry, Ralph C. Merkle,
Charles B. Musgrave, O. A. Shenderova, Susan B. Sinnott, and
Carter T. White. The institutions involved include the Materials
and Process Simulation Center at Caltech; the Department of
Materials Science and Engineering, North Carolina State
University; the Institute for Molecular Manufacturing; the
Department of Chemicals and Materials Engineering, University of
Kentucky; the Chemistry Department, United States Naval Academy;
the Naval Research Laboratories, Surface Chemistry Branch; and
the Xerox Palo Alto Research Center.
The response to this work by the story is
straightforward: it is labeled "cargo cult science." No
technical argument is advanced to explain why the results of
computational modeling and computational chemistry should be
ignored.
The story presents the book
review by David Jones as the only published argument against
the feasibility of nanotechnology. Jones cited no references, and
even a casual reading of his criticisms clearly shows that he is
almost completely ignorant of the field which he claims to be
criticizing.
The story quotes Jones as saying:
"Single atoms of more structurally useful elements [more
useful than Helium at 4 kelvins] at or near room temperature, are
amazingly mobile and reactive. They will combine instantly with
ambient air, water, each other, the fluid supporting the
assemblers, or the assemblers themselves." Earlier in the
paragraph Jones said "...the nanotechnologists do not seem
to realize the chemical obstacles in their path."
It is evident from reading Jones' criticism
that he is literally unaware of the computational and molecular
modeling work that has been done. In particular, the
molecular dynamics simulation of the hydrogen abstraction tool
provided a vivid and very clear illustration of why such reactive
species as radicals, carbenes and the like would not react with
"ambient air, water, each other, the fluid supporting the
assemblers, or the assemblers themselves." The reactions
take place in vacuum -- there is no ambient fluid with which to
react -- and the reactive species are positionally controlled --
they cannot react with the assembler because they are positioned
away from the structural components of the assembler. In a
similar way, a hot soldering iron does not react with the skin of
the operator provided it is positioned so that it does not come
into contact with the skin.
While Jones' excuse is ignorance, the writer
of the news story has no such excuse. He attended the conference
where the video of the hydrogen abstraction tool was presented;
and sat through more than one lecture discussing mechanosynthesis
-- the reactions that would be involved, the molecular tools that
would be used, and the environment in which those tools would
operate. Again, either the writer (a) is grossly ignorant of the
basics or (b) deliberately chose to omit inconvenient facts.
The following is a point by point response to the story. We
have added some relevant links to Scientific American's
text -- these are our additions. The original article, not being
on the web, had no links.
As this detailed response is rather long the
reader might wish to scan through it for points of particular
interest (or perhaps skip directly to the
conclusion). If there are any criticisms in the story that
the reader was particularly interested in, it would be worthwhile
to find the particular statement and read the critique of that
statement presented here. It is difficult to convey with a few
short quotes the rather pervasive bias which emerges when the
article is viewed as a whole.
"'That's the messiah,' confides Edward M. Reifman, D.D.S. The Encino, Calif., dentist has paid hundreds of dollars to attend a conference to hear about robotic machines with working parts as small as protein molecules. Reifman nods toward K. Eric Drexler, the avatar of nanotechnology."
In response, Reifman said "...I was quite upset to be
quoted entirely out of context regarding Eric Drexler being 'the
messiah'. That comment was mentioned in jest when I talked
briefly with the writer..."
The introduction is obviously designed to
distract attention from the basic technical issues. It
establishes that the story is not about nanotechnology so much as
it is about Drexler.
Don't get me wrong. I like Drexler, and think
he's done some remarkable work. But that's not the point. Either
nanotechnology is feasible within the presently understood
framework of physical law, or it is not. This question exists
entirely independent of Drexler. If nanotechnology is feasible,
we've got a lot more to be concerned about than the conference
registration fee or whether Reifman called Drexler the messiah.
If it's not, we still don't care about these things. Scientific
American is attempting to distract us from the core
issues.
"Drexler has just finished explaining to a strange mix of scientists, entrepreneurs and his own acolytes that nanotech may arrive in one to three decades. The world, in his view, has not fully grasped the implications of molecular machines that will radically transform the way material goods are produced."
The use of biased words such as "strange" or
"acolytes" is pervasive. It's an easy way to put a
negative spin on a story in the absence of anything technically
substantive. Some biased usages are noted in the following text,
but their sheer number makes it awkward to comment on all of
them.
This story is itself evidence that at least
some parts of the world have "... not fully grasped the
implications..." of nanotechnology. This is certainly true
for Scientific American.
"Nanotechnology is the manufacture of materials and structures with dimensions that measure up to 100 nanometers (billionths of a meter)."
This is a more subtle bit of misdirection. There are now a
myriad different definitions of the word
"nanotechnology." While this definition would be quite
unobjectionable in a story on (say) ultra high resolution
lithographic methods, in the present story -- which is focused
almost exclusively on molecular manufacturing -- its use is
inappropriate and leads to confusion. For purposes both of the
story and of this review, an
appropriate definition of "nanotechnology" is: a
manufacturing technology able to inexpensively fabricate most
structures consistent with natural law, and to do so with
molecular precision. This is radically different from
"structures with dimensions that measure up to 100
nanometers."
As the writer of the story attended the Fourth Foresight
Conference on Molecular Nanotechnology, where the basic
distinction between structures on the scale of nanometers and
structures that are molecular in their precision was repeatedly
emphasized, it is difficult to understand how he decided to
include such a misleading definition. Either he was unaware of
the appropriate definition or he chose to use a definition at
odds with the established usage in the field upon which he was
reporting.
By way of analogy, suppose a story on
"chemistry" defined a "molecule" as "an
aggregate of atoms with dimensions less than 100
nanometers." The confusion wrought by this initial blunder
would be remarkably difficult to repair.
It is important to remember that there are
three major objectives to nanotechnology: (1) Low cost. The
ability to very expensively make a few very small very precise
structures will not, per se, have a major impact. (2)
Molecular precision. While certain limits are imposed by such
factors as thermal noise and radiation damage, error rates that
are remarkably low by current standards should be feasible.
Appropriately designed structures with tens of billions of atoms
should be stable (in the sense that no single atom in the
structure would be out of place, except for the limited
excursions that thermal noise creates) for many decades. (3) Able
to make most structures consistent with physical law. While there
will no doubt be some structures that simply can't be
synthesized, current manufacturing methods are able to make only
a remarkably small subset of what's feasible.
Today's computational modeling in
nanotechnology typically focuses on the class of stiff diamondoid
structures -- structures made primarily of hydrogen and first row
and second row elements in the upper right hand region of the
periodic table. While this is a significant reduction in scope
from structures involving the entire periodic table, it makes the
problems involved in the design and analysis both of the
structures and of the synthetic pathways for making the
structures more tractable; in addition, the diamondoid structures
appear to include many structures of potentially major economic
interest. This includes (among others): molecular semiconductor
electronic devices, very high strength very light materials, many
molecular machines and, of course, many types of assemblers.
This review uses the term
"nanotechnology" as a synonym for molecular
manufacturing. The usage by Scientific American is
variable. In other contexts, where the appropriate definition
might not be obvious, the more specific terms "molecular
nanotechnology" or "molecular manufacturing" can
be used.
"Its definition applies to a range of disciplines, from conventional synthetic chemistry to techniques that manipulate individual atoms with tiny probe elements. In the vision promulgated by Drexler, current nano-scale fabrication methods could eventually evolve into techniques for making molecular robots or shrunken versions of 19th century mills."
Properly speaking, none of the items mentioned is
nanotechnology. The synthetic methods used in chemistry today can
economically make many structures, but certainly cannot make (for
example) molecular computers or low-cost shatterproof diamond
sculpted with molecular precision. Scanning probe methods are
even more limited in the range of structures they can currently
make. However, further developments in both chemistry and
proximal probe methods could be important or crucial in the
development of nanotechnology. Some of the research in these
areas can reasonably be described as "leading to
nanotechnology."
"Current nano-scale fabrication
methods" is a phrase that typically refers to today's high
resolution lithographic manufacturing methods. There is fairly
wide agreement that these methods are neither able to make
structures of molecular precision, nor are they likely to do so
in the future. Something new is needed. Drexler, among others,
has repeatedly emphasized this point.
We should remember: nanotechnology is today a
theoretical field. We want it but we haven't got it.
"In the course of a few hours, manufacturing systems based on Drexler's nanotechnology could produce anything from a rocket ship to minute disease-fighting submarines that roam the bloodstream. And, like biological cells, the robots that populate a nanofactory could even make copies of themselves. Finished goods in this new era could be had for little more than the cost of their design and of a raw material - such as air, beet sugar or an inexpensive hydrocarbon feedstock.
While manufacturing costs will be low, other factors will likely increase the purchase price of "finished goods." As an example: the manufacturing cost of software is low, but the purchase price varies over a wide range.
The Drexlerian future posits fundamental social changes: nanotechnology could alleviate world hunger, clean the environment, cure cancer, guarantee biblical life spans or concoct superweapons of untold horror."
Eric Drexler testified at the senate hearings on "New Technologies for a Sustainable World" held by Al Gore shortly before Gore became a candidate for Vice President. The transcript of the statements by both Drexler and Gore is available on the web, as is Drexler's prepared statement.
"Scientific visionaries have turned their attention from outer to inner space, as the allure has faded from dreams of colonizing another planet and traveling to other galaxies. Computer mavens and molecular biologists have replaced rocket scientists as the heroes that will help transcend the limits imposed by economics and mortality. 'Whether or not Drexler's utopian ideas are correct, they come at a time when a variety of fields have reached stasis,' says Seth Lloyd, a professor and specialist in quantum computation at the Massachusetts Institute of Technology. 'You don't come across many fields that have as bold a project as the space program was.'"
The National Space Society
might object to the claim that the allure of space development
and settlement has faded. Note also that we are going to
"transcend the limits imposed by economics and
mortality." This is a hyperbolic exaggeration by the writer.
The ability to manufacture products at a lower cost does not
"transcend the limits imposed by economics"; nor do
improvements in medical technology -- even
the rather remarkable improvements that nanotechnology is likely
to make feasible -- render us immortal. Today we live in a
world of greater material abundance than in centuries past, and
we live longer and healthier lives. In the future this trend is
likely to continue and nanotechnology is likely to play a major
role.
A RAND study, The
Potential of Nanotechnology for Molecular Manufacturing,
concludes that further research is warranted.
"Submicroscopic machines that can save or destroy the world appeal to anyone from a retired navy admiral to a technophile dentist to eager students - all of whom attended the nanotechnology conference. Reifman, the dentist, is a disciple who carries the message of nanotechnology to patients waiting nervously in his dental chair. He tells them of robots as small as a microbe that will painlessly refurbish a tooth or build a new one from scratch. 'You'll be able to be a chocoholic without guilt,' he predicts."
Referring to Admiral
David Jeremiah, USN (Ret.), former Vice Chairman of the Joint
Chiefs of Staff for Generals Powell and Shalikashvili as
"...a retired navy admiral..." is somewhat like
referring to Colin Powell as "... a retired general."
Jeremiah was more than an attendee: his
presentation on Nanotechnology and Global Security was
extremely impressive.
Scientific American conveniently
omitted researchers from the above list. The interested reader is
invited to peruse the conference
web page and examine the list
of attendees. The affiliations of the attendees are shown, so
it's easy to get a feel for which institutions are involved.
For the record, the sponsors or co-sponsors
of the Foresight series of conferences on molecular
nanotechnology have included (in no particular order): Apple
Computer; Beckman Instruments; Biosym Molecular Simulation; The
Materials & Process Simulation Center, Caltech; The
Laboratory for Molecular Robotics, USC; The Institute for
Molecular Manufacturing; JEOL; Loral Systems Manufacturing Co.;
Molecular Manufacturing Enterprises, Incorporated; Digital
Instruments; Fenwick & West; Niehaus Ryan Haller; Stanford
Univeristy Dept. of Materials Science & Engineering;
Molecular Graphics Society (USA); Global Business Network; Weil,
Gotshal & Manges; Autodesk; Mitsubishi Research Institute;
Nomura Research Institute; Xerox Palo Alto Research Center; The
Univeristy of Tokyo, RCast (Research Center for Advanced Science
& Technology); and the Micromachine Society of Japan.
"Drexler has purveyed his nanovisions for almost two decades. In recent years, however, his intricately constructed pictures of the next century and beyond have begun to be overtaken by real investigations into nanotechnology. What inspires actual researchers at the nanoscale is infinitely more mundane than molecular robots - but also more pragmatic. Nanotechnology, in this guise, may not contain the ready promise of virtually limitless global abundance and human mastery of the material world. But it may move beyond mere speculation to produce more powerful computers, to design new drugs or simply to take more precise measurements."
Drexler's first
published paper on nanotechnology, which appeared in the Proceedings
of the National Academy of Sciences USA in 1981, is now
available on the web.
Here we see the erroneous definition of
"nanotechnology" coming into action. Making more
precise measurements, making better drugs, making more powerful
computers by improving current lithographic methods: these are
all laudable accomplishments. But are they nanotechnology? Do
they give us the ability to inexpensively manufacture diamondoid
structures? No. Are they "real nanotechnology?" Not
really.
Terminological confusion has led to
misclassification of existing and near term research and
development. We do not have nanotechnology today, and are
unlikely to achieve it in the near future. It is not going to
fall out of the sky into our lap, it will not be developed by a
serendipitous experimental accident. It's going to take quite a
bit of hard work, and it's going to take planning on a time
horizon that's long by present standards.
The heartwarming story of the young scientist
who accidentally leaves the Bunsen burner on in his laboratory
when he goes to lunch, and comes back to find that he's
accidentally made a new, unique, and remarkably valuable material
is well known. The story of the same young scientist coming back
to find he has accidentally made a Saturn V booster is -- well,
let us say that it would be an unusual occurrence. Some things
require planning, organization, and the cooperative efforts of
many people working towards a common objective.
By this time we hope the reader is adept at
noting and ignoring the continuous use of biased words and
slanted adjectives: "purveyed," "real
investigations," "move beyond mere speculation,"
"actual researchers," etc. etc. etc. As might be
appreciated, a complete analysis of the biased terminology in the
story would be extremely tedious. Again, sheer volume forces us
to ignore most of the lesser examples of evident bias.
"Researchers can now manipulate atoms or molecules with microscopic probe elements, marshal the 20 basic amino acids to form new proteins not found in nature, or help organic molecules spontaneously assemble themselves into ordered patterns on a metal surface. This work certainly presents the prospect of providing new tools for the engineering community. Ironically, it also demonstrates the difficulties of using individual atoms or molecules as building blocks, given the presence of a host of physical forces that may displace them. In fact, some of Drexler's sharpest critics are engineers and scientists who spend their time toiling in the nanorealm."
The "host of physical forces" is too vague to comment on -- later in this review we'll see some more specific "problems." We'll also defer comments on the unspecified "critics" until some specific examples are presented.
"Drexler's fanciful scenarios, nonetheless, have come to represent nanotechnology for many aesthetes of science and technology. The phenomenon is not uncommon in the sociology of science. The public image of a certain field or concept, shaped by futurists, journalists, and science-fiction scribes, contrasts with the reality of the often plodding and erratic path that investigators follow in the trenches of day-to-day laboratory research and experimentation."
Yet another example of biased terminology. Drexler's scenarios
are "fanciful." The terms "infeasible" and
"contrary to physical law" have well defined meanings.
As the question of interest is the feasibility of nanotechnology,
the more hard edged statements that the proposals are feasible or
infeasible should be used. This would, of course, force a more
rigorous (and useful) discussion of the issues.
"Fanciful" is a term which means nothing, other than
the writer can't find any substantive grounds for criticism.
Next, we find we're "aesthetes."
Having slipped in the wrong definition of nanotechnology, Scientific
American is now explaining to us that we're not working
toward nanotechnology. Then they tell us that journalists
create the public image of a field and often get it wrong -- much
to the discomfiture of the practitioners -- I heartily concur!
Drexler, the 40-year-old guru of the nanoists, speaks with an exaggerated professorial tone that is faintly reminiscent of the pedantic 1960s cartoon character Mr. Peabody. Over a buffet lunch in early November at the biennial conference sponsored by his Foresight Institute - an organization he set up in Palo Alto, Calif., to help pave the way for nanotechnology - Drexler pours milk into his ice tea. He explains that the milk binds the tannins that may lead to throat cancer."
"Mr. Peabody"????
(A
picture of Mr. Peabody "and his boy, Sherman" is
available on the web).
The story has now abandoned all pretense of
discussing technology. We're back to Drexler. Drexler this,
Drexler that, Drexler the other. How he speaks, how he takes his
tea. Insinuations about his character. Scientific American
has adopted People magazine as a role model.
"During the meal, he complains about the shortsightedness of the scientific and technological research establishment in the U.S., which has largely ignored his brand of nanotechnology. Drexler is familiar with dreams that don't come true. In the 1970s he volunteered to work with space colonization advocate Gerard K. O'Neill to plan various scenarios for extraterrestrial living; he even wrote a paper on mining asteroids in his freshman year at M.I.T. Drexler and other nanoists view their technology as a means to rejuvenate a moribund space program that has no immediate plans to create retirement communities on Mars. Nanotechnology would allow the manufacture of strong, light materials that would go into space transport vehicles."
Scientific American has just published a story
based largely on ad hominem attacks on Drexler, using
systematically biased language, which constantly denigrates a
technological objective that is widely thought to be feasible.
This would appear to provide some evidence in support of
Drexler's statement. We'll have more to say on planning horizons
later.
The non-statement "Drexler is familiar
with dreams that don't come true" is another example
of the pervasive bias in the story.
Regarding the rejuvenation of a moribund
space program: The National Space
Society has a position
paper on molecular manufacturing, and the Numerical
Aerodynamic Simulation program at NASA Ames is pursuing an
initiative to develop a world class capability in computational
molecular nanotechnology.
It's also amusing to note the following quote
from Willy Ley: "The idea of a rocket-accelerated airplane
bomb must have been "in the air" in those days, for it
had occurred to me too. In early fall 1940 Schaefer and I set out
to test the idea on a few examples on paper. The result was an
article entitled "How about Penetration Bombs?" which
made the rounds of various technical and semitechnical magazines
without evoking the response we wanted. The editor of The
Scientific American wrote curtly that it was "too
far-fetched to be considered."" from Rockets,
Missiles, and Space Travel, Revised Edition, by Willy Ley,
New York Viking Press, 1957, page 172, footnote 3.
"The basic ideas behind small, self-replicating machines did not originate with Drexler. The renowned mathematician John Von Neumann, a father of the field of artificial life, ruminated about a machine that could make copies of itself. And in a much cited 1959 speech, Nobelist Richard P. Feynman talked about the ability to build things by placing each atom in a desired place. The self-assured Feynman used to toy playfully with the notion of making things small, musing on the theme with the humor of a Brooklyn-accented, Borscht Belt comic. Feynman even proposed a competition between high schools: "The Los Angeles high school could send a pin to the Venice high school on which it says [on the pinhead], 'How's this?' They get the pin back, and in the dot of the 'i' it says, 'Not so hot.'" Drexler, unlike the puckish Feynman, approaches his passion with a dour earnestness. The message: Nanotechnology is coming, we must prepare now."
It's quite entertaining to watch the mental gymnastics of
those who think nanotechnology is infeasible but who
simultaneously respect Feynman, who
said: "The principles of physics, as far as I can see,
do not speak against the possibility of maneuvering things atom
by atom. It is not an attempt to violate any laws; it is
something, in principle, that can be done; but in practice, it
has not been done because we are too big."
A similar observation holds for von Neumann.
Somehow it is necessary to believe that the self-replicating
systems proposed for nanotechnology are impossible, while at the
same time believing both that von Neumann's proposals are
entirely reasonable and that living systems are exempt
from the ban on self-replication that applies to molecular
manufacturing systems (perhaps living systems have "vital
force").
A brief
introduction to some of the concepts of self-replication relevant
to nanotechnology -- along with links to further information --
is available on the web.
If Feynman is right that we can arrange atoms
and von Neumann is right that we can design self-replicating
machines then the basic requirements for nanotechnology are
satisfied: we should be able to make self-replicating machines
that can arrange atoms. The rest of the discussion should not be
about whether this capability is feasible, but rather about the
best method of achieving it. Drexler's dourness or lack thereof
is irrelevant.
"Drexler, though, can rightly claim credit for bringing wide exposure to an enticing idea. In his 1986 work Engines of Creation, Drexler, like Jules Verne and H.G. Wells, succeeded in depicting a world altered forever by the advent of a new technology. In Engines, Drexler introduced the concept of an "assembler," a robotic device with dimensions of a tenth of a micron (a millionth of a meter) or less, that can pick up and position a reactive molecule so that it interacts with another molecule, as though it were a Lego block snapping into place. He has also described mills equipped with belts and rollers to process molecules. A battery of nanocomputers - perhaps collections of molecular rods that change position to represent distinct logic states - could broadcast instructions to trillions of assemblers at once. The computers could also instruct assemblers to self-replicate. In his book, Drexler set down a detailed description of how society would be transformed by nanotechnology. Engines presents a picture of a Manichaean balance of utopian/dystopian scenarios."
Damning with faint praise. Also note that "...Drexler set
down a detailed description...." The implication here is
that "a detailed" prediction of the future is
impossible, Drexler made a detailed prediction, therefore Drexler
is attempting to do the impossible. Both Engines and Unbounding the Future
present several plausible scenarios. Note that considering
multiple plausible scenarios about what might happen in
the future is very different from giving a single detailed
description about what will happen. Scenario planning
is an established method of guiding decision making in the face
of an uncertain future, and Drexler quite explicitly adopted this
methodology. Given that nanotechnology is feasible -- a technical
question that Scientific American seems intent on
avoiding -- then scenario planning is a useful tool for gaining
some insight into the range of issues that might arise and some
of the possible options.
Combining nanocomputers with molecular machines would allow almost anything that can be designed to be made from a variety of inexpensive raw materials, perhaps even dirt, sunlight and air. Assemblers could string together atoms and molecules so that most goods could be made from diamond or another hard material, giving the most ordinary objects a remarkable combination of strength and lightness."
An unbiased paragraph describing some things that nanotechnology should make feasible.
"The cost per kilogram of goods produced by nanomanufacturing would equal the price of potatoes. The resulting nanoworld, in which everyone is wealthy because of the drastic reduction in the cost of goods, would flummox economists, those scientists of scarcity. A jumbo airliner could be purchased for the current price of an automobile. A homeowner would pour acetone into a household manufacturing system, similar in appearance to a microwave oven. An hour later, out would come a computer, a television set or a compact-disc player. A home food-growing machine could rapidly culture cells from a cow to create a steak, a godsend to the animal-rights movement."
The potential payoff from the development of nanotechnology is indeed quite large.
"Minuscule submarinelike robots made by assemblers would extend life or reverse aging by killing microbes, by undoing tissue damage from heart disease or by reversing DNA mutations that cause cancer; the nanomachines would help revive bodies preserved in cryogenic storage by repairing frostbite damage to the brain and other organs. (Drexler, in fact, plans to sign up to have his body frozen after death.) Engines of Creation even speculates about nanotechnology providing the basis for telepathy or for radically changing one's body."
There is experimental work today involving the interpretation
of electroencephalograms to control equipment. There is also work
aimed at alleviating certain handicaps by connecting an
electronic control system to the nerves controlling a muscle if
the person's own nervous system is unable to stimulate those
nerves; and work aimed at directly stimulating the auditory
nerves of a deaf person. Future extensions of this type of
research can be envisioned. Engines,
in a section explicitly aimed at disentangling what is likely to
be feasible from what is not, discusses the feasibility of
directly connecting transducers to nerves, making possible what
might be viewed as a limited type of "telepathy."
(In some sense, ordinary speech is a type of
"telepathy," as it permits the thoughts in one mind to
be transferred to another mind).
For further information about cryonics, see http://merkle.com/merkleDir/cryo.html.
"On the dark side, assemblers would streamline the production of superweapons, allowing rapid fabrication of a tank or a surface-to-air missile. And then there is the "gray goo" problem - the possibility that nanodevices might be designed to replicate uncontrollably, like malignant tumor cells, and reduce everything to dust within days."
"Military applications of molecular manufacturing have even greater potential than nuclear weapons to radically change the balance of power." --- Admiral David E. Jeremiah, USN (Ret), former Vice Chairman of the Joint Chiefs of Staff.
"Ruminations in Engines of Creation about gray goo and extended life spans provoked guffaws from many scientists. In 1992 Drexler responded to the criticism with Nanosystems, which attempts to give his tiny machines a grounding in the underlying essentials of physics, chemistry and biology. Nanosystems heavy technical emphasis was a plea from Drexler for respectability. The subtext: I am not a flake. But the book remains largely an object of curiosity to the scientific community. It has been hard for many scientists, engineers and technicians to take seriously a section at the end that shows components of assemblers similar to large-scale mechanical devices. For example, a six-legged platform imitates the ones used to tilt flight simulators into different attitudes of yaw, pitch and roll. Its size: only 100 nanometers across, no bigger than a virus. "This is not science - it's show business," says Julius Rebek, a leading researcher in the chemistry of self-assembly at M.I.T."
We're now back to Drexler this, Drexler that, and biased
adjectives. Drexler makes a "plea," has
"ruminations," etc. etc. etc. The basic technical
issues are once again pushed aside to focus on the irrelevant.
Scientific American faces a
difficult problem with Nanosystems.
Nanosystems: Molecular Machinery, Manufacturing and
Computation is an extension and enlargement of Drexler's MIT
Ph.D. thesis. Published by Wiley in 1992, it provides a dense and
technical analysis of the issues involved in nanotechnology.
Drexler negotiated with Wiley to get a purchase price low enough
that students could afford it. Over 10,000 copies are now in
print. No technical errors of significance have been found,
despite several years of exposure to the entire scientific and
technical community, including both critics and supporters.
As the reader might appreciate, if there were
any serious technical objections to Nanosystems Scientific
American would have been more than happy to point them
out. Instead, we are treated to a series of non-statements about Nanosystems:
it "attempts", it is a "curiosity," it is
"hard" to take "seriously." Readers
are invited to look at a copy and see for themselves.
The non-statement "This is not science -
it's show business" raises an issue that is usually
overlooked: people often ask if a particular piece of
experimental work "brings us closer to nanotechnology."
The answer is: it depends on what you think nanotechnology is
going to look like. If, as the Rebek quote suggests, he thinks
that nanotechnology will not use positional control then
logically he should conclude that current experimental research
that demonstrates the basic capabilities of positional control is
not moving us closer to nanotechnology. He would presumably argue
that the recent
work at IBM Zurich which demonstrated an ability to position
individual molecules at room temperature -- while perhaps being
very nice science -- is in fact irrelevant to the goal of
developing nanotechnology. This, of course, contradicts the claim
made earlier in the story that experimental research with
scanning probe systems is "real." Evidently, the
"reality" of a particular line of research depends on
what you think nanotechnology is going to look like.
The denigration of
positional control (which sometimes appears in the self-assembly
research community) is unjustified. We note that positional
control is used extensively at the macroscopic scale. Indeed,
Homo sapiens is viewed as the tool using species -- without the
ability to hold tools and apply them in a positionally controlled
fashion to workpieces ranging from flint knives to computer
keyboards we would no doubt still be shivering in a cave. Recent
experimental work has indeed demonstrated that positional control
can be applied at the molecular level, and theoretical work has
long supported exactly the same conclusion. Recent
theoretical work on the use of a positionally controlled ethynyl
radical to selectively abstract chosen hydrogen atoms from the
diamond (111) surface provides the most recent and specific
support for the idea that positionally controlled molecular tools
can provide a mechanism for synthesizing specific complex
diamondoid structures.
"Despite his alienation from mainstream science and engineering, Drexler continues to amass devotees, particularly among computer scientists enticed by the prospect of making tangible anything they can specify with a set of three-dimensional coordinates. "Nanotechnology will reduce any manufacturing problem, from constructing a vaccine that cures the common cold to fabricating a starship from the elements contained in sea water, to what is essentially a software problem," writes physicist and science-fiction author John G. Cramer."
There is growing interest in nanotechnology from the
mainstream, though the significance of "alienation"
from the mainstream is debatable. When I submitted the first
paper describing a public key system in 1975, it was rejected by
the referee with the note that "I am sorry to have to inform
you that the paper is not in the main stream of cryptography
thinking...." That was, of course, the point of the paper.
Public key cryptography introduced a new paradigm -- a point that
the referee was too dull to grasp.
Computer scientists do indeed seem to grasp
nanotechnology quite easily. Bright folk.
"Silicon Valley, that mecca for aficionados of things small, hosts a disproportionate number of nanoists. Apple Computer has helped sponsor the Foresight Institute's conferences - the most recent one last November drew more than 300 people, double the attendance of the 1993 gathering. A researcher at the Xerox Palo Alto Research Center, Ralph C. Merkle, who made a name for himself in computer cryptography, spends his time creating models of molecular machine components. (Merkle has already signed up to have his head frozen.)"
Silicon Valley does indeed have a relatively high proportion
of highly trained, technically competent people who are used to
making their own judgements about new technologies.
Again, further information
on cryonics is available on the web.
"In 1991 John Walker, the reclusive founder of Autodesk, a California software company, donated $175,000 to help start the Institute for Molecular Manufacturing, a research organization. Most of the institute's grant money has gone to pay Drexler to work on projects such as computer simulations of molecular gears, bearings and other parts."
Drexler has not sought grants from government sources but has instead sought to fund research and education through nonprofit public foundations. He has done a remarkable amount on a shoestring budget -- I encourage donations to the Institute for Molecular Manufacturing and the Foresight Institute.
"The Drexler following includes speculative thinkers such as artificial-intelligence pioneer Marvin L. Minsky. Nanotechnology also seems to inspire government laboratories seeking to remake their image. Oak Ridge National Laboratory has let one of its modeling groups devote extensive effort to simulations of molecular bearings and shafts. Administrator Daniel S. Goldin of the National Aeronautics and Space Administration sees nanotechnology as a means of building smaller and lighter space vehicles. And the NASA Ames Research Center has scheduled a workshop for this spring to examine how its supercomputers might be used to provide models of nanodevices. Perhaps the most noteworthy trend - or the most disturbing one, to critics of the nanoist vision - is the appeal that the technology holds for students."
Students are indeed taking a strong interest in
nanotechnology:
"A new scientific truth does not triumph by convincing its
opponents and making them see the light, but rather because its
opponents eventually die, and a new generation grows up that is
familiar with it." Max Planck
"The human mind gets creased into a way of seeing things.
Those who have envisaged nature according to a certain point of
view during much of their career, rise only with difficulty to
new ideas. ... Meanwhile, I observe with great satisfaction that
the young people are beginning to study the science without
prejudice." Antoine Lavosier.
"Study groups in nanotechnology have established themselves at universities such as M.I.T. and the California Institute of Technology. "It's captured the imagination of bright, young scientists and engineers," says William A. Goddard III, a professor of chemistry and applied physics at Caltech. Goddard, an admirer of both Drexler and Merkle, occasionally works with them on simulations of molecular machine parts."
Computational chemists generally seem able to absorb the idea
of nanotechnology quite easily. One reason for this is that they
have the training and the tools to evaluate proposals for
molecular machines. Give them a PDB file and tell them that it
describes a molecular structure that's supposed to do something
useful, and they'll happily model it. Sometimes they come back
and tell you that it works and it's a great design. Sometimes
they torture it in horrible ways and it behaves in some strange
and unexpected manner.... at which point, it's back to the
drawing board.
The story attempts to portray "actual
scientists" as a group that is disjoint from
"acolytes," "devotees," and
"disciples." Unfortunately, Professor Goddard, head of the
Materials and Process Simulation Center at Caltech, is rather
obviously neither a disciple nor an acolyte. A great many others
supportive of nanotechnology are quite capable of analyzing the
technical issues and drawing their own conclusions -- and have.
The analysis of the hydrogen
abstraction tool has involved many people, including: Donald
W. Brenner, Richard J. Colton, K. Eric Drexler, William A
Goddard, III, J. A. Harrison, Jason K. Perry, Ralph C. Merkle,
Charles B. Musgrave, O. A. Shenderova, Susan B. Sinnott, and
Carter T. White. The institutions involved include the Materials
and Process Simulation Center at Caltech; the Department of
Materials Science and Engineering, North Carolina State
University; the Institute for Molecular Manufacturing; the
Department of Chemicals and Materials Engineering, University of
Kentucky; the Chemistry Department, United States Naval Academy;
the Naval Research Laboratories, Surface Chemistry Branch; and
the Xerox Palo Alto Research Center. None of these people are
"acolytes" "devotees" or
"disciples." Again, the reader is invited to peruse the
list
of conference attendees and decide whether it's possible to
neatly classify people as either "disciples" or
"actual scientists." (Or, as the old joke says,
"What am I, chopped liver?")
"Drexler and his nanoist disciples view molecular nanotechnology as a grand challenge of science and technology. And they comb the pages of journals such as Science and Nature for evidence of research advances that might lay the groundwork toward the ultimate self-replicating assembler. At the Foresight conference last fall, Merkle showed a schematic chart illustrating how the current work being done at a scale below 100 nanometers by chemists and materials scientists might one day lead to nanomachines. Lines on the left of the chart represented experimental approaches, such as probes that can manipulate atoms, tubes of graphite about a nanometer in diameter, and novel types of proteins. On the right side resided lines that corresponded to computer simulations of molecular machine parts for assemblers. In the center appeared a noticeable gap."
Back to Drexler, biased adjectives, hints and innuendoes.
As for the
"schematic chart:" I have
been using charts similar to the one at the left for several
years to illustrate the advantages of pursuing a mixed strategy
for developing nanotechnology. The basic idea is simple:
neither a purely experimental nor a purely
theoretical/computational approach will be most effective. The
strengths of the two approaches are complementary, and their
combined use will move us towards the goal more quickly than
either alone. Scientific American is advocating a
purely experimental approach (which they term "real
nanotechnology") while denigrating the value of
computational and theoretical work.
It is interesting to note that recent work at
IBM Zurich used computational models to design a molecule
that could be precisely positioned on a surface, at room
temperature and in ultra high vacuum. They then verified that the
molecule that the "computational experiments" favored
also worked experimentally. As we seek to build more and more
complex molecular machines, the use of extensive modeling will
become increasingly important. The design of an assembler will
require extensive analysis on a computer if it is to be developed
in a reasonable time frame.
The work at IBM Zurich also illustrates quite
nicely that computational experiments can be used to model
near-term as well as intermediate-term and long-term objectives.
The particular chart I used at the conference had some dotted
lines crossing the gap -- computational models of the
intermediate systems that we'll have to design and build if we
are to move from our present technology base to nanotechnology.
Most researchers whose work moves beyond computer simulations and into the laboratory do not view the challenges of nanotechnology as leading toward the goal of nanoists such as Merkle. A number of them, some of whom even capitalize on the "nano" label in promoting their work, pursue a series of more modest objectives. Differences of opinion about Drexlerian nanoism do not prevent the two camps from occasionally rubbing elbows."
The computational modeling of molecular machine proposals
creates a problem for Scientific American: it
supports the conclusion that nanotechnology is feasible.
Therefore, it's necessary to discredit computational models
without actually addressing their substance.
Keep in mind that not all simulated molecular
machine proposals work. Far from it. Failures are in fact
numerous but are usually not published (a fact I can attest to
from personal experience). What's interesting is that, with
effort, it's possible to develop molecular machine designs that
survive computer simulation.
As for pursuing "modest"
objectives: it's not unusual when selecting a goal to estimate an
expected return: multiply the payoff if it works times the
estimated probability that it will work. Goals with a modest
payoff require a high probability of success before pursuing them
can be justified. Goals with a high payoff should be pursued even
if the probability of success is modest. Nanotechnology will have
an extremely high payoff and the evidence to date (contrary to
anything Scientific American might imply) suggests a
very high probability of payoff. The only disadvantage is that
there is still significant uncertainty about how long it
might take and what resources might be needed.
"Harvard University chemistry professor George M. Whitesides presented a review of his work at the Foresight conference. Whitesides investigates how simple natural objects self-assemble by minimizing thermodynamic instabilities at a surface, such as those between air and water [see "Self-Assembling Materials," by George M. Whitesides; SCIENTIFIC AMERICAN, September 1995]. At the meeting, Whitesides described how he and his colleagues have used self-assembling hydrocarbon molecules, called alkanethiols, to form ordered rows on a gold surface. They have demonstrated how his fabrication method might be used in a process to pattern far thinner circuit lines on a computer chip that can be achieved through conventional lithographic methods. Eventually, self-assembly of small silicon cubes that contain devices that alter information might lead to news methods for manufacturing computer processors."
A good talk. Whitesides' expertise in self-assembly was
evident. Self-assembly is a powerful method of synthesizing a
very useful range of structures. The reader should keep in mind
that self-assembly is not, by itself, nanotechnology. A
simple way of seeing the difference is to note that a diamondoid
structure cannot be self-assembled, and that a self-replicating
manufacturing system is based on more than just the principles of
self-assembly (this is true even for a cell).
While self-assembly is not nanotechnology, it
should be possible to self-assemble systems that are sufficiently
powerful that they could in turn build more powerful systems.
"Whitesides does not see the goal of his work as edging toward the assembler. He distinguishes between his investigations into self-assembling monolayers and the still distant goal of achieving self-assembly by following a coded set of instructions. Biological cells use this latter approach to make copies of themselves, and so would nanoassemblers. 'What makes [Drexler's vision] exciting is self-replication, and at the moment, it is pretty much science fiction,' Whitesides says. 'Even after a fair amount of thought, there's no way that one could see of connecting this idea to what we know how to do now or can even project in the foreseeable future.'"
Another non-statement: self-replication is "science
fiction." This statement is made more intriguing by the fact
that the planet is covered by self-replicating systems.
Again, an
introductory page about self replicating systems is available
on the web.
"The complexity of making objects with individual molecular building blocks may eliminate any of the dramatic cost savings envisioned by the nanoists, except in a few clearly delineated technological areas. Fabricating computer chips has already become a form of engineering the small, with the tiniest circuit elements measuring less than a micron. The cost of a new semiconductor plant now reaches into the billions of dollars, in part because of the technical challenges posed by the need to craft ever smaller features onto the surface of a chip. Chipmakers can still justify the added expense because packing circuits more densely lead to higher computational performance and ultimately lower costs. For most other goods, nanotechnologies may receive tough competition from Mother Nature. 'Drexler's grand vision is a nice one, but sometimes some of the specifics are not entirely correct,' comments Jane A. Alexander, who established the nanoelectronics program at the Advanced Research Projects Agency. 'I once heard him say we'd make tables out of nanotechnology. Wood is awfully cheap, and trees do it very nicely.'"
Present manufacturing methods involve a very big and
increasingly expensive thing (a semiconductor fabrication
facility) to make a very small and not too expensive thing (a
computer chip -- note that silicon computer chips on a per pound
basis are quite expensive: hundreds of thousands or millions of
dollars per pound). In sharp contrast, we have examples of very
small things (cells) that make other very small things (other
cells) and do so inexpensively (perhaps a dollar a pound).
Nanotechnology proposes to use self-replication to achieve low
cost, and certainly does not propose to use something
resembling current semiconductor fabrication methods.
The question posed is therefore quite crisp:
if self-replicating systems able to manufacture a wide range of
interesting and valuable products are feasible, then low cost
manufacture of the products that such systems can make follows as
a consequence. If no such system can be designed and built, then
this proposal for achieving low cost is infeasible. Present
evidence that self-replicating systems are feasible includes the
existing biosphere, the theoretical studies of von Neumann,
the NASA Ames study Advanced
Automation for Space Missions, the work of Drexler in Nanosystems,
articles by
Ralph C. Merkle, and other articles too numerous to mention
(see the references in the NASA study and elsewhere).
To quote the open letter to Scientific
American by R. Freitas: "As co-editor of the historic
1980 NASA study of self-replicating machine systems (Advanced
Automation for Space Missions, NASA CP-2255, 1982), I was
astonished to see this concept decried as "science
fiction" in the pages of Scientific American, a magazine
which has been publishing articles praising the idea for nearly
half a century (e.g. John G. Kemeny, "Man Viewed as a
Machine," (Apr 1955):58-67; Edward F. Moore,
"Artificial Living Plants," (Oct 1956):118-126; L.S.
Penrose, "Self-Reproducing Machines," (Jun
1959):105-114). NASA's robots-assembling-robots replication
demonstration, first proposed in 1980, now occurs daily in
Japanese factories. Self-reproducing software such as worms and
viruses prowl the Internet, preying on unprotected PCs, networks
and mainframes. Machine replication has been extensively studied.
It is eminently feasible."
The argument against the feasibility of such
systems is that they are "science fiction."
As for the implied assertion that tables will
continue to be made from wood: the table on which I am typing
this is made from metal and plastic (already a change from
centuries past) and is quite heavy -- I'd have a hard time moving
it if I wanted to rearrange my furniture. I'd like to have a
table that was light enough so that I could easily move it -- a
diamondoid table would be quite nice. New technologies often
replace old, even for humble applications.
"Keeping every atom in its place may also prove exceedingly onerous at the atomic level. David E. H. Jones, a researcher in the department of chemistry at the University of Newcastle upon Tyne, who may be best known as the author of the irreverent "Daedalus" column in Nature, has provided a pointed critique of the idea that individual atoms and molecules could serve as construction elements in the ultimate erector set. Jones made his case a year ago in a review of a popular book about Drexler by science writer Ed Regis, called Nano. Regis's account generally treats the chief nanoist's ideas favorably."
Shortly after Jones made his case, we responded and rebutted his arguments point by point on the web. As Jones was grossly ignorant of the field, this was a relatively straightforward exercise. The existence of this rebuttal was conveniently omitted in Scientific American's story.
"Jones describes the contortions often required to achieve atomic control of matter. In 1989 two IBM researchers penned their employer's acronym by manipulating 35 xenon atoms with a scanning tunneling microscope - a device that dragged the atoms across a nickel surface. The atoms moved because of chemical bonding interactions that occurred when the microscope's tungsten tip came to within a tenth of a nanometer or so of each atom. Jones notes the difficulties involved: The IBM logo was created in an extremely high vacuum at the supercooled temperature of liquid helium using inert xenon atoms. Outside this rarefied environment, the world becomes much less stable. 'Single atoms of more structurally useful elements at or near room temperature are amazingly mobile and reactive,' Jones writes, 'They will combine instantly with ambient air, water, each other, the fluid supporting the assemblers, or the assemblers themselves.'"
This
issue was specifically addressed in the rebuttal. Briefly (a)
he has confused the concept of "atomically precise"
with the method of manufacture, (b) proposals that call for the
use of highly reactive compounds also call for an inert
environment, such as vacuum or a noble gas, and (c) some specific
reactions that involve single atoms, such as the
hydrogen abstraction tool discussed earlier, have been
modeled both with ab initio quantum chemistry and molecular
dynamics. It is obvious from reading Jones' comments that he was
literally unaware of this body of work.
While Jones' excuse is ignorance, the writer
of this story has no such excuse. A video of the molecular
dynamics simulation of the hydrogen abstraction tool successfully
abstracting a specific hydrogen atom from the diamond (111)
surface at 300 K (approximately room temperature) showed in clear
and graphic detail the influence of thermal noise, the specifics
of the chemical reactions involved, the nature of the environment
in which the reaction should take place (vacuum), and other
details. The video was shown at the conference that the writer
attended, and in a telephone conversation he told me he had seen
it. Other presentations at the conference discussed specific
reactions for the mechanosynthesis of diamondoid structures,
including detailed descriptions of both the reactions and the
environment in which they should take place. As these talks did
not include an easily understood video, we might forgive the
writer for having failed to understand their significance.
"Jones believes that the nanoists fail to take into account critical questions about the thermodynamics and information flow in a system of assemblers. 'How do the assemblers get their information about which atom is where, in order to recognize and seize it? How do they know where they themselves are, so as to navigate from the supply dump [where raw atomic material is stored] to the correct position in which to place it? How will they get their power for comminution [breaking up material] into single atoms, navigation and, above all, for massive internal computing?' The list continues before Jones concludes, 'Until these questions are properly formulated and answered, nanotechnology need not be taken seriously. It will remain just another exhibit in the freak show that is the boundless-optimism school of technical forecasting.'"
This was the summing-up paragraph of Jones' book review. As he stated these various problems in the earlier paragraphs, and as we responded paragraph by paragraph to his arguments, we here refer the reader to our response. A brief discussion of a method for dealing with the positional uncertainty caused by thermal noise in the context of an assembler is available.
"The nanoists' response to this fusillade is simple: read Drexler's technical tome Nanosystems, which contains a response to virtually any general point raised by detractors. Acoustic waves, for example, can be used to supply power to assemblers, an answer to one of Jones' objections."
The response was a
point-by-point rebuttal that Scientific American
was aware of and chose not to mention. A charitable attitude
towards this omission might be that they simply didn't know about
its existence. Unfortunately for this hypothesis, Scientific
American made extensive use of resources on the web (I
know -- I spent more time than I care to remember on the
telephone discussing the contents of specific web pages with
them). Further, when they asked for a critic, I provided them
with Jones' name and suggested they look at my web page rebutting
his arguments. I followed up this discussion by sending them an
e-mail which said: "David Jones ... wrote a review of
"Nano!" in Nature that was rather bluntly
critical. I don't think much of his opinions, but I don't think
he's retracted them. There's a link to my
comments from the "reactions" page, along with the
reference to his article."
As Jones' book review is literally the only
published criticism of the feasibility of nanotechnology that Scientific
American mentioned, as they devoted several paragraphs to
it, and as Jones' criticism contained gross errors that should
have been immediately obvious to anyone who attended the Fourth Foresight
Conference on Molecular Nanotechnology, their omission of the
rebuttal is ... shall we say delicately ... odd.
Of course, reading Nanosystems is
excellent advice. Had Jones done so before writing his attack it
might have made responding to his "questions" a bit
less tedious.
"Drexler contends that his critics, with their need to focus on new products or the next grant-funding cycle, have trouble thinking far enough into the future. 'To people outside who don't understand that you're talking about the year 2020 or whatever, these ideas raise confused, unrealistic expectations about the short term,' Drexler maintains. 'That makes researchers uncomfortable because it's not a yardstick they want to be measured by. It also brings in ethics and the future of the human race, which are not the usual cool, scientific, analytical concerns.'"
As a general rule of thumb, groups or individuals that have an
interest in nanotechnology have long planning horizons. NASA, for
example, often plans missions many years or even decades in
advance. Members of the National Space Society typically view the
settlement of space as a multi-decade undertaking. The Department
of Defense has internal planning documents, such as the Project
2025 report, that consider where the world might be, both
politically and technically, 30 or more years from now. Medical
applications of nanotechnology are of interest to individuals
if they can be developed within their lifetime: many of us expect
to live at least a few more decades. Many people care about the
world that their children will live in. Members
of the environmental movement want to preserve the beauty and
majesty of nature not only for the next year or two, but for
centuries and more. It's not surprising, therefore, to see
interest in nanotechnology from members of these groups.
Those who express little interest in
nanotechnology usually have planning horizons of ten years or
less. One very good experimentalist explained to me that he
wasn't interested because he wasn't going to be here in a decade.
While sympathizing with his predicament, many of us both expect
to be here for several more decades and also care about at least
some aspect of the future even after that.
As a society we need to be thinking not only
about the next month or the next year, but also the next few
years, the next decade, and the next few decades.
"For engineers who build things, finding the relevant page in Nanosystems is not enough. Drexler touts his work as 'theoretical applied science': research constrained only by physical law, not by the limits of present-day laboratory or factory manufacturing capabilities. To hard-nosed engineers, though, the juxtaposition of 'theoretical' and 'applied' quickly becomes an oxymoron. Their response to the author of Nanosystems? Come back when you can tell me how to make those things."
I believe a "hard-nosed" engineer, pretty much by
definition, would have a planning horizon of under 5 years. In
order to motivate such individuals it is necessary to either (1)
describe something that they can do in the next few years that
will provide them with some direct benefit (a product that makes
money, for example) or (2) pay them money to work on something
that will not have any significant pay-back in 5 years.
As there are many sharp engineers who are
actively looking for projects with significant short term payoffs
there is little need to tell them what they should be doing --
they'll see the payoff and will work for it pretty much
independent of any long term considerations. It is reasonable to
expect that a great deal of useful work will be done for such
reasons.
Unfortunately, it is not reasonable to assume
that years or even decades of research and development motivated
by nothing more than short term payoffs is the best or even a
feasible way to develop a relatively complex system that has a
high payoff. Anyone familiar with chess playing programs will
understand this -- it's called the horizon effect. If the program
looks only at the next five moves it will overlook a checkmate
that is ten moves deep. Continued tactical play based on a five
move look ahead might or might not ever detect the deeper
strategy -- it will certainly not detect it as quickly.
"The accumulation of small details may doom the best theories for small machines. Phillip W. Barth, an engineer at Hewlett-Packard, characterizes simulations of molecular bearings as 'computer-aided speculation.' 'The holes are bigger than the substance,' he says of Nanosystems. 'There's a plausible argument for everything, but there are no detailed answers to anything.' Barth is a leading engineer in micromechanics, a field that builds microscopic sensors and machines from silicon [see "Silicon Micromechanical Devices," by James B. Angell, Stephen C. Terry and Phillip W. Barth; SCIENTIFIC AMERICAN, April 1983]. Barth observes a lack of discussion of a number of basic engineering considerations that could make many of Drexler's nanodevices impossible to build. Drexler's nanobearings may be molecularly stable. But Nanosystems, he notes, does not address the stability of structures synthesized during intermediate steps in building the bearings."
So far, the critics have provided no unanswerable (or even
very difficult) problems -- small or otherwise -- to accumulate.
Barth's concern that intermediate structures must be stable is a
significant design constraint -- just as the concern that the
intermediate states of (say) a house must be stable is also a
significant design constraint. It might be prudent to assemble a
wall before trying to raise it, and the roof should
presumably be put in place after there are adequate
supports to hold it. The argument that nanotechnology is
infeasible because the stability of intermediate structures
imposes a design constraint is remarkably weak, particularly when
we consider the astronomically large number of intermediate
structures that are possible if we wish to make a structure of
any reasonable size.
We consider one approach for dealing with the
stability of intermediate structures. We note, firstly, that the
stability of structures with no dangling bonds and in which the
strain on the bonds is relatively modest can be evaluated
relatively easily. We could adopt a set of design rules
enumerating a set of allowed local structures and setting
explicit limits on allowed local strain. The simplest design
rules would be for structures that closely resembled diamond.
Such structures have been proposed in Nanosystems and
elsewhere, and arouse (as Barth notes) little concern.
Next, we note that during synthesis we might
create a structure which has dangling bonds. Such dangling bonds,
particularly if there are many of them, can make the analysis of
the stability of the structure difficult. To consider a well
known example, the 7x7 reconstruction of the non-hydrogenated
silicon (111) surface is not generally viewed as intuitively
obvious and took considerable effort to work out. If we used
intermediate structures similar to the this, we'd have our work
cut out for us.
A simple strategy, therefore, is to ban
intermediate structures with many dangling bonds. Thus, we deem
the non-hydrogenated silicon (111) surface as "too complex
to analyze."
Can we still synthesize a very wide range of
structures within this constraint? One strategy for doing so
follows. We start with a structure which has no dangling bonds.
We make a small local modification to the structure by (if
needed) first abstracting a few hydrogen atoms (or otherwise
making a very small region reactive). The stability of such a
small region can be analyzed adequately using current ab
initio quantum chemistry methods. Further, if we are
synthesizing a large structure in which there are a few common
repeating and relatively small patterns, we would expect that the
local structure representing the local intermediate state would
also be repeated -- so we need only analyze in detail a few small
"unstable" regions in great detail to ensure that they
would not engage in some unexpected reconstruction. Finally,
having made the desired local modifications, we determine if
there are any dangling bonds. If there are, we cap them off by
adding hydrogens to them.
In brief summary, this strategy changes the
problem from "Start with relatively simple and stable
structure X and synthesize relatively simple and stable structure
Y (where Y is very different from X)" to "Start with
relatively simple and stable structure X and synthesize
relatively simple and stable structure X' (where X' differs from
X by only a few atoms); then from X' synthesize relatively simple
and stable structure X'' (where X'' differs from X' by only a few
atoms); then ..... ; then from XN synthesize
relatively simple and stable structure Y (where Y differs from XN
by only a few atoms)."
This kind of issue has been well known in the
nanotechnology research community for some years. The 1993
Foresight Conference had a demo of the MASS (Molecular Assembly
Sequence Software). The purpose of this software is to
provide an aid for the development of synthetic sequences
intended to make diamondoid structures. The idea is to start with
the final structure and work backwards to some starting structure
by a sequence of steps, each one of which can be mediated by a
suitable positionally controlled molecular tool. Chemists will
recognize this as a variant of retrosynthetic analysis, but with
the added assumptions that (a) synthesis takes place in vacuum,
(b) the object being synthesized is a stiff diamondoid structure
and (c) positionally controlled molecular tools are available.
Discussions of the issues involved preceded 1993 by several
years.
"Unresolved details, moreover, may not be so trifling. 'Energy is a fundamental concern,' Whitesides declares. 'It is no good to say it comes from somewhere - acoustic waves or whatever. If we can forget the details of energy supply, we have a perpetual motion machine.'"
An interesting series of statements. Energy is indeed
important, but providing a source for it is "no good?"
Followed by the implied claim that we're providing perpetual
motion? To think that Drexler and myself; the researchers at
Caltech, Oak Ridge, NASA Ames, NRL, NCSU and elsewhere; the
students at MIT, Caltech, and other universities; and the many
others who have looked over the proposals for molecular machines;
would have anything to do with a proposal that either required or
provided perpetual motion is not only insulting, it's ... very
... very ... dumb! Though not all that
unusual for this story.
Being charitable,
however, perhaps there is a concern about how to convert
mechanical energy into chemical energy. A simple example of a
mechanosynthetic process to do this is the conversion of a
precursor to the hydrogen
abstraction tool to the activated form. Consider the
precursor shown to the left. Clearly, if we pull hard enough on
the two handles, something will break. If we design in a
"weak link," then it will break where we want it to
break -- in this case, between the silicon and carbon. The result
is the activated hydrogen abstraction tool, which is chemically
very reactive.
This general mechanism -- pulling on a
structure until something breaks -- can be used to create a wide
variety of radicals. It is a simple method of converting
mechanical energy into very reactive chemical structures.
"The present inability to build an assembler - coupled with elaborate speculation about what the future may hold - gives nanotechnology a decidedly ideological or even religious slant, in Barth's view. In early January he posted a message to an Internet bulletin board (sci.nanotech) suggesting that subscribers comment on whether molecular nanotechnology has the makings of a mass social/political movement or a religious faith in the traditions of Marxism or Christianity. Barth bolsters the case for nanoism as a form of salvation by citing a passage from a new magazine called NanoTechnology: 'Imagine having your body and bones woven with invisible diamond fabric. You could fall out of a building and walk away.'"
It's useful to recall, at this point, that the original
question was whether or not nanotechnology, defined as the
ability to inexpensively make most structures that are consistent
with natural law, is or is not feasible. There are very good
reasons to believe that it is. The relevance of Marxism and
Christianity to this technical question is, shall we say,
unclear. The inclusion of random quotes to "bolster"
this red herring serves little useful purpose.
The nanoists' legacy may be to stoke science-fiction writers with ideas for stories. The latest genre in science fiction employs nanotechnology as its centerpiece. A follow-on to the cybernetic fantasies of authors such as William Gibson, it is sometimes even called "nanopunk." The world depicted by nanowriters goes beyond cybernetic mind control and downloading one's brain into a computer. It postulates ultimate control over matter. "It seems like nanotech has become the magic potion, the magic dust that allows anything to happen with a pseudoscientific explanation," says Ist-van Csicsery-Ronay, Jr., an editor of the journal Science-Fiction Studies, published by DePauw University."
Again, the objective is to draw attention away from the technical issues. Space travel is feasible even if not all science fiction stories about space travel are correct or even sensible.
"A collection of 'nano' stories that appeared last year features the imaginings of noted science-fiction writers, such as Poul Anderson. The volume, Nanodreams, even contains an introductory essay by Drexler on the merits of science fiction as a means of exploring the societal implications of a nanotechnological future. 'Saying something sounds like science fiction should not be regarded as a form of dismissal,' Drexler said in a recent interview. 'Much of what science-fiction writers described in the 1950s happened, and you need to distinguish between antigravity and flying to the moon, between time travel and making a robot that works in the factory.'"
Drexler's comment about sums it up.
"Nanodreams includes a story in which the pain experienced by a fetus during an abortion is telecommunicated to nanomachines that reproduce the sensation within the father of the child - and then, finally, kill him. Another nanotale describes a company that has just achieved a breakthrough by making nanomachines that can repair tissue damaged by a bullet wound. In one scene a poster on a laboratory wall depicts Albert Einstein handing a candle to Drexler."
And the relevance of this to the technical feasibility of nanotechnology is?
"The fantasies of nanoists posted on Internet bulletin boards and World Wide Web sites often outstrip the imaginings of the best science-fiction writers. Take the often discussed idea of a utility fog: nanobots that link together to create materials and objects in a desired form and shape, from paint to furniture. 'When you got tired of that avant-garde coffee table, the robots could simply shift around a little, and you'd have an elegant Queen Anne piece instead,' reads one description on the Web."
More biased language. How would someone from 1896 react to what is common place today?
"Chemistry has distant roots in alchemy, the belief that transmutation of materials will bring health and wealth (though perhaps not ultimate mastery of interior decoration). Nanoism resembles a form of postmodern alchemy - and one that awards cash for molecular machine parts. Toward the end of November's Foresight conference, an announcement was made about a new prize, named for Feynman."
Another non-statement. If we eliminate the statements in the story that have no bearing on the technical issues, most of the story simply vanishes. The few statements that remain are easily rebutted.
"The prize of $250,000 comes courtesy of Jim Von Ehr, an executive at Macromedia, a software company in San Francisco, and Marc Arnold, a St. Louis venture capitalist. It is to be awarded for the fundamental breakthroughs that will usher in the era of molecular nanotechnology: a robot arm and a computing component for an assembler."
Historically, prizes have been very effective at motivating research in a field.
"For the time being, the nanoists can only wait for these breakthroughs to arrive, while continuing to formulate their computerized models of molecular machine parts. It may be a long time coming. In fact, Drexler himself has said that his fortitude has been weakened by jibes from critics and that he might consider a calling other than nanotechnology. "I'm tired of it," he says."
As the reader might appreciate, the kind of ad hominem attacks
used in this story would make anyone tired. The fact that
researchers pursuing fundamentally new and valuable objectives
commonly have to endure such an ordeal is a societal problem of
great significance. How many discoveries were not made, or were
delayed, because the people in a position to make them decided it
just wasn't worth the pain?
The "waiting for breakthroughs"
theme is based on the view that science advances by unexpected
and serendipitous discoveries. If the purpose of scientific
research is to discover new laws of nature and unexpected
phenomena, then this view is reasonable. The important event is
the "breakthrough," i.e., the unexpected discovery of
evidence showing that our existing understanding of nature is
either incomplete or in fact incorrect.
However, the whole thrust of nanotechnology
is to develop what is feasible within the framework of existing
and accepted natural law. This is a very different objective from
the "pure science" objective of determining what those
laws are. We assume that existing natural law -- supported by a
great deal of experimental evidence -- is correct. When there are
serious questions about how to apply existing natural law to a
particular structure, we avoid using that structure. An example
of this principle was given earlier, where the 7x7 reconstruction
of the silicon (111) surface was deemed "too complex,"
and the structure was therefore avoided. This principle has broad
applicability in the design process where we are free to avoid
designs that create problems.
The idea of accidentally and serendipitously
discovering a Saturn V launch vehicle illustrates the absurdity
of thinking that the pursuit of pure science is the only
worthwhile objective. Other objectives are also worthwhile and
call for other methods. In particular, if we think that
nanotechnology will involve the design and manufacture of
relatively complex artifacts, then the design process itself
becomes a significant component of the overall effort. This is a
concept well understood in computer science, where the design
process entirely dominates software development.
It's also intriguing to juxtapose the
criticism that greater analysis and greater detail is needed in
the proposals for molecular machines in general and assemblers in
particular before we can believe they are feasible with the
criticism that providing greater detail and analyzing molecular
machine proposals in depth by computer modeling is a waste of
time. In many instances the "critics" are really
saying: "I don't understand how you could do that."
What they really want is not experimental evidence of a
breakthrough, but a design proposal that illustrates how to
achieve the desired objective within the framework of existing
accepted and well understood natural law.
"Nanoists' convictions about the inevitability of a breakthrough evoke memories of another idea once posed by Feynman, their adoptive mentor. In a commencement speech given to the 1974 graduating class at Caltech, Feynman noted that some Pacific Islanders religiously awaited the return of the U.S. troops who had landed in World War II. He described the elaborate preparations the islanders made for the return of the planes that would bring them advanced technological accouterments and limitless wealth. Fires mark the sides of the runways. A man plays air-traffic controller by sitting in a hut with carved wooden headphones from which pieces of bamboo stick out, like antennas. The believers wait patiently in this preindustrial imitation of an airfield."
Scientific American has another major problem
with Feynman.
If Feynman is taken at face value, then nanotechnology should be
feasible. Jones
mentioned this as one of the arguments in favor of nanotechnology
but then dropped the subject, never explaining either (a) where
Feynman was wrong or (b) why Feynman's observations were
inapplicable.
To get around this awkward problem it is
necessary to neutralize Feynman without actually addressing the
technical issues (or attacking Feynman). The paragraphs above and
below are an attempt to do exactly that.
"'They're doing everything right,' Feynman said. 'The form is perfect. It looks exactly the way it looked before. But it doesn't work. No airplanes land.' Similarly, some scientific endeavors rely on wish fulfillment - and an inability to consider why something may not work, Feynman noted. 'So I call these things cargo cult science,' he concluded, 'because they follow all the apparent precepts and forms of scientific investigations, but they're missing something essential, because the planes don't land.' Until the nanoists can make an assembler and find something useful to do with it, molecular nanotechnology will remain just a latter-day cargo cult."
I think Feynman would not be pleased to have his name and
words attached to this technically empty argument. Readers are
invited to read There's plenty
of room at the bottom and draw their own conclusions.
As noted earlier, Carl Feynman (the son of
Richard Feynman) was also dismayed at the misuse of his father's
essay.
What, then, are the technical arguments advanced against the
feasibility of nanotechnology? One book review by David Jones -- thoroughly
rebutted shortly after it came out -- was quoted. The
existence of the rebuttal was conveniently omitted. A concern
about the stability of intermediate structures was expressed. One
strategy for dealing with this concern was presented here. A few
people were quoted in a way which suggested they had a technical
criticism of the feasibility of nanotechnology. The quotes were
rebutted.
The story has repeated and irrelevant
descriptions of Drexler, ranging from his tone of voice to how he
drinks his tea to whether or not he's "tired of it."
I'd certainly be tired of such biased and inaccurate journalism.
In previous eras the messenger who told the king something he
didn't want to hear was killed. This was no doubt momentarily
pleasing to the king, but it doesn't actually deal with the
substance of the message. And it ensures that no one tells the
king anything he doesn't want to hear -- which can cause
problems.
Scientific American should stop
evading the fundamental technical question: given the currently
accepted understanding of natural law, is nanotechnology feasible
or is it not?
To date, all the evidence of which I am aware
strongly supports the conclusion that it is. The obvious
strategy, given this conclusion, is to go ahead and build
molecular manufacturing systems. It is, however, difficult to
build something before it has been designed, particularly when
the design is itself non-trivial. Thus, we need to design such
systems. The
available tools of computational chemistry are sufficient to let
us do this. That is, it is possible to design and
computationally model both molecular components and indeed an
entire (simple) assembler. While we might debate how hard it
will be to build the resulting designs with our existing tools,
and whether they will work perfectly the first time they are
tried (unlikely), it is clear that this process will
It's likely that we'll actually have to design several such
systems. More powerful systems, able to synthesize complex
diamondoid structures, would seem harder to make directly with
our existing capabilities. Intermediate systems also need to be
designed, systems which are less powerful but more amenable to
near term synthesis. Such simpler systems could then be used to
build the more powerful systems.
In closing: a modest proposal. Let's go ahead
and design these things. It will take a lot of work, and it seems
unlikely to provide a near term payoff (say, in the next five or
ten years). But if we don't design them it's very difficult to
see how we can build them. Ever.
And the long term payoff is immense.
You might wish to send e-mail to Scientific American, with a cc to Foresight Institute (foresight@foresight.org) expressing your opinion of their story. (Polite letters are more effective.) A selection of these letters has been posted at the Foresight web site.