Nanotechnology

This debate originated from an email exchange with Kevin Kelley, Wired Senior Editor. I'd written to sound out Wired's interest in a follow-on to my earlier article to develop ideas from my recent book, Superdistribution: Objects as Property. In response to a comment in Kelly's web page (above) I happened to remark "This "loss of control" spells trouble for Drexler's nanotech's dream." Kelly responded "I agree. It makes nano more like life. I've tried to have similar conversations with Drexler, but I would love to have your take on this for Wired IF. Could you state your case against nanotech in 1,000 words or less?. Kelly assigned Rod Simpson, Wired Associate Editor, to organize such a debate. Eric Drexler and Mark Miller declined to participate, at least partially because of Gary Wolfe's The Curse of Xanadu feature article in Wired. So Simpson mailed me a copy of Ed Regis's book, "Nano: The Emerging Science of Nanotechnology" and arranged for him to debate instead. And here we are.

Round 1 of 4
Brad Cox	In the book "Nano: The Emerging Science of Nanotechnology", Ed Regis wrote: "The fact was that if Drexler's assemblers could arrange the right molecules the right ways, then they could build you practically anything. Every imaginable miracle device would be within grasp: self-cleaning carpets for example; "active rugs" whose fibers rippled like cilia, moving dust and dirt off and away; or the gasoline tree, a genetically altered plant that dispensed gasoline instead of sap; or the super-duper home shopping network, in which the blueprints for any desired product were sent by optical fiber to your own personal all-purpose household constructor, whose assemblers then manufactured that object at home for free. Nanotechnology would be the universal building engine, the molecular cornucopia." The Nanotechnology belief-system, as expressed in this excerpt, is a statement of religious faith, not science; of myth, not reality. This is the difference between faith in the emmaculate conception and the scientific logic of gynecology and pediatrics. This is faith that the great feathered serpent would deliver the Aztecs from the Spaniards, not the logic of how wooden swords with obsidian edges might fare in confrontations with horses, cannon and syphilis. This is a faith for those for whom the heaven paved with gold of the gospel hymnals don't quite cut it, not the logic of engineering and science. In particular, it is the faith that the centrally planned design approaches that rose to supremacy in the industrial age will continue right through the information age, a faith that both creationism and evolution will now bow to the nerd. My distrust of nanotechnology arises from two lines of argument. The first is the assumption that atoms really behave like sticky billiard balls when combining to form molecules; that is, that chemicals reactions are governed by Newtonian physics. My initial research focus at Ryerson Labs at the University of Chicago (where Richard Feynman did much of his work) involved just this area, building programs to compute how atoms behave when joined into molecules. Although I've read Drexler's scholarly work on this topic ("Nanosystems: Molecular Machinery, Manufacturing and Computation"), I'll not develop this line of argument here, leaving this specialized argument to those whose quantum physics skills haven't grown as rusty as mine. In this debate I want to challenge the core premise underlying not just the nanotechnology faith, but the core faith of other domains from economics to organizational dynamics. This is the faith that whatever evolution can do, design can do better; that the spontaneous orders emerging from evolutionary interaction of autonomous distributed agents with their environment can be improved on by that centrally-planned activity the engineering community calls design.
Ed Regis	Just received your post, Brad. Incredible, by any standard. You quote a paragraph of mine and then use the word "faith" nine (9) separate times (count 'em!) by way of characterizing your attitude toward nanotechnology. (You also use the word "myth" once, apparently to avoid too much repetition. Thank God for small favors!) But despite having used the word "faith" nine (9) times, you don't give us even so much as one reason for actually believing that nanotechnology "is a statement of religious faith." Not one! Then you claim that you'll have two lines of argument. One, you don't like the notion that atoms behave like sticky billiard balls. But after boasting of your own vast credentials in this area, you immediately proceed to confess that you won't develop "this specialized argument" because your quantum physics is rusty. Well, what can we do but take you at your word? Your next, and final, point is that you want to challenge the "core premise" that whatever evolution can do, design can do better. Unfortunately, you don't develop that line of argument, either. All you do is to make that bald claim, a statement of your future intention. In fact, you don't present any argument whatsoever! So for crying out loud what the hell is there for me to argue against? You remind me of the pregnant lady who constantly claims that she's ready to give birth, that the hour is near, that the labor pains are coming fast and furious, that her water is about to break (don't blame me, you brought up the gynecology stuff!) ... but who never delivers an actual baby. When does the real "debate" begin? I can't wait!
Round 2 of 4
Brad Cox	Ed, science historians (Thomas Kuhn; The Structure of Scientific Revolutions) argue that faith and paradigm mean about the same thing. Since you object to the word faith, I'll phrase my argument as a conflict between paradigms of mankind's place in its environment. Nanotechnologists advocate the industrial age view that mankind can impose order exogenousously, by an external designer. I claim, following Occam's Razor, that it is more economical to view order as primarily emerging endogenously and spontaneously within evolutionary systems, which puts design in a secondary role. I can't make this clear if the domain can range from self-cleaning carpets to pollution cleanup devices. I'll contrast the two approaches with a specific problem; namely a cure for the common cold. Rhinoviruses, AIDS and chestnut blight fungi are hard to control because they quickly evolve in response to changes in their environment. In effect, each of millions of infectious agents continually exploits their local knowledge of scarcity, abundance and opportunity, passing each competitive advantage to its progeny. Traditional science such as biomedicine uses model systems, such as tissue cultures, to gain formal statistical knowledge of large populations. This knowledge is not the same as the tacit, local or informal knowledge each agent has of its particular circumstances. A nanotech designer (or central planner) can process formal knowledge effectively However a central planner has limited access to local tacit knowledge. This is more available within distributed evolutionary systems, be they trillions of rhinoviruses invading a respiratory system or millions of self-interested entrepreneurs in an economy. The problem for nanotechnology is that formal knowledge is not sufficient to bring a nanomechanical manipulator arm to bear. Exactly which atom on which molecule should be manipulated? Exactly which other atoms are in the way, given that every atom and molecule are in rapid thermal motion. Formal knowledge isn't sufficient for the same reason that knowledge of maps and average traffic conditions can't get us to work safely. For that we need the tacit knowledge of particular conditions that we gain by keeping our eyes open as we drive. Nanotech advocates regularly note that nanomechanical motors do exist in each paramecium's flagella. However this demonstrates what evolutionary distributed processes can accomplish. Calling them "motors" in no way supports the claim that centrally planned design could do likewise. Those familiar with Friedrich Hayek's work will recognize here the same argument that predicted the collapse of Russia's centrally planned economy. And its why Michael Rothchild's Bionomics Institute has been trying to displace the design-centered industrial age metaphors the press invariably uses in discussing evolutionary systems (for example "pumping up" or "cooling down" an economy, "reengineering" a corporation, and so forth).
Ed Regis	You portray nanotechnologists as advocating the "view that mankind can impose order exogenously, by an external designer." But as an emerging technology, nanotechnology does not "advocate" any specific view regarding the origin of order. Nanotechnology is an attempt to build macroscopic objects by manipulating atoms mechanically, placing them where they need to go in order to create the desired object. It, along with any other technology, obviously *presupposes* that human designers "can" impose order upon nature. That presupposition, I take it, is uncontroversial and trivially true, as is illustrated by the existence of New York City where a primeval forest once stood, plus centrally-planned and intentionally designed human artifacts like highways, the Hoover dam, and the Panama canal. What you're really arguing, then, is not that externally imposed design is impossible but that it's somehow inferior to bottom-up, locally-based emergent systems. You cite rhinoviruses and H.I.V. as phenomena that, because they're responsive to local conditions, are inherently superior to the structures created by the top-down, "centrally-planned activity the engineering community calls design." All this is fine as far as it goes. Still, top-down central planning and externally-imposed designs often beat spontaneously-evolved agents all to hell. Between 1966 and 1977, doctors working under the central planning of the World Health Organization, and by using the spray injectors and bifurcated needles that had been created by "that centrally-planned activity the engineering community calls design," eradicated the smallpox virus from planet earth. A disease that at one time was the biggest killer in history, and which had maimed and blinded millions of people, was completely removed from nature. The fact that those viruses were "autonomous distributed agents" which had evolved in response to local conditions using tacit knowledge and all the rest, did not save them. Without exception, every last one of those clever and shrewd virus particles is dead and gone. I think the only credible conclusion to be drawn from this is that sometimes design and planning is superior and sometimes it isn't. In any case, nanotechnology need not be the "centrally-planned activity" you portray it to be. If and when it comes into being, it will have resulted from a bunch of parallel autonomous distributed agents called human beings using their tacit and formal knowledge, and working in distributed labs on distributed computers and other distributed lab machinery. Finally, the fact that atoms are in random thermal motion does not prevent the existence or accurate replication of DNA molecules, either in living cells or in commercial DNA synthesizers. It need not impede nanotechnology. "Exactly which atom on which molecule should be manipulated?" That depends on what you want to build.
Brad Cox	Ed, we're not communicating. Eradication of smallpox was the accomplishment of the biomedical community using the traditional approach of traditional medicine; exploiting the evolutionary nature of distributed systems. This was not done by a nanotechnologist designing an anti-smallpox agent with a CAD package. And replication of DNA in evolved systems proves nothing about design-centric nanotech approaches. This is like arguing that the success of distributed western market economies proves that a centrally planned economy would work in Russia. Finally, "That depends on what you want to build" evades the question. I offered a specific example of something we might want to build; a cure for the common cold. How is your central designer going to get the detailed local tacit knowledge necessary to apply a micromanipulator arm to that? Again, I emphasize that formal knowledge of average behavior is not sufficient, any more than knowledge of maps and average driving conditions are sufficient for safe commuting. But curing the common cold is clearly a hard problem so let me propose an easier one. Consider how design and evolution relate to aircraft manufacturing. How many wings should a plane have? Should they be above the fuselage or below? What should the tail section look like? How many passengers should it carry? How far should it fly? What should planes be made of? What color should they be painted? Then contrast how the Wright Brothers answered these questions to how we'd answer them today. Clearly design played a role. But contrast this with the role that evolution played as competing designs struggled against each other in the last half century of commerce and wars. The language we use to speak of knowledge invites us to believe that the 747 was produced by the formal knowledge of aircraft designers. However the designers were merely part of the conduit through which distributed tacit knowledge was expressed as diverse aircraft designs competed in a distributed aviation market . So my question is still unanswered. Explain how your nanotech central designer will decide even such "trivial" macroscopic questions as the size, shape and color of my common cold cure pill or airplane examples. Then we might move along to harder nanotech issues such as exactly which atom on which molecule should be manipulated while evading other atoms that may be in the way. In other words, how is a nanotech central designer going to acquire the massive tacit knowledge to perform in a distributed market in the trivial macroscopic case. Then we might move on to much harder cases such as understanding, designing, and manipulating atoms in thermal motion. How can a central nanotech designer do this more effectively than the evolutionary distributed solutions that mankind has been using for millenia?
Ed Regis	The eradication of smallpox was not the achievement of a nanotechnologist, damn right. But since I didn't say it was, why do you bring this up? Did I miss something? You're also right that DNA replication in living cells proves nothing about a design-centric approach. But I didn't say that it did. I mentioned DNA in answer to your doubts regarding the creation of stable molecular structures in the face of random thermal vibration. Did you get a phone call at that point? You seem to be having a whale of a time following your own line of argument, Brad - but who can blame you? I mean, anyone who imagines that Hayek's predictions about Russia have anything to do with whether nanotechnology is scientifically possible is not someone straitjacketed by traditional western notions like "relevance" and "logical connection." As for your seven (7) questions about airplanes, look, I'm really not an aeronautical engineer, and I did not sign up for a debate about aircraft design. How those questions prove that nanotechnology is impossible is way over my head, and so I have to admit defeat here. You win! Still, in all honesty I feel compelled to inform you that you missed the three most important questions of all: (1) What kind of rivets should we use, flat rivets or pop rivets? (2) What time do we land? (3) If God had meant us to fly, why didn't he give us tickets? No one in their right mind is going to deny that aeronautical engineers learn from experience, that they get feedback from their successes and failures, that they adjust their designs in light of performance in a variety of empirical conditions. All this is entirely obvious trite platitude. That's just how human epistemology works: we use reason and we use sense experience and we revise our preconceptions in the face of actual practice. Did anyone ever deny this? You ask: "Explain how your nanotech central designer will decide [certain issues]." Now, Brad, this really won't do at all, even if you did get a phone call. Sometimes you just have to pay attention, like it or not. In my last post I wrote: "If and when [nanotechnology] comes into being, it will have resulted from a bunch of parallel autonomous distributed agents called human beings using their tacit and formal knowledge, and working in distributed labs on distributed computers and other distributed lab machinery." Does that sound like I believe that some Big Master Nanotechnologist in the Sky is going to solve all of our problems as if by magic? You claimed in your first post that nanotechnology is "a statement of religious faith." So far, you haven't made good on that claim. Care to try?
Brad Cox	Ed, you brought up smallpox, not I. I answered that the old-fashioned approach we call biomedicine did this, not the revolutionary new "science" of nanotech. Nobody disputes that biomedicine exists. The dispute is whether the difference is sufficiently compelling to justify a new name for biomedicine and claiming this is a paradigm shift. Apart from some wooly-minded extropian star-gazing (uploading, indeed!), Merkel's design of atomic-scale planetary gears is the main example of serious nanotech design I've seen. While I admire his work, it triggers the same ambivalence old-style AI once did; of extraordinary technical proficiency heading in the wrong direction. When the AI bubble finally burst its advocates got on a path that could lead to real progress by proclaiming genetic algorithms and agent-based simulations were "a branch of AI". A big part of Nanotech seems to be repeating this same old shell game. The part that is genuinely new is the part that claims direct centrally-planned design of atomic-scale structures. I've argued here that centrally-planned design has not been as capable as many as hoped at several levels of granularity. These include economy-sized structures in Russia, company-sized structures (corporate "reengineering"), plane-sized structures in the airline market, and rhinovirus-sized structures. I've left the prognosis for atomic-sized structures for readers to infer on their own since the limitations to knowledge are similar at each level. To focus on nanotech in particular, the smallest nanotech manipulator arm ends with a single atom and is thus as big as the object it's probing. But both the target atom, and other atoms in the way, are rapidly moving. This is the same limitation to knowledge that a blind man would experience if the only way of aiming the next shot is to first shoot a queue ball to sense where the other balls are positioned. Except that with nanotech the limitations of knowledge are far worse because the balls are in motion around a tether whose attachment point is also unknown. Except for "designed" versus "evolved" and "endogenous" versus "exogenous", I'm out of words for explaining the difference between Merkel's Nanotech bearings and the bearing of a paramecium's flagellum. The best I can do here is to point out once more that these are the product of opposite (designed versus evolved) approaches and let the readers judge which one to bet on. Enough of this. Ed, go ahead and do it. Until you do, I hope you'll forgive me for not yet acclaiming the emergence of a radically new science of nanotechnology.
Ed Regis	In your first post, you claimed to have read Eric Drexler's book "Nanosystems: Molecular Machinery, Manufacturing and Computation." Unfortunately, you don't seem to have understood a word of it. Chapter 5, entitled "Positional Uncertainty," is wholly devoted to the problem of manipulating molecules that are in thermal motion. That problem is not unsolvable. For one thing, your vision of atoms flying around like popcorn applies only to fluids; in solids atoms merely jiggle in place. Typically, the magnitude of the jiggling is small with respect to the molecular operations that need to be performed. In any case, you could dampen excess jiggling by lowering the temperature of the environment, or by attaching the molecule to a larger, stiffer structure, or by confining it within an enclosure, effectively holding it in a vise. And, of course, you could do all three things simultaneously. Thermal vibration, therefore, presents no barrier to the controlled molecular transformations that are required for nanotechnology. Your other alleged show-stopper is the problem of deciding "which atom on which molecule should be manipulated while evading other atoms that may be in the way." But there needn't be any other atoms in the way. You could get rid of unwanted atoms by evacuating the chamber and then introducing back into it only those atoms needed for the desired reaction. Indeed, this was exactly the technique used by physicist Hans Dehmelt to confine individual atoms in particle traps. Further, had you read Drexler's chapter 13, entitled "Molecular Sorting, Processing, and Assembly," you would have learned how molecular mills, sorters, and other mechanisms can physically eliminate unwanted atoms and create ordered streams of feedstock molecules that can be presented to the manipulator arm one by one. Your real quarrel, however, is not with nanotechnology as described by Drexler, Merkle, me, or by any other human being, living or dead. Your real quarrel is with the "nanotech central designer" you've repeatedly conjured up for us. To call this concoction a straw man is an insult to straw. Straw is a real thing with known uses; the "nanotech central designer," by contrast, is a private phantom of your own creation. You have an avowed fondness for evolution, which you imagine is somehow "better" than conscious design. But why does it matter? Do hummingbirds fly "better" than the Concorde does? The Concorde gets from New York to Paris in 3 hours and 45 minutes, which is what matters to its paying passengers. Will nanotechnology be "better" than the products of evolution? Who cares? If it can build useful objects out of cheap raw materials at slight cost, if its cell-repair machines can float through your bloodstream keeping you healthy, young, wise, and good-looking, why should it have to do anything more?