Slope of technology curve
One of the premises in the current MNT (molecular nanotechnology) debate is there may be â€œrapid development from a basic fabricator to a flood of advanced products,â€ taking a world with no noticeable MNT presence by storm. This assumes the basic fabricator will be created in its mature state, capable of replicating itself at high speed. Given a mature fabricator with no lower-performance ancestors the world economy would be transformed in a matter of months as MNT products displace many existing ones, and social chaos would ensue as millions lose their jobs in the face of MNT competition.
But history has few examples of new technologies bursting out like Athena from the forehead of Zeus. Most have false starts, unrelated applications of partial successes, and then low-performing complete examples that are improved over years and decades. Molecular manufacturing probably will follow this pattern. Over the coming decades we will see gradual improvements in the complexity of products that can be built and the speed of production.
A different argument can be made: that although there will probably be a continuous improvement from low-performing to high-performing manufacturing, a very small (quick) region of that curve will have a lot of impact.
The graphic below shows five stages of product complexity:
â€¢ shaped crystal â€“ diamondoid shapes made for use in other products
â€¢ simple product â€“ devices such as a op-amp chip or diamondoid block and tackle
â€¢ complex product â€“ a central processing unit, electronic or using mechanical rods
â€¢ reprogrammable â€“ rather than having molecular manufacturing devices designed to make a specific product, we now have general purpose ones that can make a variety of complex products
â€¢ autoproductive â€“ nanofactories that can build duplicates of themselves
The production rate for each type of technology will improve over time, from a rough prototype producing only one gram of material per day to single-product-optimized factories producing one gram per second. A new level of MNT complexity will have to concentrate on niche markets until its production rate is improved to where it can compete against the high production rates of its simpler predecessors. For self-replicating technology the early generations will serve more as proof of concept than as transformative devices. By the time they can replicate themselves enough times to have an impact a descendent with ten times the production rate will have been developed that will zoom past them. http://www.kelthaven.org/images/mmprogress.gif
In the early stages, before nanofactories become useful as general-purpose commodity manufacturing instruments, their products will be very expensive, and they will be able to produce only very small volumes of whatever they build. At this phase, they will not have much impact.
In the late stages, when a nanofactory can build another nanofactory in a day or so, it seems certain that the impact will be immense. The question, then, is how long the time period will be between low-volume (expensive) production and high-volume (cheap) production. When phrased this way, the division appears dichotomous--that is, sudden.
If molecular manufacturing cannot be used to build nanofactories, but only simpler things like CPUs, there will be no revolution yet. But we will be gathering skill in design. Once the first autoproductive nanofactory is built, the only question is how quickly its design will advance from 1) fabricating its mass in more than a month to 2) fabricating its mass in less than two days.
Since the design of nanofactories will be digital and engineerable (like software), it appears to some researchers (e.g. Chris Phoenix, CRN) that a mere performance improvement will not take long at all--especially since there will by that point be clear recognition of the economic benefits to be gained.
It may seem that MNT advocates ignore the possibility of manufacturing devices simpler than an autoreplicating assembler but which may have a major impact on the economy. High-performance MNT products will take the place of components that form the bottleneck in current systems. This will shift market share away from old industries and enable new ones to form. As the technology advances new products that could only be made through MNT will be developed and more new industries will form around them.
Although the impact of such products appears large by today's standards, MNT advocates recognize that such impact could be dwarfed by full exponential manufacturing. The impact of pre-nanofactory MNT, as large as it is, appears to be absorbable by society. However, the impact of full-on autoproductive (exponential) manufacturing may be somewhat more disruptive.
MNT products will follow the usual expansion curve as productivity improves. The initial low-production rate version will concentrate on high-value items for wealthy customers. As the technology is improved the products will spread into more market niches. As it reaches maturity it will start cranking out commodity products, displacing existing producers. So the â€œshaped crystalâ€ technology will begin by supplying structural components for microdevices, move up to providing reinforcing fibers for aircraft, and then a couple of decades later a contractor will show up at a construction site holding the entire project's supply of diamondoid trusses under his arm. The nanofactory paper already discusses how this may happen for assemblers on p. 58. The production rate can change by two orders of magnitude depending on the maturity of the control system. Initial assembler designs will be limited by their control and material supply technologies, even if the manipulator arms could handle atoms a thousand times faster.
If there is slow progress in improving production rates, it will be frustrating to nano-designers but arguably a benefit to the rest of society. A prototype assembler that needs two months to duplicate itself would be useless for commercial production but would emphatically settle remaining disputes over how practical the technology is. That will be the spur to settle policy issues before the wave of exponential production arrives--though some (e.g. Chris Phoenix, CRN) feel that this will be far too late to make wise policy.
The policies for handling exponential production MNT will be an outgrowth of preceding policies, including the ones put in place to handle the predecessor MNT technology. MNT will provide devices with unprecedented capability for spying, healing, computing, building, and more even before full MNT is developed. This will require new policies to handle them. If the early nanotech policies are not wisely chosen they may conflict with exponential production--and unless they are designed explicitly with exponential production in mind, they will probably be unable to handle it at all. Near-term policy discussions should provide a basis for good policies to regulate the era of linear MNT while recognizing that exponential MNT policies will have to deal with qualitatively different problems.