PUBLISHED DECEMBER 2003
Of
Chemistry, Nanobots, and Policy
Introduction
The ability to build
products by
molecular manufacturing would create a radical improvement in the
manufacture of technologically advanced products. Everything from computers to
weapons to consumer goods, and even desktop factories, would become incredibly
cheap and easy to build. If this is possible, the policy implications are
enormous.
Richard Smalley, a
prominent nanotechnologist, has tried for several years to debunk this
possibility. Most recently, he participated in a
published exchange with Eric Drexler, another prominent nanotechnologist,
who has been the primary proponent and theorist of molecular manufacturing (also
called
molecular nanotechnology, or MNT).
This paper examines the
arguments presented by each side and concludes that Smalley has failed to
support his opinion that MNT cannot work as Drexler asserts. Much of Smalley's
discussion is off-topic, and his assertions about the limitations of enzyme
chemistry are factually incorrect—a fatal weakness in his argument. He therefore
does not provide a useful criticism of MNT. Trying to bring the debate back on
topic, Drexler spends most of his time restating his earlier positions. Despite
these problems, the current exchange represents a significant advance in the
debate, since Smalley's new focus on realistic chemistry (instead of the earlier
“magic fingers”) permits detailed analysis of the technical merits of his claim.
The answer to the question
of MNT’s capabilities will have a large effect on nanotechnology policy, and
further research is urgently needed to find this answer. Smalley's factual
inaccuracies and continued failure to criticize the actual chemical proposals of
MNT strongly suggest that his denial of the possibility may be unfounded. In
view of this, while we agree with Smalley that some scenarios of molecular
manufacturing are worrisome, we reject his conclusion that the possibility of
MNT should be denied in order to avoid scaring children.
This paper reviews the history of the MNT
debate, analyzes the technical arguments on both sides, then briefly discusses
the feasibility and desirability of further research and the potentially
disastrous implications of continuing to ignore the possibility of molecular
manufacturing.
History of the Debate
Molecular nanotechnology
was first proposed by Richard Feynman in 1959. In a talk entitled “There's
Plenty of Room At the Bottom”, Feynman asserted, “But it is interesting that
it would be, in principle, possible (I think) for a physicist to synthesize any
chemical substance that the chemist writes down. Give the orders and the
physicist synthesizes it. How? Put the atoms down where the chemist says, and so
you make the substance.” In the 1980's, Eric Drexler elaborated on this vision
and called it 'nanotechnology', projecting its consequences in the popular book
Engines of Creation and working out a limited version of programmable
chemistry in his MIT Ph.D. thesis.
In 1992, Drexler expanded
his MIT thesis into the technical book
Nanosystems, which outlined a proposal for
building manufacturing systems based on programmable synthesis of nanoscale
diamond components. This proposal may be labeled limited molecular
nanotechnology (LMNT) to distinguish it from the broader vision of synthesizing
“any chemical substance that the chemist writes down.” LMNT theory was developed
in increasing detail in subsequent years. Meanwhile, commentators, including the
media and science fiction authors, seized on the projected consequences of
unlimited MNT—especially the so-called grey goo scenario in which a
self-replicating nanobot eats the biosphere. Policy organizations, in particular
the Foresight Institute (founded by Drexler), began to call for attention to the
capabilities and problems implied by MNT.
In the mid to late 1990's,
the U.S. and other governments, inspired by the promise of nanotechnology and
the initial scientific research into the nanoscale, began to provide significant
funding for such research. Many scientists discovered that they were doing forms
of nanotechnology and joined the program. This caused a split between nanoscale
technologies that were easy to fund, and molecular nanotechnology, which was not
yet a mainstream field of research. The scientists working on nanoscale
technologies and the administrators funding them had several incentives to try
to discredit molecular nanotechnology, including justifying the current funding
decisions and avoiding any association with grey goo and other doomsday
scenarios.
In September of 2001,
Richard Smalley published an article in Scientific American titled, “Of
Chemistry, Love and Nanobots,” and subtitled, “How soon will we see the
nanometer-scale robots envisaged by K. Eric Drexler and other molecular
nanotechnologists? The simple answer is never.” Smalley asserted that chemistry
is not as simple as Drexler claims—that atoms cannot simply be pushed together
to make them react as desired, but that their chemical environment must be
controlled in great detail. Smalley contrived a system that might do the job, a
multitude of “magic fingers” inserted into the working area and manipulating
individual atoms. He then asserted that such fingers would be too fat to fit
into the required volume, and would also be too sticky to release atoms in the
desired location. He concluded that since his contrived method couldn't work,
the task was impossible in a mechanical system. Drexler and colleagues published
a
point by point rebuttal of Smalley's position, to which Smalley never
responded.
In April of 2003, Drexler
wrote an open letter to Smalley, asserting that Smalley's fingers were no more
than a straw-man attack since Drexler had never proposed any such thing,
accusing Smalley of having “needlessly confused public discussion of genuine
long-term security concerns,” and calling for him to help set the record
straight. In the absence of any response, Drexler followed up with a second open
letter in July, noting that in 1999 and 2003, Smalley had stated the possibility
of building things “one atom at a time,” and asking for closure on the issue.
These letters prompted the
debate published in the December 1 issue of
Chemical and Engineering News. In the second part of this four-part
exchange (the first part being the April letter), Smalley begins by praising
Drexler for agreeing that fingers won't work. Smalley agrees that something like
an enzyme or ribosome (components of cells) might be able to do precise
chemistry—but, according to Smalley, only under water. He then suggests an even
stranger alternative—that Drexler's nanofactory might contain complete
biological systems—and spends most of the space describing the limitations of
underwater chemistry. Finally, he asks, “Or do you really think it is possible
to do enzyme-like chemistry of arbitrary complexity with only dry surfaces and a
vacuum?”
Drexler replies that, as
noted in his book Nanosystems, his
proposal does assert that chemistry in dry surfaces and a vacuum (“machine-phase
chemistry”) can be quite flexible and efficient, since holding a molecule in one
place can have a strong catalytic effect. He mentions chemical vapor deposition
systems as an example of “dry” chemistry, and points out that, “Further,
positional control naturally avoids most side reactions by preventing unwanted
encounters between potential reactants”—in other words, it doesn't take a lot of
subtlety to avoid making the wrong product. Drexler also spends significant
space in his reply talking about other design issues of molecular manufacturing
systems, the need for an integrated and targeted research program, and the
policy implications of failing to act: “The resulting abilities will be so
powerful that, in a competitive world, failure to develop molecular
manufacturing would be equivalent to unilateral disarmament. U.S. progress in
molecular manufacturing has been impeded by the dangerous illusion that it is
infeasible.”
Smalley's final answer is a direct attack on
machine-phase chemistry. It is the most detailed technical criticism that
Smalley has yet published. He claims that chemical reactions must be controlled
through a many-dimensional hyperspace and that this cannot be achieved with
simple robotics. Smalley repeats his claim that although enzymes can do precise
and reliable chemistry, they can only work in water. [This claim is untrue; see
below.] Smalley ends the debate with a two-paragraph appeal to others in the
chemical community to join him in protecting children from being scared by
stories of monstrous self-replicating nanobots from Drexler's dreams.
Technical Analysis of the Debate
If Smalley's goal is to demonstrate that machine-phase chemistry is
fundamentally flawed, he has not been effective; he has not even demonstrated a
problem with Drexler's proposals. Since 1992, Drexler has proposed that dry
machine-phase chemical synthesis can be used to build intricate nanometer-scale
objects. Smalley's strategy, both in the 2001 Scientific American
article and in the current debate, has been to equate Drexler's proposals with
something unworkable and then explain why the latter can't work. Thus Smalley's
comments do not directly address Drexler's proposals, but attempt by example to
show fundamental problems with his underlying theory. However, both of Smalley's
attempts have failed, and the second failure is noteworthy for what it reveals
about the weakness of Smalley's position.
Smalley's 2001 Scientific American article focused on the impossibility
of using molecular 'fingers' to manipulate each atom involved in the reaction.
Drexler has never proposed separate manipulation of each atom; instead, he
claims that much simpler control will suffice in a well-designed robotic system
where chemicals can be kept apart until they are properly positioned. Besides,
as Drexler pointed out in his open letter, enzymes and ribosomes do not need
fingers. Thus challenged, Smalley responded by equating Drexler's proposal not
just with enzymes, but with the entire apparatus of biological life. Smalley
began by agreeing that an enzyme-based system could do precise chemistry, but
then attempted to show that enzymes would not provide the capabilities that
Drexler needed.
When Smalley substituted enzymes for his 'Smalley fingers,' he lost the debate.
According to Smalley, enzymes can only work in water, and underwater chemistry
cannot build technologically interesting materials such as crystals of steel or
silicon. If Drexler plans to avoid water, Smalley asks, "What liquid medium will
you use? How are you going to replace the loss of the hydrophobic/hydrophilic,
ion solvating, hydrogen-bonding genius of water in orchestrating precise 3
dimensional structures and membranes?" But Smalley is flatly wrong about the
ability of enzymes to function without water.
As far back as 1983, an article in Science described enzymes working not
only in other liquids, but in vapor phase without any solvent at all. One of the
authors of that article, Prof. Klibanov, wrote in 1994, "...using an enzyme in
organic solvents eliminates several obstacles that limit its usefulness in
water. For example, most compounds that interest organic chemists and chemical
engineers are insoluble in water, and water often promotes unwanted side
reactions. .... Consequently, once it was established that enzymes can work in
organic solvents with little or no water, R&D in the area surged." In other
words, enzymes often work better without water. Smalley's objection collapses.
In his closing statement, Smalley finally confronts machine-phase chemistry
directly rather than by example. He argues that chemistry requires great
subtlety of control in order to prevent undesired reactions: "You need to guide
the reactants down a particular reaction coordinate, and this coordinate treads
through a many-dimensional hyperspace. I agree you will get a reaction when a
robot arm pushes the molecules together, but most of the time it won’t be the
reaction you want." Smalley is asserting that any chemical reaction can proceed
in a wide variety of ways, depending on each motion of each nearby atom, and
that without the ability to control each atom separately the result of the
reaction cannot be controlled. This may be true for underwater chemistry,
especially protein folding. But in vacuum chemistry without water on stiff
surfaces, it is possible to exclude all or nearly all undesired reactions by
controlling the collective positions of the reactants so that the molecules can
only touch at the location of the desired reaction. Atoms will not magically
jump out of position to spoil the reaction. As Smalley himself stated (in
Scientific American), atoms "move in a defined and circumscribed way."
In his final statement, Smalley asserts, "I have never seen a convincing
argument that [Drexler's] list of conditions and synthetic targets that will
actually work reliably with mechanosynthesis can be anything but a very, very
short list." But the evidence shows that Smalley has not carefully studied
Drexler's work. The dry enzyme result was cited in a 1994 paper of Drexler's. In
addition, Smalley's apparent uncertainty (in his first statement) about whether
Drexler was proposing wet or dry chemistry, and his repeated distortion of
Drexler's proposals, suggest a substantial lack of familiarity with Drexler's
work. Smalley's failure to see a convincing argument can be attributed to this
lack of attention, and does not indicate any identifiable problem with Drexler's
proposals.
In the absence of a cogent objection to respond to, Drexler could only restate
his earlier work. His description appears to be consistent with the description
in Nanosystems, which has not been scientifically criticized in the
decade since its publication. The validity of his position can be inferred from
repeated failures to debunk it. At this point, scientific investigation rather
than debate will be needed to test Drexler's theories; there appears to be no
simple argument that can disprove his conclusions.
Discussion
The question of whether machine-phase chemistry
can be used to construct machines is vitally important. As both Smalley and
Drexler recognize, such a capability would enable radically powerful and compact
manufacturing systems with potentially extreme consequences. We might expect
that both participants in this debate would have put their strongest arguments
forward.
Smalley's task was to demonstrate that Drexler's proposals for machine-phase
chemistry cannot lead to a workable nanoscale manufacturing system. Smalley
began by inventing molecular “fingers” and describing why they don't work. Then,
he invented a nanofactory based on wet chemistry, and described why it cannot
produce many useful products. Not until the end did he address Drexler's actual
proposals, and his argument at that point depended heavily on a clearly
incorrect understanding of enzymes. In addition to being largely off-topic, and
apparently contradicting his own statements of 1999 and 2003 that were
referenced in Drexler's open letter, Smalley’s argument is sprinkled with
factual errors about chemistry, as noted above.
Smalley's final technical criticism of machine-phase chemistry is not
convincing. It appears to be based on the idea that machine-phase chemistry
taking place in vacuum with positional control will have as many unwanted
reaction pathways as wet chemistry, but will have less ability to avoid them.
However, most of the unwanted reaction pathways in wet chemistry are a result of
the presence of water itself, or result from the lack of positional control
experienced by floppy floating biomolecules. It appears that the lack of degrees
of freedom in machine-phase chemistry may eliminate undesired reactions even as
it simplifies the possible pathways of the reactants. In theoretical terms,
then, machine-phase chemistry may be at least as flexible and reliable as wet
chemistry; Smalley's arguments do not seriously challenge this possibility. In
addition, the reduction in degrees of freedom may make it quite a bit easier to
design desired machine-phase reactions than protein-based reactions, since
protein is notoriously complex.
Drexler's task in this debate was to defend his assertions about the feasibility
of molecular manufacturing, in particular Nanosystems, against Smalley's
attack. Unfortunately, Smalley's repeated straying from the topic did not
provide Drexler the chance to respond to meaningful criticism. However,
Drexler's statements are consistent with his own earlier assertions, do not
contradict any known physical law, and address several practical engineering
details of machine-phase chemistry and how to use it in a manufacturing system.
Drexler ends his statements by calling for further research, beginning with an
independent scientific review of molecular manufacturing concepts. This call
clearly is justified by the evidence to date.
Smalley's last word is an appeal to other scientists to close ranks and oppose
further discussion of molecular manufacturing, in order to prevent “our
children” from being scared by the possible consequences. This is, to put it
mildly, unwarranted and premature. Despite access to two forums and three
chances to express convincing arguments against Drexler's theories, Smalley has
been unable to do so. This does not prove Drexler right; however, it raises the
distinct possibility that Smalley is wrong. For Smalley to urge that debate be
terminated at this point is unscientific and irresponsible.
Policy implications
Current nanotechnology policy in the U.S. and several other countries is based
on the belief that molecular manufacturing as described by Drexler is
impossible. Smalley, with his reputation as a Nobel Prize winning chemist and
nanotechnologist, has been a major exponent of that belief. But he has
demonstrated that he is unable to make a cogent case against Drexler's theories.
It is time for independent scientific investigation of mechanical chemistry, not
merely continued authoritative but unsupported scientific statements of
impossibility.
If molecular manufacturing will work as Drexler describes, preparation for its
consequences must begin well in advance. Vehement opposition from credentialed
nanotechnologists has prevented any significant efforts to prepare for the
possible development of this technology, or even to assess whether and when it
might be developed. As nanoscale sensing, manipulation, and chemistry are
developed further, the situation may become rapidly more dangerous. Recent
technical work by CRN has raised the possibility that the final stages of
development may be extremely rapid. Given the poor quality of MNT criticism thus
far, it would be foolish to bet our future on the hope that no new policy will
be necessary, without a much more detailed examination of the theory behind MNT.
Failure to anticipate the development of molecular manufacturing could have
serious consequences. Simple physics theories, conservatively applied, predict
that the technology will be dangerously powerful. A working molecular
nanotechnology will likely require the design and enforcement of policies to
control the use of compact advanced manufacturing systems and their products.
But panicked last-minute policy will be bad policy—simultaneously oppressive and
ineffective. The military implications are even more perilous. Molecular
manufacturing systems are expected to be able to produce weapons as powerful as
nuclear bombs, but much more selective, easier to manufacture, and easier to
use. If a powerful nation suddenly realizes that molecular manufacturing is
possible, and discovers that rival nations are already making material progress,
they may react violently, or may enter into an arms race that will probably be
unstable and thus may result in war with weapons of unprecedented power.
On the positive side, molecular manufacturing may be able to mitigate many of
the world's humanitarian and environmental crises. Advancing its development by
even a year or two could alleviate untold suffering, raising standards of living
worldwide while sharply reducing our environmental footprint. However, rapid and
effective humanitarian use may also depend on sound policy developed well in
advance.
During the past decade, increasingly detailed proposals have been developed for
the architecture and technology of molecular manufacturing systems. Such
proposals cannot be developed fully in the absence of laboratory work and
targeted research, but we now know enough to initiate action based on existing
work. Machine-phase chemistry—the proposal that Smalley has failed to
criticize—can and should be investigated in detailed chemical simulations. The
theories about nanoscale physics in Nanosystems should also be
investigated; such studies may be expected to produce results relevant to other
nanoscale technologies. We can, and should, begin to quantify the expected
capabilities of LMNT-type systems: What substances and devices can they build?
How rapidly can they work? How easy will it be to design products for LMNT-type
manufacturing systems? How much will it cost to create such a system, and how
quickly will that cost decrease over time?
If no flaw can be found in the proposals of limited molecular nanotechnology, it
must be assumed that LMNT may work as described. In the past decade, no flaw has
been found. The proposals are sufficiently detailed to support a much more
thoughtful critical study than has yet been done, and such a study would result
in further refinement of the proposals. The responsible course of action is not
to hide from imaginary monsters, but to direct increasing energy toward
examining both the theory and its implications.
The principal author of CRN's analysis and commentary was
Chris Phoenix, Director of Research. Please
contact us
for more information.
OTHER DEBATE COMMENTARIES:
