A publication of the Foresight Institute
Nanotechnology-based manufacturing techniques should yield great increases in productivity and wealth. Improvements in two techniques in particular will greatly decrease resource requirements: the incorporation of voids, and wearproofing.
Whenever possible, objects can incorporate carefully shaped
voids to save cost and mass. Generally, voids are more useful for
large systems or those under low loads. They can range in size
from arbitrarily large down to a fraction of a nanometer wide;
the upper limit is set by device size, the lower by the scale of
atoms. For structures under light compressive loads, voids formed
in fractal patterns can yield maximum efficiency.
Today's bulk manufacturing can produce large, irregular voids at reasonable cost, as in foam rubber and insulation. Nanomachines should be able to produce uniform voids down to one atom across, thereby cutting the mass, cost, energy, and time needed for production. The biggest gains will be for objects with structural loads in pure compression or mixed compression and tension; fortunately this includes the majority of objects we use, such as furniture, doors, most walls, and appliances. The void fraction of these could be very high, perhaps 99% or more. Highly loaded objects (e.g., engine parts) will benefit less, and highly loaded tension systems (e.g., cables and pressure vessels) will benefit little.
Incorporating voids, combined with scavenging heavy pre-nanotechnology parts, will allow us to recycle old systems into multiple new ones without new material resources, reducing the need for mining and refining.
|Van der Waals cylinder-and-sleeve bearing
© K. Eric Drexler
Wear limits the lives of mechanical and structural systems, which often attain a reasonable lifetime only by having worn-out parts replaced. (An annoying example is the modern automobile). Wear is cumulative and can seem exponential, as worn parts increase wear on other parts. The aim of wearproofing is to head off the wear process, with the increasingly ambitious goals of longer-lasting parts, zero-wear parts, and finally self-repair. Using nanotechnology, we can expect improvements in:
Wear on tools can be reduced even for bulk processes by
forming parts using non-contact methods such as explosives,
lasers, electron beams, plasma torches, water jets, and
electromagnetic forming instead of drill bits, grinding wheels,
and the like. There may be uses where such macro-tools forever
outperform nano-tools: perhaps in well drilling, tunneling, and
Synergies between the above techniques can be expected; for example, making an object with voids but covering the surface with diamond. And besides saving energy in manufacturing, we can expect to do so in transport as well: objects will last longer and so need to be delivered less often, they will weigh less when they do need transport, and with nanoproduction systems--quiet, small, flexible, and clean--manufacturing on-site becomes a possibility.
But eventually, we can expect self-repair to solve the wear problem.
Jerry Fass is a part-time science writer based in Wisconsin. He also coordinates FI's journal monitoring project.
|Foresight Update 4 - Table of Contents|
[Note: A related article on the computational limits of the human brain is available in Update 6.]
Today it is commonplace to compare the human brain to a computer, and the human mind to a program running on that computer. Once seen as just a poetic metaphor, this viewpoint is now supported by most philosophers of human consciousness and most researchers in artificial intelligence. If we take this view literally, then just as we can ask how many megabytes of RAM a PC has we should be able to ask how many megabytes (or gigabytes, or terabytes, or whatever) of memory the human brain has.
Several approximations to this number have already appeared in the literature based on 'hardware' considerations (though in the case of the human brain perhaps the term 'wetware' is more appropriate). One estimate of 1020 bits is actually an early estimate (by Von Neumann in The Computer and the Brain) of all the neural impulses conducted in the brain during a lifetime. This number is almost certainly larger than the true answer. Another method is to estimate the total number of synapses, and then presume that each synapse can hold a few bits. Estimates of the number of synapses have been made in the range from 1013 to 1015, with corresponding estimates of memory capacity.
A fundamental problem with these approaches is that they rely on rather poor estimates of the raw hardware in the system. The brain is highly redundant and not well understood: the mere fact that a great mass of synapses exists does not imply that they are in fact all contributing to memory capacity. This makes the work of Thomas K. Landauer very interesting, for he has entirely avoided this hardware guessing game by measuring the actual functional capacity of human memory directly (See "How Much Do People Remember? Some Estimates of the Quantity of Learned Information in Long-term Memory", in Cognitive Science 10, 477-493, 1986).
Landauer works at Bell Communications Research--closely affiliated with Bell Labs where the modern study of information theory was begun by C. E. Shannon to analyze the information carrying capacity of telephone lines (a subject of great interest to a telephone company). Landauer naturally used these tools by viewing human memory as a novel 'telephone line' that carries information from the past to the future. The capacity of this 'telephone line' can be determined by measuring the information that goes in and the information that comes out, and then applying the great power of modern information theory.
Landauer reviewed and quantitatively analyzed experiments by himself and others in which people were asked to read text, look at pictures, and hear words, short passages of music, sentences, and nonsense syllables. After delays ranging from minutes to days the subjects were tested to determine how much they had retained. The tests were quite sensitive--they did not merely ask 'What do you remember?' but often used true/false or multiple choice questions, in which even a vague memory of the material would allow selection of the correct choice. Often, the differential abilities of a group that had been exposed to the material and another group that had not been exposed to the material were used. The difference in the scores between the two groups was used to estimate the amount actually remembered (to control for the number of correct answers an intelligent human could guess without ever having seen the material). Because experiments by many different experimenters were summarized and analyzed, the results of the analysis are fairly robust; they are insensitive to fine details or specific conditions of one or another experiment. Finally, the amount remembered was divided by the time allotted to memorization to determine the number of bits remembered per second.
The remarkable result of this work was that human beings remembered very nearly two bits per second under all the experimental conditions. Visual, verbal, musical, or whatever--two bits per second. Continued over a lifetime, this rate of memorization would produce somewhat over 109 bits, or a few hundred megabytes.
While this estimate is probably only accurate to within an order of magnitude, Landauer says "We need answers at this level of accuracy to think about such questions as: What sort of storage and retrieval capacities will computers need to mimic human performance? What sort of physical unit should we expect to constitute the elements of information storage in the brain: molecular parts, synaptic junctions, whole cells, or cell-circuits? What kinds of coding and storage methods are reasonable to postulate for the neural support of human capabilities? In modeling or mimicking human intelligence, what size of memory and what efficiencies of use should we imagine we are copying? How much would a robot need to know to match a person?"
What is interesting about Landauer's estimate is its small size. Perhaps more interesting is the trend--from Von Neumann's early and very high estimate, to the high estimates based on rough synapse counts, to a better supported and more modest estimate based on information theoretic considerations. While Landauer doesn't measure everything (he did not measure, for example, the bit rate in learning to ride a bicycle, nor does his estimate even consider the size of 'working memory') his estimate of memory capacity suggests that the capabilities of the human brain are more approachable than we had thought. While this might come as a blow to our egos, it suggests that we could build a device with the skills and abilities of a human being with little more hardware than we now have--if only we knew the correct way to organize that hardware.
This article is also available on Dr. Merkle's Web site.
Dr. Merkle's interests range from neurophysiology to computer security. He recently spoke on nanotechnology and biostasis at the Life Against Death Conference in San Francisco.
|Foresight Update 4 - Table of Contents|
One of the contributions by K. Eric Drexler to nanotechnology
was his success with estimating the behavior of nanomachines by
using simple mechanical calculations. Ultimately, however, these
exploratory engineering calculations remain approximations only.
Serious nanomachine design will require much more. Almost
certainly it needs very powerful computers able to carry out
dynamic calculations on large molecules. These calculations need
lots of computer power. Specialized chemical workstations with
prices in the range of $200,000 already exist.
To speed nanotechnology along what we really want is lower price computers, ideally costing no more than a Mac II. There is a wide open road to just such a computer. Technology for chemical design workstations costing about $40,000 exists right now, for the trouble of assembling a system from standard boards (unfortunately not done yet). The same parts will cost far less in a few years (so Popular NanoMechanics may start publication soon).
The technology depends on the Transputer, a chip specially designed for parallel processing. Computer System Architects sells IBM PC boards with 16 Transputer chips and 16 megabytes of memory for $28,000.
Chemical Design Ltd, a British company, already sells a chemical design workstation, the MITIE 1000, which can contain as many as 36 independent Transputers. The smallest MITIE 1000 sells for $170,000. The MITIE calculates as much as 72 times faster than a VAX 8600, analyzing the conformation and dynamics of large molecules at supercomputer speeds.
The MITIE contains a microVAX as a host machine; the remaining modules run on the VAX. Chemical Design has about 250 customers around the world, including Glaxo, Rhone-Poulenc, Fisons, Dupont, American Cyanamid, Merck, and Hoffmann-LaRoche. The program ChemX contains specific modules for building and displaying the molecule (ChemCore), modelling molecules (fitting, analyzing the conformation: ChemModel), designing proteins (ChemProtein), and carrying out calculations to find minimum energy states (ChemQM). There are also library modules to maintain a large database (ChemLib: the recommended size of hard disk for a single-user system is 70 Megabytes).
Any molecular machine we design must be chemically stable in the environment for which we design it. We must therefore make sure not just that the molecule would be stable if isolated from all other chemicals but also that the system will withstand likely chemical attack. Molecules will try to attain minimum energy states, and their excited states are also of interest. To resolve all of these issues will require very fast chemical design software. Ultimately software for nanomachine design will do much more, but even existing chemical design software running on an affordable workstation puts us far ahead.
What about the software? Unfortunately, porting software to a parallel computer usually requires a total rewrite of any modules in which you expect to use parallelism. Porting should be a cooperative effort between someone versed in parallel computing and someone versed in chemical design software. When someone will get a chemical design show on the road, porting software to a MAC II-transputer system, isn't clear to me. My own expertise lies in parallel computing. Anyone interested can reach me through Foresight Update.
Dr. Thomas Donaldson currently writes software for a transputer machine for the FEM market. He pioneered the idea of artificial enzyme systems as an approach to cell repair.
|Note: One example of recent use of
parallel computers for complex molecular dynamics
calculations can be found on the web at:
From Foresight Update 4, originally published 15 October 1988.
Foresight thanks Dave Kilbridge for converting Update 4 to html for this web page.