Home > Papers > Browse
Ten-Year Assembler Timeline and Weather Forecast
By Chris Phoenix, November 2002
Like most things in nanotech, assemblers are a big topic. Is an assembler even possible? What do we need to do to develop them? Who is working on them? When will we have them? And what will we be able to do with them? That's too many questions for one essay; I'll focus on the technology required to build one, and by extension, the schedule we might see. The point of this essay, as the title implies, is that it's impossible to give a sensible timeline for the development of an assembler. There are too many factors that could cause delays--and many of the factors have nothing to do with technology. But weather will happen, and technology will advance.
First, I should say what an "assembler" is. As commonly used in nanotech, an assembler is a programmable manufacturing system (factory) that builds products, including copies of itself, with every atom or molecule placed according to plan. A factory that can duplicate itself would be an incredibly significant invention--an Industrial Revolution in a box. Although some well-established scientists and publications have asserted that such a thing is impossible, their arguments are not substantive and have been answered extensively elsewhere (for example, here) so I will not waste time debating them. In the remainder of this essay, I will take it for granted that the question is not whether an assembler is possible, but when it will happen.
When will we have the first assembler? The timeline for an assembler depends a lot more on politics than it does on technology. If the wrong people are convinced that assemblers are impossible, the necessary R&D could be delayed by years or even decades. If no one tries to build it, it won't get built until it is relatively trivial to do. Thirty years from now, if no one has done it yet, an assembler might make a good PhD thesis. But if a large government pours money into the project, it might succeed in less than a decade. Whether this will happen depends on whether the government in question recognizes the military and societal importance of an unlimited, portable manufacturing capability. (Of course, existing governments might not want an "industrial revolution in a box" to be available publicly; it would disrupt a lot of existing systems, although it would also alleviate a lot of material scarcity.)
This essay will be outdated within three years. The last three years have seen the Human Genome Project, the gigahertz PC, covalent chemistry done by STM, the National Nanotechnology Initiative, and a vast number of other--equally important--advances and inventions. Not to mention the staggering growth of information and techniques in each of a dozen "enabling technology" fields such as biotech and MEMS. The next three years will see even more change and invention. An assembler project started three years from now would have a rather easier task than one that was started today. We will know a lot more, and there will have been many new breakthroughs that make assemblers easier to build. Most organizations could not justify starting an assembler project today.
Only the government would be willing to start such a project today, and then only for military purposes. Such a project would have a very different timeline than a non-military development pathway. What would a government-sponsored "assembler project" look like? It would probably be secret, due to its military significance. It would be extremely multidisciplinary; it would combine tools and techniques from biology, chemistry, materials science, computer science, and robotics. Not to mention optics and several others I haven't thought of yet. Depending on the degree of urgency, it could range from a small study group to a ten-thousand-person effort to investigate many parallel options for solving each problem. Once the basic theoretical work was done to demonstrate feasibility and establish a road map, the assembler project could look a lot like the Manhattan Project. However, it would be easier to hide than a nuclear program, for several reasons. An assembler project would not require arcane physics knowledge or a massive industrial base. (You can tell when a country is starting a nuclear weapons program because they start building containment structures and shopping around for uranium, and their nuclear physicists stop publishing.) An "assembler project" would not look very different from any other large industrial research effort, and would be quite hard to detect. The enabling technology that requires the most work may be robotics; if a nation's top roboticists stop publishing, beware!
Suppose a government decided that it needed to have assemblers as soon as possible. When would the first assembler be developed? Several years ago, Ralph Merkle designed a rather simple architecture for a basic hydrocarbon assembler (here). Assuming there's nothing fundamentally wrong with his design, then the problem reduces to several steps: 1) Refining his design, or coming up with a better one; 2) Working on solutions for all the sub-problems; 3) Integrating and testing all the best solutions found; 4) Using that first crude assembler to build a more useful one. Steps 1 and 2 require the most creativity; 2 and 3 require the most money. If many solutions for each sub-problem were investigated in parallel, which might require 10,000 or more researchers, then step 2 might require only a few years. (New sub-wavelength optical sensing technologies will make step 3 a lot easier than it was even a year ago.) So, if a government were willing to throw a few billion dollars at the problem, we could probably have an assembler in well under a decade--perhaps as little as five to eight years.
Which governments might do it? There are at least four: the U.S., China, Japan, and Europe. (From what I hear, Russia does not have the ability to launch and sustain a big project.) Each of these has the resources to pull it off. There are three variables: 1) desire; 2) project architecture; 3) degree of secrecy and self-sufficiency. For a multi-billion dollar, multi-year project, funding must be obtained at a high level; although the resources required would probably be far less than for, say, the development of the Stealth bomber, it's still not a trivial commitment. Architecture of the project would be critical. How much effort would be wasted in bureaucracy? How much trial and error would be permitted? How would the project access the scientific expertise it needed? Finally, secrecy and self-sufficiency are important because they will determine whether the rest of the world is surprised by the first assembler, or whether everyone has time to see it coming (and cooperate or compete, either of which would speed it up). Given recent advances in sub-wavelength sensing and nanometer-scale fabrication, I think we're at the point where a Manhattan Project-type approach could produce an assembler with similar resources, time scale, and secrecy as the original Manhattan Project. I suspect, though I have no way of knowing, that it will take another one or two years before the U.S. is even tempted to begin such a project. I have no idea what the perception of assemblers is in other countries, or whether they have the national character required to organize a project along these lines. (We may not even be able to do it ourselves anymore. I suspect China could, if they wanted to--and they may already want to.) Without a large, secret, and effective government/military project, commercial efforts will probably make it happen first.
Now let's assume that an assembler won't happen until some private company can justify it in their R&D budget. Suppose that it won't happen until a commercial effort requiring only $100 million can be reasonably sure of success in less than three years. When are we likely to reach this point? A lot of technology would have to be in place first. The optical technologies I mentioned in the previous paragraph would have to be developed, accepted, and matured. New chemistry would be needed, including a lot more work with chemical reactions done by STM. Robotics would have to mature quite a bit--we've had pick-and-place assembly for a while, and Japan has some factories almost fully automated, but there are a lot of problems yet to be solved. $100 million would not pay for much problem-solving, only integration. I'll pull a number out of the air and guess ten years until enough technologies are in place to make a commercial project feasible, and five more before someone actually makes a convincing business case. Once the first commercial assembler project begins, many other companies will compete; this will ensure rapid development.
Two days ago as I write this, it was announced that polio virus had been made from scratch by a small team of researchers. The ability to make viruses simply by synthesizing DNA on a desktop machine is highly significant; I rate it higher than the cloning of Dolly the sheep. Although military or commercial interests will surely create an assembler at some point, it's also important to ask at what point a small team will be able to do it--at what point the technology will be readily available. Several technology trends indicate that we will have the ability to build nanometer-scale machines, or micrometer-scale complex molecules, within a few decades as a side effect of other technologies. These technologies include large-molecule synthesis (we can already do 120-kilodalton proteins in milligram quantities direct from DNA--see here), dip-pen nanolithography, MEMS, and too many others to list. As the machines we can build get smaller, and the pieces we can build get bigger and more complex, the integration (bootstrapping) problem becomes a lot simpler. It seems safe to say that within thirty years, forty at the most, assemblers will be commonplace--unless the technology is so far ahead at that point that no one bothers with assemblers anymore.
I have a few final points to make. First, are there any current assembler efforts? As far as I know, there are not. Zyvex is the closest, but it is doing basic research and developing intermediate products. It is certainly helping; in addition to its own research, it is publicizing the concept and acting as a focal point for university research. But at this point I think there are too many sub-problems for a single smallish company to solve alone. Second, will the nay-sayers succeed in blocking the assembler for decades to come? Call me optimistic, but I don't think so. The concept of nanotechnology has gone from crazy to futuristic to widely accepted in less than two decades. Assemblers have already appeared many times in mainstream sci-fi such as Star Trek. I expect that the first built-from-scratch bacterium will happen within four years of the built-from-scratch virus; that and other advances will drive home the plausibility of the idea. Finally, other nations that want to compete with the U.S. or to develop themselves more rapidly than industrialization allows may undertake an assembler project; with the possible exception of supercomputers, none of the required technology is protected at all.
In summary, it appears that a manufacturing system based on nanotechnology, capable of producing copies of itself and making a wide and useful range of products, may be produced in less than ten years; likely in fifteen; and almost certainly within thirty years. There are several motivations--military, economic, and political--for developing such a system, and even today there are several nations that could support an early assembler program. The basic required technologies exist today; within ten years they will be mature and usable by a relatively small commercial project. Military, and later commercial, competition will ensure that the assembler is developed rapidly.