Question 1: Tell us about the recent activities of the Center For Responsible Nanotechnology. Are you pleased with the progress of CRN?

Yes, I'm very pleased. We continue to get invitations to speak at conferences and on panels. I recently gave a presentation to the National Academies' review of the NNI. And we have developed a product, "Nano-Workshops," which will help organizations decide how to respond to the development curve of nanotechnology. We're currently signing up organizations for the first few workshops at reduced cost.

Question 2: You and Eric Drexler recently published a document called "safe exponential manufacturing", in which you argue that molecular assemblers are no longer necessary for molecular manufacturing. Did you abandon the molecular assembler paradigm because it was unfeasible, or because it evoked fears of runaway replication?

The idea of assemblers has been fictionalized and extended to mean a robot that can rearrange any molecule and turn anything into anything else, and might turn into gray goo if they got loose. Such a universal machine was never proposed, and would be extremely difficult if not impossible to build.

Molecular assemblers, in the sense of small, externally controlled, self-contained construction robots using presupplied feedstock, were proposed in Engines of Creation in 1986. Machines of this sort are not impossible, but would be difficult to design and use.

As far back as Drexler's book Nanosystems in 1992, it was realized that free-floating construction robots would be less efficient than fastening those robots down into a factory framework. That was not done to alleviate fear, but simply because it's a better design. But since the idea of gray goo has continued to plague molecular manufacturing, we thought it was worth explaining that the nanofactory approach was even farther from gray goo than the molecular assembler idea.

Question 3: Tell us about the nanofactory. What advantages does it have over the earlier molecular assembler concept, other than to minimize the "grey goo" risk.

Independent molecular assemblers would have to contain a lot of equipment in a small package, which would make them needlessly complex and inefficient. Their activities would have to be coordinated over shifting and unreliable communication links. Their product would have to be first extruded and then combined, which poses several difficult engineering challenges.

In a nanofactory the functional pieces of an assembler would be integrated into a structural framework. That makes communication and planning a lot easier, because you know exactly where everything is in relation to everything else. It also makes design more efficient. For example, you could share one computer among a lot of construction systems, which is good because a computer could be quite large relative to molecular machines. Pieces of a large product could be built and stuck together while the whole thing was contained within the protected nanofactory environment. So there are a lot of advantages.

Gray goo would require many mechanisms and functions that a molecular assembler would not have, so I'm not sure there was actually any risk to be minimized. But a nanofactory, with no free-floating robots at all, should eliminate any remaining worries.

Question 4: Dr. Eric Drexler has shown an animation of a desktop nanofactory, and this animation shows tiny machines making reactions happen between individual molecules. But does this animation really show how mechanosynthesis would occur?

Yes, the animation shows a way that mechanosynthesis could occur. The reactions in the most recent version have been tested in simulation, not just made-up. So when it shows a molecule being held on the tip of a molecular pyramid, and then another pyramid swings in with a structure at the tip that removes some of the atoms from the bound molecule, they've done a simulation that shows that this is what would actually happen if those molecules were built and maneuvered that way.

Of course, there are different kinds of mechanosynthesis. It might even be done in water. And there's a huge number of different possible reactions. And there are several different kinds of mechanisms that could be used. So this particular design is not a proposal for the best way to build the factory. But it is not just a cartoon or an artist's conception; as far as we know, a system built along these lines could actually work as shown.

Question 5: You seem to be increasingly sanguine about the timetable for the development of a working nanofactory. While most others emphasize the enormous technical difficulties of developing molecular manufacturing systems, you present the problems as fully tractable. Why are you more confident than your colleagues?

"Enormous technical difficulties" and "fully tractable" are not mutually exclusive. There will definitely be enormous difficulties. What I emphasize is that the difficulties are worth overcoming. So far, most people working in most fields of nanotechnology have not taken the time to understand the power of the molecular manufacturing approach. And they don't see how the various benefits work together to make it somewhat easier than they expect.

If you read science fiction from the 1950's, even the most technical authors like Heinlein use analog computers, and the computer's circuits are still big enough to repair by hand. You have atomic rockets with computers controlled by complex 3D shapes, cams, that the pilots load by hand. I just read a story in "World's Best Science Fiction 1970," in which a desktop publishing machine in the year 2000 used "various folding black plastic hoods." These authors simply didn't realize what could be done with digital computers. Molecular manufacturing will be the same kind of breakthrough, and for similar technical reasons.

I don't think it'll be easy. But I think that someone who understood the theory, and had access to a few billion dollars, would be able to launch a very credible effort starting today. Ten years from now, when the theory is more widespread, and the enabling technologies are better developed, it will take a lot less money.

Question 6: Many mainstream researchers still claim that desktop nanofactories are infeasible if not impossible. How much success have you had in convincing skeptics of the feasibility of molecular manufacturing?

There are different reasons for skepticism. Some are skeptical simply because it doesn't make sense to them; they've never understood how the parts of the theory fit together. Others can't see beyond the practical difficulties. They see what we can do today with a particular technology, and they don't see how that technology can be extended far enough, so they write off the whole field. There are some political and business reasons for skepticism. Basically, some of the skeptics want to avoid public fear of gray goo, and they choose to attack the whole field and concept rather than study and educate on the difference between molecular manufacturing and free-floating goo-bots.

I have had some success with skeptics. For example, William Illsey Atkinson wrote a book called Nanocosm that was viciously scornful of molecular manufacturing. After a few email exchanges, he wrote, "I grind my teeth to say it, but you have managed to do something I'd thought was impossible: make me reconsider my position that the nanoassembler cannot be achieved in principle." http://www.nanotech-now.com/Atkinson-Phoenix-Nanotech-Debate.htm

In talking with scientists, I have had mixed success, depending on whether they wanted to discuss or just to argue. Sometimes we have had to agree to disagree. For example, I might say, "There are several known examples of X, and theory says there should be a lot more, so it's reasonable to include X in our plans." And they might say, "But the theory isn't complete yet, and you can't count on anything until it's actually demonstrated in the lab." Well, we know that the basic concept is workable, so I think it's wrong to insist that every detail be demonstrated before we start to talk about the implications. If we do that, we will have no ability to plan ahead, because we won't be able to start talking until the first nanofactory is built and running.

The people who are most skeptical of molecular manufacturing are the ones who have spent the least time studying it. Some of the most famous skeptics have written whole articles objecting to ideas that were never proposed. When skeptical scientists actually study molecular manufacturing, they are likely to say, "Yes, I admit that might work eventually, but I think it won't be the best approach." Well, time will tell. But no one uses analog computers anymore.

Question 7: A simple version of the desktop nanofactory could extrude sheets of a diamondoid material. Such a machine would be quite valuable, given the impressive strength/weight ratio of diamondoid.. Since this desktop nanofactory would only be making one material, it would be much easier to design than a desktop machine capable of producing myriad components. Moreover, it would be more difficult for the mainstream scientific community to argue that a relatively simple nanofactory was impossible. Have you given thought to proposing a simpler nanofactory that could only produce diamondoid sheets, as a precurser to more advanced desktop nanofactories?

The trouble is that the only known way to build a large diamond-building nanofactory is with another nanofactory. Any nanofactory that could build nanofactories as well as sheets of diamond would be able to build a lot of other products as well. So this is not the way to sneak up on the problem.

There is more than one possible approach. One is to develop, probably with great effort, the capability to build one tiny programmable diamondoid-building nanofactory. Then use that to make bigger and bigger nanofactories. It might be hard to get funding for this approach.

Another approach is to build a nanofactory out of weaker materials that can build weak materials. It might even be based on biopolymers, built and assembled underwater. Instead of construction materials, use it to build computer circuits, which don't have to be strong and which are extremely valuable. It might also be useful for building sensors and medicines. This would prove the concept of a programmable molecular fabrication system that can build more fabrication systems. And a nanofactory using relatively weak molecules could then be used to build more advanced designs out of better materials.

Question 8: How much funding and time would be needed to ascertain if molecular manufacturing is truly feasible? How long would a feasibility study take? Have any detailed feasibility study proposals been written?

The right question to ask is not whether it's feasible yes-or-no, but what a useful manufacturing system would look like and what it would take to build it. The real issue is whether it's better to access the nanoscale with large tools doing special-purpose finicky operations, or whether it's better to invest in building general-purpose programmable manufacturing tools at the nanoscale. It's clear that we can do that; the question is whether it's worth doing. Preliminary theory says yes, it's very much worth doing. The study would flesh out the details, developing and evaluating a concrete proposal. Molecular manufacturing tools could take a lot of different forms. So there are a lot of different studies that could be done.

The first step would be to design a system of molecular types that could implement actuators, computers, sensors, and reliable mechanically guided molecular synthesis. Drexler's book Nanosystems made a good start on that for the molecular type of 3D carbon lattice (diamondoid). Nanosystems also provides a solid theory basis for nanoscale mechanical engineering. It took one very smart person a few years to write that book. I don't think it would take very long for a group of biochemists to do a similar study of biomolecules.

The second step would be to figure out what kinds of products could be made with a manufacturing system based on the chosen molecules, and what the performance of the system would be. For diamondoid, we know that even a primitive nanofactory design could make a wide range of extremely valuable products, producing more than its own mass each day. I would expect that any nanofactory design would be able to build computer circuits, and that could make it worth billions of dollars per year.

The third step would be to design a roadmap to build the first nanofactory. You only need to build one tiny nanofactory because you can use that to build the rest. But building the first one is far from trivial, and the development cost and time could be large if you have to develop a new kind of chemistry or make tools to make tools to make tools. The fourth step is to look at the economics and uncertainties, to decide whether it's worth investing.

I would think all of this could be done, for several different representative molecular systems, by ten to fifteen well-chosen people in six months to a year. A preliminary estimate might be done by three people in three months. I suspect at least a few governments and corporations have already done this. But I would expect the results to be kept private.

We really need a well-funded public study along these lines. Because the potential risks and benefits are both so great, it's urgent that we gain a better understanding of this technology.

Question 9: If a desktop nanofactory could be built within the next five years, it would seem that angel investors would be willing to invest substantial amounts of money into such a venture. Why aren't angel investors investing more money into the desktop nanofactory paradigm?

I think developing a diamondoid nanofactory in a five-year time frame is quite unlikely, and would probably take billions of dollars to have a reasonable chance of success. This would be a "Nanhattan Project" scenario. Angel investors would not do this.

A biopolymer nanofactory might be significantly easier to develop. This has not been studied as much. I suspect a biopolymer nanofactory would be worth doing, but I don't know for sure. Until recently molecular manufacturing has commonly been equated with diamondoid, so other molecular systems have not been looked at. There may be opportunities there.

Question 10: Do you think that an official debate between the proponents and the critics of molecular manufacturing will ever happen? So far, the only real debate has occurred between Drexler and Smalley.

I'm not sure I'd call that a real debate. Smalley's argument showed that he was not familiar with some significant twenty-year-old work in enzyme chemistry, so one of his foundational points--that enzymes work only in water--is clearly invalid. Some of his other arguments were not addressed to actual molecular manufacturing proposals. Drexler, for his part, chose to reiterate the basic theory of molecular manufacturing rather than address Smalley's arguments in detail.

As I said above, molecular manufacturing has until recently been equated with diamondoid. One of my goals has been to point out that molecular manufacturing is broader than that. (This is also one of Drexler's goals; he's always favored starting with a biopolymer approach. But his association with diamondoid has made it hard for him to get that point across.) This is the kind of understanding that will reduce the controversy.

I don't expect there ever will be a decisive debate. Instead, I expect that there will be a gradual meeting of the minds, helped by lots of smaller technical discussions. As molecular manufacturing theory develops, and experiments confirm more and more of the basic approach, it will be realized that there's nothing impossible or even radical in the technical requirements. Eventually, there will be nothing to argue about. The reaction will follow the standard pattern, from "That's ridiculous" to "that's trivial" to "I always said so." The shift will be more political than technical.