The Drexler-Smalley Debate
on Nanotechnology: Incommensurability at Work?
Abstract: In a recent debate, Eric
Drexler and Richard Smalley
have discussed the chemical and physical possibility of constructing
molecular assemblers – devices that guide chemical reactions by
placing, with atomic precision, reactive molecules. Drexler insisted on
the mechanical feasibility of such assemblers, whereas Smalley
resisted the idea that such devices could be chemically
constructed, because we do not have the required control. Underlying
the debate, there are differences regarding the appropriate goals,
methods, and theories of nanotechnology, and the appropriate way of
conceptualizing molecular assemblers. Not surprisingly,
incommensurability emerges. In this paper, I assess the main features
of the debate, the levels of the emerging incommensurability, and
indicate one way in which the debate could be decided.
Keywords: nanotechnology, Drexler-Smalley
debate, molecular assemblers, incommensurability.
1. Introduction 
Many debates about nanotechnology emerge from
particular visions of the field. We find, for example, visions of a
future dramatically changed by the new technology, with the production
of materials and objects with atomic precision in a remarkably short
time by self-replicating nanobots (Drexler 1986); but we also find the
fear that nanotechnology will quickly run out of control, leaving us
powerless behind (Joy 2000). As with most extreme views, it is unlikely
that any of these scenarios is completely correct. However,
particularly in less radical forms, they may capture something right
about certain developments of the field.
In this paper, I examine a recent debate in
nanotechnology, which was also motivated by different visions of the
field. But, in this case, the different visions involved distinct ways
of conceptualizing what is (and is not) feasible in the area, and even
alternative standards of assessment of such feasibility judgments. In
the debate, we find all the interesting features of scientific debates
more generally: curious arguments (and often not unproblematic ones),
powerful images, unexpected conceptual shifts, the use of diverse
standards, and a good bit of rhetoric. What emerges from the exchange
examined here is an interesting perspective on how scientific debates
can be conducted and interpreted – and why, sometimes, it is so hard to
settle them. Nanotechnology, even at the metalevel, never stops to be
The debate involves two significant
the one hand, we have Eric Drexler, one of the visionaries of
nanotechnology. He clearly conceived of a world completely transformed
by the developments in the area. A crucial component of his view takes
central stage in the exchange below: the notion of a molecular
According to Drexler, such an assembler would be able to build
virtually anything with atomic precision and no pollution. His vision
was first presented in the 1980s, in Engines of Creation
(Drexler 1986), with the more technical details articulated later in
the early 1990s, in Nanosystems
(Drexler 1992). Drexler is the chairman and cofounder of the Foresight
Institute, an institution that aims to help prepare society for
advances in technology, with particular emphasis on nanotechnology.
Drexler’s main background is in engineering, and as we will see, it is
from the perspective of an engineer that he approaches nanotechnology.
As will become clear below, this explains important features of his
vision of the field.
On the other hand, we have Richard Smalley.
University Professor of chemistry, physics, and astronomy at Rice
University, Smalley was awarded the 1996 Nobel Prize in chemistry for
the discovery of fullerenes. His current research is deeply immersed in
nanotechnology, focusing, in particular, on the chemistry, physics, and
potential applications of carbon nanotubes. With his main background in
chemistry and physics, Smalley approaches nanotechnology with an eye
for what can actually be implemented and controlled in the laboratory.
His approach is not only informed by the relevant chemical and physical
theories, but it relies deeply on the actual chemical and physical practices
to determine the feasibility of proposed views.
What is the issue in the debate between
Drexler and Smalley? Briefly put, the question is whether molecular
are possible. As conceived of by Drexler, molecular assemblers are
"devices able to guide chemical reactions by positioning reactive
molecules with atomic precision" (Drexler 2003a, p. 38). More
specifically, the issue is whether it is physically and chemically
possible to construct such assemblers; i.e., whether the
construction of a molecular assembler is compatible with
accepted physical and chemical principles. Drexler claims it is.
In his picture, molecular assemblers are basically mechanical
devices, controlled by computers to "guide the chemical synthesis of
complex structures by mechanically positioning reactive molecules"
(Drexler 2003a, p. 38).
Smalley disputes the viability of this mechanical picture, challenging
the possibility of obtaining the precise control of nanophenomena
presupposed by Drexler. According to Smalley, the required control
cannot be had – not even in principle.
2. The debate
The debate starts with Smalley questioning
Drexler’s proposal with two arguments: the so-called fat fingers
and sticky fingers objections. Smalley’s point is that it is
not possible to pick up and place individual atoms
with the precision required by Drexler: computer-controlled ‘fingers’
will be too fat and too sticky for that (Smalley 2001). The talk of
fingers in this context may seem strange, given that, literally, there
are no at the nanoscale. However, as we will see, this talk plays an
important rhetorical role in Smalley’s argument, which can be seen as a
kind of reductio of the mechanical features of Drexler’s
conception. What Smalley wants to highlight with this language is the
difficulty of actually implementing Drexler’s vision, according
to the standards set by Drexler himself. I will consider each argument
The fat fingers objection takes
seriously the mechanical
nature of Drexler’s conception of molecular assemblers, and attempts to
show that the unfeasibility of the conception is ultimately due to the
mechanical assumptions it requires. As we saw, for Drexler, an
assembler will "mechanically [position] reactive molecules" with
"atomic precision", and in this way, it will be able to "guide
the chemical synthesis of complex structures" (Drexler 2003a, p.
38, italics added). What happens if we take literally the idea of mechanically
locating each atom with atomic precision? This would require,
according to Smalley, nanobots with manipulator arms
– this is the point where the mechanical features are taken at face
value. But given that the fingers of the nanobot arm must themselves be
made of atoms, there would not be enough room at the nanometer scale to
allow the control required to precisely locate each atom. After
all, to have complete control
of the chemistry, too many fingers in too many arms would be needed.
And there is simply not enough room for that. In Smalley’s own words:
Because the fingers of a manipulator arm must
themselves be made out of atoms, they have a certain irreducible size.
There just isn’t enough room in the nanometer-size reaction region to
accommodate all the fingers of all the manipulators necessary to have
complete control of the chemistry. [Smalley 2001, p. 77]
According to the sticky fingers
the precise control over the positioning of atoms required by Drexler
cannot be achieved, given that the atoms of the manipulator arms will
interact with other atoms in unintended ways. Just by
positioning an atom in a given place is not enough to guarantee that it
will interact only with the atoms we want it to
interact with. As Smalley points out:
Manipulator fingers on the hypothetical
self-replicating nanobot are […] too sticky: the atoms of the
manipulator hands will adhere to the atom that is being moved. So it
will often be impossible to release this minuscule building block in
precisely the right spot. [Smally 2001, p. 77]
With these two arguments, Smalley thinks that
Drexler’s mechanical case for molecular assemblers is fundamentally
However, Smalley also raises an additional
worry. In his view, Drexler needs self-replicating molecular
assemblers to implement his vision; otherwise, the rate of production
would be too slow. A single non-replicating assembler would
take a long time to produce only a mole of something:
Imagine a single assembler: working furiously,
hypothetical nanorobot would make many new bonds as it went about its
assigned task, placing perhaps up to a billion new atoms in the desired
structure every second. But as fast as it is, that rate would be
virtually useless in running a nanofactory: generating even a tiny
amount of a product would take a solitary nanobot millions of years.
(Making a mole of something – say, 30 grams, or about one ounce – would
require at least 6 × 1023 bonds, one for each atom. At
the frenzied rate of 109 per second it would take this
nanobot 6 × 1014 seconds – that is, 1013
minutes, which is 6.9 × 109 days, or 19 million
years.) [Smalley 2001, p. 76]
In contrast, self-replicating nanobots
be much more efficient. With the ability to self-reproduce, very
quickly they could create a whole army of assemblers, which in turn
would be able to produce things at a much faster rate.
For fun, suppose that each nanobot consisted
of a billion atoms (109
atoms) in some incredibly elaborate structure. If these nanobots could
be assembled at the full billion-atoms-per-second rate imagined
earlier, it would take only one second for each nanobot to make a copy
of itself. The new nanobot clone would then be ‘turned on’ so that it
could start its own reproduction. After 60 seconds of this furious
cloning, there would be 260 nanobots, which is the
incredibly large number of 1 × 1018,
or a billion billion. This massive army of nanobots would produce 30
grams of a product in 0.6 millisecond, or 50 kilograms per second. Now
we’re talking about something very big indeed! [Smalley 2001, p. 76]
According to Smalley, the implementation of
Drexler’s vision requires more than just molecular assemblers; these
assemblers need to self-replicate as well.
How does Drexler respond? First, with regard
self-replication requirement, even though Drexler himself had an
important role in forming the impression that self-replication was
necessary for the success of nanotechnology (Drexler 1986), things have
changed on this front. Drexler has recently been developing, in
collaboration with Chris Phoenix, models that do not require
self-replication to implement large-scale systems of productive
nanomachinery (see Drexler & Phoenix 2004). The details of these
models, however, still remain to be seen.
Second, with regard to the fat fingers and
sticky fingers objections, Drexler insists, as noted above, that his
assemblers do not manipulate individual atoms. They manipulate reactive
molecules (Drexler 2003a, p. 38). Given that Smalley’s two main
objections were based on the difficulties associated with manipulating individual
atoms, they just miss the target.
In reply to Drexler’s response, Smalley
a second version of the fat and sticky fingers objections, extending to
reactive molecules the arguments that were initially couched in terms
of individual atoms:
The same argument I used to show the
infeasibility of tiny fingers placing one atom at a time applies
also to placing larger, more complex building blocks.
Since each incoming ‘reactive molecule’ building block has multiple
atoms to control during the reaction, even more fingers will be needed
to make sure they do not go astray. Computer-controlled fingers
will be too fat and too sticky to permit the requisite control.
Fingers just can’t do chemistry with the necessary finesse. [Smalley
2003a, p. 39, italics added]
Thus, the original complaint about the
of controlling chemical processes with the needed refinement can be
easily extended to reactive molecules as well. If anything, in
Smalley’s view, the second version of the ‘fingers’ objections is
stronger than the first, given that the precise manipulation of a whole
reactive molecule requires more ‘fingers’ to control the
multitude of atoms involved than what is required by the manipulation
of just a single atom. Thus, the initial difficulty comes back
– now multiplied by each atom involved in the process.
In response, Drexler thoroughly rejects the
talk of fingers. It is not only that this talk cannot be taken
literally; there are simply no such fingers at the nanometer
scale. As he points out:
Like enzymes and ribosomes, proposed
neither have nor need these ‘Smalley fingers’. The task of positioning
reactive molecules simply doesn’t require them. [Drexler 2003a, p. 38]
In a curious way, both Smalley and Drexler
the nonexistence of such ‘fingers’, albeit for very different reasons.
Smalley rejects these ‘fingers’ as part of his reductio of the mechanical
approach to assemblers, which he correctly takes to be Drexler’s view.
Drexler, in turn, denies commitment to these (obviously nonexistent)
objects as part of his attempt to defuse Smalley’s objection.
But once this is clear, we can see the
significance of Smalley’s ‘fingers’ objection: it challenges Drexler to
spell out not the mechanical, but the chemical processes
underlying Drexler’s conception of molecular assemblers. The objection
ultimately disputes the feasibility of controlling the chemical
reactions that would inevitably take place if a mechanical molecular
assembler were ever produced. In this way, by skillfully shifting the
issue from the mechanical to the chemical domain, the objection defies
the viability of Drexler’s proposal.
However, once it is agreed that there are no
fingers at all at the nanoscale, Smalley raises a new
challenge. If the process of placing reactive molecules does not
involve fingers, and if molecular assemblers are to use enzymes and
ribosomes in this process – as Drexler himself acknowledges (Drexler
2003a, p. 38) – further difficulties emerge. After all, we
should now take seriously the need for describing the chemical
processes involved in the implementation of a molecular assembler; in
other words, the chemical details have to be articulated.
In particular, several points need to be spelled out. For example:
How is it that the nanobot picks just the
molecule it needs out of this cell, and how does it know just how to
hold it and make sure it joins with the local region where the assembly
is being done, in just the right fashion? How does the nanobot know
when the enzyme is damaged and needs to be replaced? How does the
nanobot do error detection and error correction? [Smalley 2003a, p. 39]
Without answering questions of this sort, it
unclear how a molecular assembler – with the particular type of control
and precision required by Drexler’s proposal – could actually be
constructed, even in principle.
The outcome of these considerations is what
can be called Smalley’s dilemma.
Supposing that Drexler’s molecular assembler will use something like
enzymes and ribosomes, then either the assembler is a water-based
entity, or it is not. If it is a water-based entity, then it is limited
in what it can achieve; for instance, it cannot produce
anything that is chemically unstable in water. (And how will it then
produce steel, cooper, aluminum, or titanium?) If the assembler is not
water based, then the chemistry that underlies it eludes us. In
Smalley’s own words:
The central problem I see with the nanobot
self-assembler then is primarily chemistry. If the nanobot is
restricted to be a water-based life-form, since this is the only way
its molecular assembly tools will work, then there is a long list of
vulnerabilities and limitations to what it can do. If it is a
non-water-based life-form, then there is a vast area of chemistry that
has eluded us for centuries. [Smalley 2003a, p. 40]
In either case, according to Smalley, there is
trouble. The first horn seems to bring major limitations to what could
be achieved by a water-based assembler (e.g. nothing that is
unstable in water could then be produced). The second horn, with a
non-water-based assembler, requires a chemistry whose details we may
not have completely mastered yet.
Interestingly enough, Drexler’s response to
the dilemma does not address any of the two horns.
Instead, he returns from chemistry to mechanics.
Talking about Feynman’s famous 1959 talk (Feynman 1960), Drexler
Although inspired by biology (where
regularly build more nanomachines despite quantum uncertainty and
thermal motion), Feynman’s vision of nanotechnology is fundamentally
mechanical, not biological. Molecular manufacturing
concepts [that is, Drexler’s own approach] follow this lead. [Drexler
2003b, p. 40, italics added]
With the acknowledgment of Feynman, Drexler
then rejects the need for accommodating the details of chemical
processes that, prima facie,
seem to be required for the implementation of his own vision. By
emphatically putting himself back into a purely mechanical world, he
denies any role for biological or strictly chemical processes in his
Nanofactories contain no enzymes, no
living cells, no swarms of roaming, replicating nanobots.
Instead, they use computers for digitally precise control,
conveyors for parts transport, and positioning
devices of assorted sizes to assemble small parts into larger
parts, building macroscopic products. The smallest devices position
molecular parts to assemble structures through mechanosynthesis
– ‘machine-phase’ chemistry. [Drexler 2003b, p. 41, italics added]
Without a doubt, Drexler emphasizes here the
mechanical features of his conception of assemblers, invoking
conveyors, computers, and positioning devices to assemble structures.
We are now miles away from any chemical understanding of a molecular
assembler. This is perhaps the position Drexler wants to be in.
Presumably, he sees it as a safe place from which to disarm Smalley’s
dilemma, given that the latter does not arise for a nonchemical
conception of assemblers.
This may be so, but the move has its cost
as Smalley does not fail to point out in his final reply, instead of
exploring the chemical details that need to the articulated for
Drexler’s conception to get off the ground, Drexler simply returned to
his mechanical view, bringing back the same difficulties along the way.
For Smalley, a purely mechanical conception of molecular assemblers is
miles away from anything that could actually be implemented – even in
principle – due to the unfeasibility of the required control. With
noticeable disappointment, Smalley notes:
I see you have now walked out of the room
where I had led you to talk about real chemistry, and you are
now back in your mechanical world. […] Much like you can’t make
a boy and a girl fall in love with each other simply by pushing them
together, you cannot make precise chemistry occur as desired
between two molecular objects with simple mechanical motion
along a few degrees of freedom in the assembler-fixed frame of
reference. Chemistry, like love, is more subtle than that. You need to guide
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 2003b, p. 41,
However, with Drexler’s return to the
view, we are back to the main trouble: the level of control over
reactive molecules that is presupposed by this view simply cannot be
obtained. In the passage that follows, Smalley emphasizes just this
Chemistry of the complexity, richness, and
needed to come anywhere close to making a molecular assembler – let
alone a self-replicating assembler – cannot be done simply by mushing
two molecular objects together. You need more control.
There are too many atoms involved to handle in such a clumsy
way. [Smalley 2003b, p. 41, italics added]
However, if a purely mechanical approach to
assemblers does not quite work, what is the alternative? Not
surprisingly perhaps, Smalley’s final conclusion insists on the need
for returning to a chemical conception of assemblers, as a way
to try to obtain, at least in part, some of the required control. As he
To control these atoms you need some sort of
molecular chaperone that can also serve as a catalyst. You need
a fairly large group of other atoms arranged in a complex, articulated,
three-dimensional way to activate the substrate and bring in the
reactant, and massage the two until they react in just the desired
way. You need something very much like an enzyme. [Smalley
2003b, p. 41, italics added]
In other words, to get the control Drexler
needs, it is crucial to appeal to a chemical
understanding of the phenomena: instead of conveyors, computers, and
positioning devices, we have catalysts, reactants, and enzymes. Even
then, it is not entirely obvious that one can fully implement Drexler’s
overall vision. After all, chemical processes are often capricious,
subtle, and delicate – in ways that repeatedly elude us.
3. A partial diagnosis: incommensurability at work?
After reviewing the main features of the
debate, it is hard to resist the temptation of giving at least a
partial diagnosis. Although I do not intend to be comprehensive, I want
to highlight significant features that should help us understand some
of the moves made above.
3.1. Different conceptions of molecular assemblers
First, we clearly have here two radically
different approaches to molecular assemblers. On the one hand,
there is Drexler’s mechanical conception, which is developed as
(conceptual) prototype. It examines, from a mechanical point of view
and purely theoretically, in what way molecular assemblers are
possible, by essentially formulating a theoretical model in which the
relevant physical principles are not violated. The irony is that, as an
engineer, Drexler only provides theoretical artifacts, rather
than physical ones. For Drexler, however, this is not at all a
problem. It is simply part of his theoretical applied science
project, which does not aim at providing experimental results, but
develops instead only a "theoretical analysis demonstrating the possibility
of a class of as-yet unrealizable
devices" (Drexler 1992, p. 489, the first italic is mine). Instead of
producing physical devices, the aim is to generate theoretical results.
In much the same way, the aim of interpreting a physical theory (say,
quantum mechanics) typically is the formulation of theoretical
results regarding the possibility
of certain aspects of the world (on the assumption that the theory in
question is true), rather than the generation of new experimental
results. The activity of interpretation may not be the most typical
activity in scientific practice, but it is a significant part of it
On the other hand, there is Smalley’s chemical
approach to molecular assemblers, which challenges the feasibility
of Drexler’s mechanical conception. As a chemist, Smalley
insists on the production of detectable and controllable
effects, emphasizing the need for accommodating the actual, chemical
details that are part of the phenomena. (This is precisely what Drexler
is unwilling to do.) However, as we saw, Smalley’s challenge goes
deeper, given that it disputes even the feasibility in principle
of actually implementing anything like a mechanical molecular
assembler, due to the difficulty of having the required control.
As a result, and very briefly put, we are
faced here with a disciplinary clash (between chemistry and
engineering), with different conceptions of the nature
of molecular assemblers (chemical versus mechanical), and with distinct
that may lead to their construction (effective implementation versus
conceptual exploration). It is perhaps not surprising that we have
hardly any agreement in the debate!
3.2. Different levels of incommensurability
Given the significant differences between the
two approaches, the picture that emerges is one of incommensurability
(see, e.g., Kuhn 1970, Feyerabend 1981, Siegel 1980,
Hoyningen-Huene 1993, and Sankey 1994). After all, there are no common
to assess the adequacy of each conception. According to the standards
that Drexler set out to himself – namely, to articulate theoretical
artifacts – his approach is perfectly adequate. His criteria of
adequacy require only the mechanical feasibility of molecular
assemblers, in the sense that the phenomena in question are not
incompatible with any known physical (and perhaps chemical) principles
– even though we may not have the slightest idea of how to actually
implement and construct the devices under consideration. For
Drexler, the process of actual construction will come later.
But we also saw that, in response to
challenge, Drexler’s own conception seems to shift, back and forth,
between mechanical and chemical representations of molecular
assemblers. Due to the nature of these shifts, we clearly have here
incommensurability of a conceptual nature. Drexler’s considered
view, however, seems to favor the mechanical conception, which
makes his proposal undoubtedly open to Smalley’s criticisms. Smalley
challenges, in fact, even the feasibility in principle
of such assemblers. Why?
Because Smalley criticizes the core of
Drexler’s approach: the requirement of positioning reactive molecules
with atomic precision. That is, Drexler demands (a) a perfect
control of the position where each reactive molecule will be
placed, and (b) a perfect control of the way
in which a given reactive molecule will interact with other molecules.
Smalley challenges both assumptions. If we were to implement anything
like Drexler’s proposal in the lab, we would face insurmountable
difficulties. Given the huge number of atoms present in the phenomena,
we would not have the precise control to determine in which way a given
reactive molecule would interact (against (b)). Thus, it would not be
possible to position precisely the reactive molecule (against (a)).
Smalley, in turn, adopts a radically
different conception of the nature of molecular assemblers.
With his chemical
conception, assemblers are subject to all the vagaries of chemical
processes. And it is this conception that grounds Smalley’s criticism
of Drexler’s idea of atomic precision. If the chemical
involved in the interactions between reactive molecules are taken into
account, it becomes clear that we cannot simply have the required
control envisaged by the mechanical approach.
Smalley also challenges the methods
Drexler to implement his proposal. The construction of theoretical
artifacts – as the outcome of Drexler’s theoretical applied science –
is not enough to establish the feasibility of molecular assemblers as
Drexler conceives of them. After all, any attempt to actually implement
such assemblers (for example, by trying to construct them in the lab)
will immediately face trouble, given the relatively limited control
that we can actually have over chemical reactions at the nanoscale.
The points just made indicate that there are
least three levels of incommensurability here: cognitive, conceptual,
and methodological (see, e.g., Kuhn 1970, Laudan 1984, and
(i) Cognitive incommensurability emerges when there are no common
standards to assess the adequacy of certain theories about
the phenomena under examination. (ii) Conceptual
incommensurability is the outcome of the lack of common standards to adjudicate
concepts used to describe the phenomena. (iii) And finally, methodological
incommensurability arises from the lack of common standards of assessment
of the reliability of the different methods used. How do these
levels of incommensurability bear on the present discussion?
(i) The debate here involves cognitive
incommensurability in that each side adopts different theories to
articulate the corresponding conception of assembler: mechanical
theories in Drexler’s case and chemical theories in Smalley’s. Each of
these theories is, of course, adequate in its respective domain,
but given the dramatically different ways in which Drexler and Smalley
conceptualize the domains (one mechanically, the other chemically), it
is unclear how one could assess the overall adequacy of the
theories without simply begging the question against the rival proposal.
(ii) The debate also involves conceptual
incommensurability, given the radically different ways in which
molecular assemblers have been conceptualized: Drexler conceives of
them in basically mechanical terms, whereas Smalley is highly sensitive
to the chemical features involved in the phenomena. But how could we
assess the adequacy of such concepts without simply prejudging the
nature of the assemblers themselves? Depending on the view of
assemblers we adopt (a chemical or a mechanical view), we obtain very
different answers regarding the adequacy of the concepts in question.
(iii) Finally, the debate includes methodological
incommensurability as well, given that each view has a different method
of articulation of molecular assemblers. Drexler’s theoretical applied
science approach insists that we should first develop theoretical
artifacts, establishing the theoretical possibility of such assemblers.
Smalley, in turn, with a chemically grounded view, highlights the need
for controllable and detectable results before we could even talk realistically
about the possibility of such objects. Unless we could, in principle,
develop techniques of implementation of molecular assemblers –
identifying the relevant operations to be performed in the lab – it is
hard to judge how such assemblers are technologically possible.
The fact that a device is theoretically possible (that is, its
existence does not violate any laws of physics or chemistry) is not
sufficient to guarantee that we can construct that device, and
hence establish that it is possible in the actual world, given our
technology. Drexler agrees, of course, with the distinction between
theoretical and technological
possibility, and in fact, theoretical applied science often moves ahead
of technology (Drexler 1992). But for Smalley, without accommodating
the practical details of what actually goes on in the lab,
without taking into account the technological aspects of
current chemistry, we cannot claim to have established even the theoretical
possibility of the devices in question. We need more than lack of
inconsistency with physical and chemical principles. The technology
that goes on in the lab is as much part of science as the theories that
are articulated there. Given that the production of a molecular
assembler crucially relies on that technology, we need to consider the
latter as well.
Note that the fact that Drexler and Smalley’s
views are incommensurable does not
entail that they are incomparable. The absence of common standards of
assessment only entails that evaluative judgments cannot be made
without begging some questions, such as assuming the set of standards
of one view to judge the adequacy of the other. Concepts, theories, and
methods can, of course, be compared. We have been doing this all along.
What may not happen is that we will be in a position to decide –
without circularity – the adequacy of these concepts, theories, and
methods, given the lack of a common standard of adequacy.
Why is it significant to identify the various
of incommensurability found in the debate between Drexler and Smalley?
Because this helps to explain in which ways the debate has been
inconclusive, and why it is inevitable to end up with the impression
that Drexler and Smalley are simply talking past each other. With
different conceptions of assemblers and with different methodological
strategies to articulate such assemblers (i.e., strategies that
aim to show the feasibility of such assemblers and to sketch how the
latter could, in principle, be constructed), it is not surprising that
there is no agreement as to how the debate could be settled. Without
common standards of evaluation, or common methods of assessment and
construction of assemblers, it is hard to see how to resolve this
debate without simply begging the question against one side or the
By highlighting the incommensurability
the discussion, we can also understand another feature of the debate:
the many layers in which it takes place. As noted above, we find not
only different conceptions of molecular assemblers (chemical
versus mechanical), different methods
of construction or implementation of such assemblers (actual
implementation versus conceptual exploration), but also, more
generally, different goals for nanotechnology research – given
the different visions
underlying Drexler’s and Smalley’s projects. As we saw, Drexler’s
vision for nanotechnology is one of atomic precision and perfect and
complete control over molecular reactions. It is essentially an engineer’s
vision. Smalley’s vision, in turn, insists on the production of
detectable and controllable phenomena, and takes as a crucial part of
scientific activity the manipulation and stabilization of the
phenomena. This vision challenges the viability of a notion of control
that is not grounded on what can actually be performed in the
lab. It is essentially a chemist’s vision. And, as was pointed
out, in each of these levels, we have incommensurability.
3.3. An alternative way of interpreting the debate:
instruments at work
The considerations just made implicitly
alternative strategy to analyze the debate between Drexler and Smalley.
Perhaps with some adjustments, this alternative could provide a way to
‘settle’ the dispute without (hopefully) begging any questions.
As is well known, Larry Laudan developed a
interesting framework to assess scientific debates: the reticulated
model (Laudan 1984). The idea is that scientific practice is
articulated in terms of three interrelated levels: goals, methods, and
theories. The level of goals involves the aims and values
shared by a particular scientific community. These goals include
certain ways of assessing and structuring scientific research, for
example, searching for and valuing empirically testable and informative
theories over mere conceptual sketches of possible experiments. The
level of methods deals with methods of theory construction and
theory evaluation, as well as the particular experimental strategies
used to implement, control, and stabilize the phenomena. Finally, the
level of theories includes the various theories and theoretical
assumptions adopted by a particular community to explain and predict
According to this picture, scientific change
involves change on at least one of the three levels, but never
changes in all of them at once. Thus, we could use the ‘shared’ level
(say, the level of theories) to assess the adequacy of the remaining
levels (say, goals and methods), and in this way, try to settle the
debate. For instance, suppose that a given community has as one of its
goals to construct a machine that accelerates objects with a speed
faster than that of light. But if the community also accepts a theory
that states that no object could travel faster than light, this would
establish the unfeasibility of the goal. Thus, the community could
invoke that theory to revise the goal.
Of course, this simple model does not cover
the crucial elements of scientific practice. We also have, at least,
the level of scientific instruments (for a fascinating and
sophisticated account, see Baird 2004); and instruments cannot be
identified with any of the three previous levels. (i) Although theories
are often invoked in the construction and manipulation of instruments
(including the interpretation
of the results), instruments are, of course, much more than theories,
and play a significantly different role in scientific practice. For
instance, instruments provide the tools in terms of which experiments
are possible, allowing scientists to probe details of the physical
world that would otherwise be unavailable to them. (ii) Although the
use of instruments require, of course, ingenuity and technique, the
skills demanded go well beyond whatever methodological rules
that may be adopted in scientific practice. Learning such skills
involves special requirements and abilities, such as to be able to
calibrate the instrument and to distinguish artifacts of the instrument
from genuine information it provides. (iii) Finally, the goals
of instrumental practice need not be the same as those of theoretical
practice, given that the former is concerned with details of the
instrumental apparatus that need not be the primary concern of the
latter. Thus, instruments are a crucial additional level of
consideration in scientific practice.
For simplicity’s sake, let us consider
scientific practice as involving certain aims, methods,
theories, and instruments.
Bearing this in mind, we can now return to the Drexler-Smalley debate
and identify the levels in which it has been conducted. As noted above,
there are differences in all of the first three levels. We have
distinct aims: Drexler’s theoretical applied science project is
ultimately concerned with the production of theoretical artifacts,
whereas Smalley insists on the need for the construction of detectable
and controllable phenomena. There are different methods:
Drexler invokes theoretical exploration to establish the
possibility of certain devices, whereas Smalley insists on the actual
implementation of the relevant phenomena in the lab. Finally, there
are different theories: Drexler’s mechanical approach
to molecular assemblers emphasizes the mechanical features of
the phenomena, whereas Smalley insists on the need for accommodating
the relevant chemistry.
Despite the disagreements at these three
levels, the picture changes if we consider the fourth level, that of instruments.
Here, at last, we find agreement
between our authors. Both agree that the use of appropriate microscopy
devices is crucial for the implementation of the phenomena in question,
and necessary for the actual construction of a
assembler (assuming that it can be done). After all, it is through
these instruments that the scientific community has the control it has
over nanoscale phenomena. And it is only in terms of
that the community might be able to build an assembler. After all,
given the size of such assemblers, the mediation of appropriate
instruments is indispensable to control them.
With this minimal agreement, we can now work
way upward, and assess the debate from the point of view of
instruments. Given that instruments are indispensable to the
construction, stabilization, and control of phenomena at the nanoscale
– and both sides of the debate agree on that – a purely theoretical
approach to molecular assemblers that does not take into
account the need for such instruments misses a crucial point
of what needs to be accommodated. And Smalley’s insistence on the need
for the production of controllable and detectable devices can be seen
as an emphasis on just the need for appropriate instruments.
In this way, we see how Smalley is ultimately
justified in making the requirement he makes, without begging the
question against Drexler. After all, both parties share their
commitment to the indispensability of appropriate instruments to
Smalley, however, articulates this commitment further, introducing the
requirement that detectable results should be produced as part of the
determination of the possibility of molecular assemblers. After all,
given that instruments are indispensable for the construction of such
assemblers, to determine whether the latter are possible, it is crucial
to be able, at least in principle, to produce detectable
results. In this way, the overall proposal Smalley advocates seems more
Of course, this does not establish the
adequacy of Smalley’s criticism
of Drexler. This is a separate issue, and is open to the
incommensurability charge discussed above. For, as was noted, the
criticism relies on concepts, methods, and theories that are not
shared by Drexler. However, the emphasis on instruments
indicates one way in which the debate could be decided. After all,
there is a common perspective – the commitment to the indispensability
of instruments – that is shared by both sides, and from which the
overall adequacy of the two proposals can be determined, without
assuming points that are contentious in the debate.
As we saw, the debate between Drexler and
Smalley has many levels and involves a variety of moves. Given the
dramatic differences in concepts, aims, theories, and methods, and the
difficulty of finding common standards of assessment of them, it is
understandable that we are faced with many levels of incommensurability.
However, by exploring the shared commitment
instruments – as the basic source of stable information about the
phenomena under consideration – it is possible to overcome, in part,
the incommensurability and decide the debate. Not in the sense of
conclusively settling the issue, which is not to be had in any case.
But at least in the sense of appreciating what needs to be done to
carry out the visions that underlie each proposal. By identifying the
crucial role that instruments play in the articulation of these
visions, we also see the role these visions can play in shaping
 My thanks go
Adams, Davis Baird, David Berube, R.I.G. Hughes, Loren Knapp, Cathy
Murphy, Alfred Nordmann, Chris Robinson, Joachim Schummer, and Chris
Toumey for extremely helpful discussions. An earlier version of this
paper was presented at a workshop on the Drexler-Smalley debate at the
University of South Carolina’s NanoCenter. I wish to thank all those
who attended for their contributions, and Joachim Schummer for his
encouragement and help. The material is based upon work supported by a
grant from the National Science Foundation, NSF 01-157, NIRT. All
opinions expressed here are mine and do not necessarily reflect those
of the National Science Foundation.
 Of course,
Drexler has actually not constructed a molecular assembler. The
question of the possibility
of constructing such a device would be irrelevant if the device had
already been constructed. It is enough for Drexler’s purpose to
establish the theoretical possibility of such a construction,
sketching how it could be performed in principle.
If no known physical and chemical laws are violated in the
construction, the resulting process is, at least, theoretically
possible – even though we may not have the slightest idea of how to implement
the process and thus actually construct the assembler.
 Note that,
according to Drexler, molecular assemblers will not manipulate
individual atoms, but only reactive molecules. I will return to
this point below.
 Or, at least,
implementing a molecular assembler, presumably we would need to
accommodate the chemical details needed in the theoretical description
of the latter.
Drexler could have
challenged the second horn, noting that there have been studies of
several chemical and biological processes that are not
water-based. But, in this case, it might not be so clear how Drexler
could still maintain the mechanical
nature of his assemblers, given that the relevant work would have to be
done by the appropriate chemical and biological processes.
 The literature
incommensurability is, of course, huge (see, e.g., Kuhn 1970,
Feyerabend 1981, Siegel 1980, Hoyningen-Huene 1993, Sankey 1994, and
the references quoted in these works). But this is not the place to
review it. For the purposes of this paper, I will only focus on the
issues that are significant for the present debate.
 This is a bit
rough. Presumably, Drexler would agree on the relevance of chemical
theories for his overall approach, which goes beyond
his account of molecular assemblers (see Drexler 1992). However, if we
focus only on Drexler’s conception of assemblers,
we get a more ambivalent picture regarding the role of chemistry. As we
saw in his response to Smalley, Drexler shifts back and forth between a
mechanical and a more chemical understanding of assemblers. However,
given that Drexler’s considered view seems to be the mechanical
one, the crucial role is ultimately played by mechanical
 The community
typically also shares Smalley’s commitment to the need for the relevant
instruments as part of chemical practice. It is therefore not
surprising that most members of that community will also accept
Smalley’s critical assessment of Drexler’s proposal. This is
expected, of course, given that the values, methods, and theories of
that community are being assumed. Drexler, however, does not share
them. This is another expression of the incommensurability involved in
Baird, D.: 2004, Thing Knowledge: A
Philosophy of Scientific Instruments, University of California
Drexler, E.: 1986, Engines of Creation:
The Coming Era of Nanotechnology, Anchor Books, New York (expanded
edition with a new afterword, 1990).
Drexler, E.: 1992, Nanosystems: Molecular
Machinery, Manufacturing, and Computation, John Wiley & Sons,
Drexler, E.: 2003a, ‘Open Letter to Richard
Smalley’, Chemical & Engineering News, 81, 38-39.
Drexler, E.: 2003b, ‘Drexler Counters’, Chemical
& Engineering News, 81, 40-41.
Drexler, E., Phoenix, C.: 2004,
Nanotechnology: Feasible, Potentially Safe, and Unnecessary’
(unpublished paper, in preparation).
Feyerabend, P.: 1981, Realism, Rationalism
and Scientific Method (Philosophical Papers, volume 1), Cambridge
University Press, Cambridge.
Feynman, R.: 1960, ‘There’s Plenty of Room at
the Bottom’, Engineering and Science, 23, 22-36.
Hoyningen-Huene, P.: 1993, Reconstructing
Scientific Revolutions: Thomas S. Kuhn’s Philosophy of Science
(trans. by A. Levin), University of Chicago Press, Chicago.
Joy, B.: 2000, ‘Why the Future Doesn’t Need
US’, Wired, 8, 1-11.
Kuhn, T.: 1970, The Structure of
Scientific Revolutions (second edition), University of Chicago
Laudan, L.: 1984, Science and Values: The
Aims of Science and their Role in Scientific Debate, University of
California Press, Berkeley.
Sankey, H.: 1994, The Incommensurability
Thesis, Avebury, Aldershot.
Siegel, H.: 1980, ‘Objectivity, Rationality,
Incommensurability, and More’, British Journal for the Philosophy
of Science, 31, 359-384.
Smalley, R.: 2001, ‘Of Chemistry, Love and
Nanobots’, Scientific American, 285, 76-77.
Smalley, R.: 2003a, ‘Smalley Responds’,
Chemical & Engineering News, 81, 39-40.
Smalley, R.: 2003b, ‘Smalley Concludes’, Chemical
& Engineering News, 81, 41-42.
Department of Philosophy, University of South Carolina, Columbia, SC
29208, USA; email@example.com
Copyright © 2004 by HYLE and Otávio Bueno