Upload
jeffrey-funk
View
726
Download
1
Embed Size (px)
Citation preview
Thinking About the Future of Technology and Emerging Entrepreneurial Opportunities:
Rates of Improvement and Economic Feasibility
by
Jeffrey L. Funk
Associate Professor
National University of Singapore
Division of Engineering and Technology Management
9 Engineering Drive 1, Singapore 117576: EA-5-34
Christopher L. Magee
Professor of Practice
Engineering Systems Division
Massachusetts Institute of Technology
N52-395, 77 Massachusetts Avenue, Cambridge, MA
1
Thinking About the Future of Technology and Emerging Entrepreneurial Opportunities:
Rates of Improvement and Economic Feasibility
Abstract
This paper uses data on rates of improvement to discuss when new technologies or systems
composed from them might become economically feasible. Technologies must provide some
level of performance and price for specific applications before they will begin to diffuse and
technologies that experience rapid rates of improvement are more likely to become
economically feasible for a growing number of applications than are other technologies.
Drawing from a large data base on rates of improvement, this paper describes a set of
plausible futures that are very different from ones that are presented in public forums.
2
1. Introduction
Reaching conclusions about when new technologies might become economically feasible
without understanding rates of improvement is all too common in today’s world. Think of the
clean energy debate. In spite of the regular coverage by the mass media of clean energy
technologies such as solar, wind, batteries for electric vehicles, and bio-fuels, there is very
little mention in the press or in technical writing of the relative rates of improvement that
these technologies are experiencing. Thus, even well-educated people have little chance of
understanding their rates of improvement or their probabilities of becoming economically
feasible in the near future.
Instead, the public debate revolves around what Nobel Laureate Daniel Kahneman1 calls
“instinctive and emotional” thought. People tend to assess the relative importance of issues
by the ease with which they are retrieved from memory and this is largely determined by the
extent of coverage in the media. The media talks about solar, wind, battery-powered vehicles,
and bio-fuels and thus many people think these technologies are experiencing rapid rates of
improvement when many are not (e.g., wind, 2% a year; Li-ion batteries, 5%2) in spite of the
large improvements that are needed before they will become economically feasible. Second,
judgments and decisions are guided directly by feelings of liking and disliking, with little
deliberation and reasoning. Kahneman recounts a conversation he had with a high-level
financial executive who had invested in Ford because he “liked” their products without
considering whether Ford stock was undervalued. Similarly, some people believe in demand-
based subsidies for clean energy because they “like” the notion of energy from the sun and
wind; but few of these people consider their rates of improvement and thus their probabilities
of becoming economically feasible.
We believe there is a better way for firms, universities, and governments to participate in
the clean energy debate and discussions of demand-based subsidies and the future of
3
technology in general. Daniel Kahneman labels this mode of thinking “slower, more
deliberative, and more logical.” We argue that the rates of improvement are a necessary part
of this slow deliberative thinking and it can help firms, universities, and governments better
understand when new technologies might become economically feasible and thus make better
investment, research, policy, and educational decisions. We use the term economic feasibility
in reference to a cost and performance comparison by the marketplace between the new and
old technologies (or other competing technologies). We distinguish this economic feasibility
from organizational, legal, and regulatory challenges that also exist particularly for large
complex systems and ones with strong network effects.
A second variable of importance for understanding economic feasibility is the amount of
improvements that are needed before a technology in the laboratory becomes economically
feasible or before a commercialized technology becomes economically feasible for a growing
number of applications or it leads to changes in the way we design higher level systems.
Combining the amounts of improvements that are necessary with the rates of improvement
defines a 2-by-2 matrix (See Figure 1) in which the x-axis represents rates of improvement
and the y-axis represents amount of improvement needed. Technologies in the upper right
quadrant have recently or will soon become economically feasible. Ones in the bottom left
will probably never be economically feasible and ones in the other two quadrants may or may
not become economically feasible in the near or far future.
This paper first summarizes technologies that are experiencing improvements of greater
than 10% per year and the technical drivers of these improvements. Second, the impacts of
these rapid rates of improvement on economic feasibility and thus plausible futures of
technology are then described. The dramatic differences between these plausible futures and
the ones that are ordinarily discussed in public forums, including the mass media,
government agencies, and universities, suggest there is value in doing this type of analysis
4
and in collecting additional data on rates of improvements and the degree of improvements
needed. Third, this paper summarizes the implications of these plausible futures and how
governments, firms, and universities can better think about the futures of technology.
2. Technologies Experiencing Rapid Rates of Improvement
Technologies that experience rapid rates of improvement are more likely to become
economically feasible than are technologies that experience slower rates of improvement
(particularly since the data here and elsewhere show that rates of improvement are relatively
constant with time). They are also more likely to quickly diffuse than are other technologies
since rapid rates of improvement in performance and cost lead to increased profitability for
firms that adopt the new technology, which the early diffusion research found to have the
largest effect on rates of diffusion3. They are also probably more likely to induce cognitive
biases in managers than are slower moving technologies because rapid rates are particularly
difficult for most people to comprehend.
Summarizing those technologies that are experiencing rapid rates of improvement is highly
problematic. Ideally, for each technology, we would have rates of improvement data for all
the relevant dimensions of performance and cost, data on customer preferences, and a
comparison with the competing technologies, both existing and new ones, along each
dimension that is important to existing and potential customers. Such a comparison would
also be for all the systems that might be made economically feasible by the improvements in
the various component technologies.
New systems are of particular interest since new technologies often make new systems
possible and many of the technologies discussed in this paper continue to make new forms of
transportation, environmental, energy, health care, and other systems economically feasible.
One goal of this paper is to help firms, government employees, professors and students think
more effectively about these systems including their decades of expected lifetimes and the
5
need for adapting to new technologies as they become economically feasible. Furthermore,
since these systems involve a large variety of technologies, many of which are new ones, it is
difficult for a single firm, much less a single person, to understand all the technologies that
are impacting or might impact in the future of a specific system.
Nevertheless, it is not possible to present this level of detail in a single paper. This paper
summarizes data and analysis that are presented in much more detail in other publically
available sources4); the summary begins with a list of rapidly improving technologies (See
Table 1). The data in Table 1 was found in Science, Nature, IEEE, and other science and
engineering journals through extensive reading and searches. Improvements in the
performance and cost of these technologies are represented by improvements in specific
technical ratios that engineers and scientists have chosen to measure performance and cost
because these ratios capture the economics of the technology. In these output-to input ratios,
output is typically a dimension of performance and input is cost or some surrogate of cost
such as volume, weight, area, energy, power, or time. Improvements in these ratios often
represent improvements in multiple dimensions of both performance and cost.
Most of the technologies shown in Table 1 have already been commercialized and thus their
biggest effect on the future will be through their impact on existing and new systems.
Technologies that have not yet, or only recently been commercialized include LEDs for
lighting, OLEDs for lighting and displays, organic, quantum dot and Perovskite solar cells,
superconductors for energy transmission and Josephson junctions, quantum computers,
carbon nano-tubes, new forms of non-volatile memory and cellulosic ethanol.
Returning to Table 1, we characterize the improvements as annual rates of improvement
since most of the time series data are straight lines on a logarithmic plot. We define
“currently” as time series data that includes data from the last 10 years. Thus, although some
of the time series shown in Table 1 include data from 30 years ago, the technology is included
6
in Table 1 because the times series includes data from the last 10 years and the older parts of
the time series data are retained for completeness. The technologies with fewer data points
are ones that are relatively new (and thus potentially very interesting) and thus much less data
is available.
This section’s discussion of the improvements is organized by the drivers of them where
this discussion builds from a previously published paper by one of the authors in California
Management Review and related research5 The recently published paper described two
drivers of improvements: 1) creating new materials (and often their associated processes) to
better exploit their underlying physical phenomena; and 2) geometric scaling. Some
technologies directly experience improvements through these two mechanisms while those
consisting of higher-level “systems” indirectly experience them through improvements in
specific “components.” The most rapid improvements are primarily from a subset of these
two mechanisms. First, creating new materials (and processes for them) can lead to sustained
rapid improvements in performance and cost when new classes of materials are continuously
being created. Second, technologies that benefit from reductions in scale (e.g., integrated
circuits) have experienced much more rapid improvements than have technologies that
benefit from increases in scale (e.g., engines).
2.1 Creating Materials to Exploit Physical Phenomena
Many of the technologies in Table 1 benefit from creating new materials to better exploit
their underlying physical phenomena where there is a tight linkage between creating
materials and the processes for making them. We use the term “create” because these
materials do not occur naturally and thus scientists and engineers must literally create them
and the processes used to manufacture them. Beginning with the top of Table 1, scientists and
engineers increased the performance and reduced the cost of LEDs by creating materials that
7
better exploit the phenomenon of electroluminescence where many of them can be defined as
new classes of materials. Creating new combinations of semiconducting materials with
gallium, arsenic, phosphorus, indium, and selenium enabled new LED colors and directly
increased efficiencies (measured in luminosities per Watt) while indirectly reducing their
costs. Similar improvements occurred with organic LEDs (OLEDs) as small molecules,
polymers, and phosphorescent materials were created.
Similar arguments can be made for photo-sensors, solar cells, organic transistors, quantum
dot displays, carbon nanotubes for electronic and structural applications, and superconducting
Josephson junctions and cables. New combinations of semiconductors and other materials
were created that convert more photons to electrons than do other materials in both photo-
sensors and solar cells. New organic materials were created that have higher mobility for
transistors than do other materials. New semiconductors and processes for them were also
created that better exploit the phenomenon of quantum dots and have higher efficiencies, i.e.,
they efficiently translate electrons into photons. Improvements in carbon nanotubes and
grapheme mostly came from new processes that enabled lower costs. The new processes also
enabled higher purity and density for carbon nanotube transistors and larger area sheets for
grapheme.
New superconducting materials were created that have higher “critical” temperatures,
current densities, and magnetic fields. As the name implies, superconducting materials have
zero resistance and thus infinite conductance when the temperature falls below a “critical
temperature” and the current and magnetic fields stay below their critical values. Scientists
and engineers have created more than 30 superconducting materials with the highest critical
temperature being above 170o Kelvin or about -100 C o. The increases in critical temperatures,
current densities, and magnetic fields are relevant for MRIs, energy applications, and
electronic devices such as superconducting Josephson Junctions. Superconducting Josephson
8
Junctions are potentially important because they have several orders of magnitude lower
power consumption and higher speeds than do transistors.
2.2. Increases in Scale
Many technologies benefit from increases in scale; these include engines, transportation
equipment6, and production equipment where the benefits of increases in scale can be
explained using the concept of geometric scaling. Since many of the technologies in Table 1
(and those not in the table) benefit from increases in the scale of equipment used to produce
them, this sub-section focuses on production equipment and how some equipment benefits
more from increases in scale than do other types of equipment.
Chemical plants, other material processing plants, and many kinds of flat material based
technologies such as displays benefit more from increases in physical scale than do assembly
plants7. For example, with chemical plants, the costs of pipes vary as a function of radius
whereas the outputs from pipes vary as a function of radius squared. Similarly, the costs of
reaction vessels vary as a function of surface area (radius cubed) whereas the output of a
reaction vessel varies as a function of radius cubed. These advantages of increases in physical
scale have been confirmed in empirical analysis where capital costs of chemical plants rise
much slower than does output as the physical scale of the pipes and reaction vessels are
increased8.
More recent technologies that benefit from increases in the physical scale of production
equipment are cellulosic ethanol, carbon nanotubes, superconductors, displays, solar cells,
and other electronic products that can be roll printed. The cost of cellulosic ethanol has
dropped for the same reason that costs have fallen for other chemicals; costs rise slower than
do output as the physical scale of the equipment is increased9. The energy costs of carbon
nanotubes have fallen as the scale of their facilities has been increased. This is because heat
loss is typically a function of surface area and carbon nanottube production is a function of
9
volume; similar results have been found with aluminum10.
For displays, analyses have found that the capital cost per area output (substrate area per
hour) has fallen as the size of the substrate and production equipment has been increased.
Similar cost reductions have occurred with solar cells, organic transistors and other types of
displays as the scales of the substrate and the production equipment have been increased and
such cost reductions are expected to occur with quantum dot displays as their substrate sizes
are increased. Roll-to roll printing also benefits from increases in scale and costs of roll-to
roll printing are much lower than with traditional manufacturing processes11. Technologies
that can be roll printed include some types of solar cells, displays, and other electronics.
Organic materials including ones for solar cells and displays can often be roll printed.
Similar arguments can be made for the wafer sizes of ICs and LEDs, which are currently
produced on 2 to 4” wafers before they are cut into single LEDs. Since ICs are produced with
12” wafers and soon to be 18” wafers, the cost of LEDs is expected to continue falling as
larger wafers are implemented12 and as new materials are created. However, by itself,
increases in the scale of production equipment do not lead to the rapid rates of improvement
that are found with reductions in scale (see next section) and with creating new materials.
Thus, we cannot expect rapid rates of improvements from technologies that only benefit from
increases in the scale of production equipment.
2.3. Reductions in Scale
Some of the most rapid rates of improvement have been achieved with technologies that
benefit from reductions in scale and the concept of geometric scaling helps us understand
when this will likely occur. As an aside, these benefits are not related to the usual economies
of scale but instead are associated with the physical laws that govern a technology and the
technology’s geometry.
Reducing the scale of transistors, storage regions, and other dimensional features has led to
10
many orders of magnitude improvements in the cost and performance of ICs, magnetic and
optical discs, and newer types of ICs such as MEMS and bio-electronic ICs. This is because
for these technologies, reductions in scale lead to improvements in both performance and
cost. For example, placing more transistors or memory cells in a certain area of an IC
increases the speed and functionality and reduces both the power consumption and size of the
final product, which are typically considered improvements in performance for most
electronic products. The reductions in scale also lead to lower material, equipment, and
transportation costs. The combination of both increased performance and reduced costs as
size is reduced has led to many orders of magnitude improvements in the performance to cost
ratio of many ICs. For example, three orders of magnitude reductions in transistor length
have led to about nine orders of magnitude improvements in both the cost of an individual
transistor and the number of transistors on a chip13. Similar arguments can be made for
magnetic and optical storage. Reductions in the magnetic storage area enabled increases in
the magnetic recording density of magnetic cores, drums, disks, and tape, which led to
improvements in both speed and cost. For optical discs, reductions in the wavelength of light
emitted by semiconductor lasers are needed to reduce the size of storage cells.
Looking to newer technologies, similar arguments can be made for MEMS, bio-electronic
ICs, and DNA sequencing equipment. MEMS are used in motion sensors for Nintendo’s Wii,
nozzles for ink jet printers, in the sensing for micro-gas analyzers, and in the building blocks
for optical computing (e.g., waveguides, couplers, resonators, and splitters) and they are
fabricated using some of the same equipment and processes that are used to construct ICs.
Reductions in the scale of the relevant dimensions dramatically increase the performance of
some types of MEMS and also the number of transistors available for processing the
information. For example, reductions in feature size lead to higher sensitivity, lower energy
usage, faster response time, and lower costs for a mciro-gas analyzer and many types of
11
sensors including bio-electronic ones14. Bio-electronic ICs are basically a MEMS with micro-
fluidic channels and they are used to sense and analyze biological material in for example
point-care diagnostics, to provide better forms of drug delivery in for example IC-controlled
smart pills, and to control artificial implants15. Finally, although DNA sequencers use a
variety of different materials and processes, all of them involve reductions in the scale of the
relevant features and these reductions in physical scale are the major reason for the multiple
orders of magnitude reductions in the cost of sequencing and synthesizing DNA16.
One technology field that benefits from both reductions in scale and in creating new
materials is nanotechnology, particularly nanotechnologies with single digit feature sizes.
Improvements from nanotechnology mostly arise by creating materials that benefit from
single nanometer feature sizes and the number of material classes has grown quickly over the
last decade. Such materials include many classes of carbon nano-tubes (CNTs), graphene,
nano-particles, nano-fibers and many other “ultra-thin” materials. For example, the number of
materials that have been constructed with single-, double- or triple atom thicknesses has
already exceeded 10 and is growing quickly17. The creation of new classes of materials and
improvements in their performance has been made in spite of their low rates of production.
Rapid rates of improvements through creating new materials and processes without high
levels of production is common for many of the technologies discussed earlier; these include
OLEDs, organic transistors, and new forms of solar cells.
2.4. Impact on Higher-Level Systems
Many of the technologies in Table 1 experienced improvements in cost and performance
because specific “components” in the “technological system” have experienced rapid
improvements while having a large impact on the overall systems performance and cost.
Rapid improvements in electronic components, in particular ICs, have led to rapid
12
improvements in computers18 where the improvements in ICs are driven by reductions in
scale. As one computer designer argued, by the late 1940s computer designers had recognized
that “architectural tricks could not lower the cost of a basic computer; low cost computing
had to wait for low cost logic”19, which eventually came in the form of better ICs. For
example, an order of magnitude improvement in the numbers of transistors per chip about
every seven years (See Table 1) led to similar levels of improvements in computations per
second and per kilowatt hour of computers20.
Similar arguments about the role of ICs can be made for other electronic products such as
digital cameras, eBook readers, video games, high density television, set-top boxes, servers,
and routers, and also for even higher level systems such as corporate information systems21.
Not only do the cost of ICs and other electronic components make up more than 95% of the
cost of many electronic products22, the performance of these products is largely determined
by the speed, functionality and power consumption of ICs. Furthermore, improvements in
computers have led to improvements in the performance and cost of medical equipment such
as magnetic resonance imaging (in addition to magnetic materials) and computer assisted
tomography23.
Rapid improvements in electronic components such as vacuum tubes, ICs, lasers, and
photo-sensors (See Table 1) also led to rapid improvements in data speeds and spectral
efficiency of both wireline and wireless telecommunication. Improvements in electronic
components enabled faster data speeds and higher bandwidth for single cable, coaxial cable
and more recently optical cable where improvements in optical fiber also required
improvements in the purity of glass24. For wireless, although the introduction of cellular
systems, smaller cells, and better protocols for these cellular systems were needed to achieve
improvements in the data rates for wireless communication, rapid improvements in ICs
enabled the implementation of these cellular systems, smaller cells, and better protocols and
13
also improvements in mobile phones. Cellular systems required faster switching speeds,
which were enabled by the improvements in computers and ICs mentioned in the previous
paragraphs25.
3. Thinking about the Future
Many of the technologies discussed in the previous sections will continue to experience
rapid improvements. Some of these technologies will become economically feasible for their
first applications while many will become economically feasible for a growing number of
applications. More importantly, we live in a “systems”-based world in which improvements
in components impact on systems and the number of ways in which we can combine
components into new higher-level systems is growing exponentially. Very high-level systems
that are being impacted by the improvements include homes, offices, buildings, health care,
communication, and transportation.
Identifying those technologies that are experiencing rapid improvements and the drivers of
their rapid improvements enable us to more effectively think about the future of these
technological systems. Knowing these technologies and their drivers of improvements
enables us to think about the future in a “slower, more deliberative, and more logical” mode
than an “instinctive and emotional” mode as characterized by Kahneman. In turn, this enables
us to better understand when new technologies become economically feasible, devise better
R&D policies, and find better solutions to global problems such as urban congestion and
sustainability.
Technologies that benefit from reductions in scale are experiencing some of the most rapid
rates of improvements over long periods of time. Improvements in ICs and magnetic storage
will likely continue for the next 20 to 40 years as further reductions in scale are made. With
respect to ICs, smaller wavelength light sources (e.g., 13 nm extreme ultraviolet) for
photolithography and three-dimensional (3D) ICs will probably give us another 15 years in
14
Moore’s Law before newer technologies become economically feasible. The new 13nm
wavelength light is more than 1/10 as small as the previous wavelength light source and is
expected to enable feature sizes that are less than 5 nm. 3D ICs enable increases in the
number of transistors per chip by building up as opposed to reducing feature sizes. New
layers are added with ten or even 100 layers of transistors or memory cells being
contemplated. Optical wave channels will probably be used as one of these layers in the
future, thus enabling faster interconnect speeds between transistors.
Newer technologies such as carbon nanotubes, superconducting Josephson junctions and
quantum computers are also experiencing rapid rates of improvement (See Table 1) and thus
may become economically feasible before the improvements mentioned in the previous
paragraph are completed. Carbon nanotubes have the potential to replace silicon as the
channel material in the next 10 to 20 years if improvements in the purity, density, and
directionality of carbon nano-tubes are continued26. Josephson Junctions, which use
superconductors, are much faster and consume much less power than do conventional ICs
and they can also be used to construct quantum computers. Google’s recent purchase of
quantum computers has caused some to believe that they are already economically feasible
for some applications. Furthermore, since the number of “Qubits is steadily rising and the
economic feasibility of quantum computers is a non-linear function of their number, the
economic potential of quantum computers are probably extremely large. All of this suggests
that improvements in computers will continue to occur and these improvements will continue
to change the way we manage large systems.
3.1 Analysis and Control of Systems with Computers
Continued improvements in ICs and other electronic components will enable continued
improvements in the electronic products mentioned in an earlier section. Just for computers,
15
continued improvements in them through better ICs and through new types of them such as
quantum computers will enable more extensive control over our world independent of
whether one agrees with this trend. Improvements in computers enable us to better control
and analyze homes, offices, factories, laboratories, buildings, health care, communication,
and transportation. For analysis, improvements in computers enable us to analyze more of the
output from scientific equipment, whose output is growing very rapidly. This equipment
includes particle accelerators, telescopes, DNA sequencing equipment, and other types of
medical equipment, all of which create vast amounts of data.
Perhaps more importantly, improvements in computers and information storage are
facilitating the emergence of so-called “Big-Data” analytic services27, whose hardware and
software sales are expected to reach $23.8 Billion by 201628. These services are basically
large mathematical models that are used to make predictions. Thus, in addition to pursuing
more efficient algorithms, big-data proponents build models that include hundreds of
variables since the cost of computation is very low. For example, data from the Internet are
enabling better translations and better predictions of flu trends, inflation, health problems,
loan defaults, and rising food prices. Similar computation and data combinations also enable
analysis of more complex socio-technical issues such as the chances of riots or terrorism.
Many of these big-data analytics are also being enabled by improvements to the Internet.
In addition to the improvements in glass fiber, lasers, photodiodes, and other components,
improvements in photonics will ensure that the current barrier (conversion between electrons
to photons) will not significantly slow progress. Converting electrons to photons and photons
to electrons is becoming the bottleneck in the Internet and improvements in photonics are
addressing this bottleneck. Rapid improvements in photonics are also causing more of the
data communication between and within computers to be done with optical channels. For the
future, the trend is towards more optical communication between chips on circuit boards and
16
later for communication within individual chips. For chips, this may well result in the optical
channels becoming another layer in a 3D IC.
Even bigger economic advances are potentially in our future as improvements in sensors
enable a larger variety of data analyses, new types of electronic systems and new forms of
control. The cost and performance of camera chips continues to be improved as feature sizes
are reduced and similar developments are occurring with MEMS-based sensors as their
feature sizes are reduced beyond their current levels to those found in ICs. To put this in
perspective, feature sizes for MEMS are currently about one-half to one micron while ICs
were being fabricated with such feature sizes in 1980. Subsequent reductions in the feature
sizes on ICs after 1980 enabled new types of electronic systems such as personal computers,
mobile phones, video games, and the Internet to emerge. Similar reductions in feature size
over the next 10-20 years for MEMS can enable new forms of electronic systems such as
better gas chromatographs and other sensors, better mobile phone filters and other passive
components, better inkjet printers, and better bio-electronic ICs to become widely used as
they become as cheap as pocket calculators. In combination with conventional ICs and lasers,
improvements in MEMS are also driving improvements in 3D scanners and printers,
holograms, eye-tracking devices, and many other sensors for factories, vehicles, buildings,
dams, bridges, power plants, infrastructure, and other systems. All of these sensors will
enable better maintenance and management of systems and this can lead to lower energy
costs and more sustainable systems.
For example, consider lighting. In addition to the greater efficiencies available with LEDs
than with incandescent and fluorescent lights, their small sizes enables more aesthetic designs
for lighting fixtures, their costs are falling partly because the size of wafers are being
increased, and by using sensors, we can create lighting systems that only illuminate those
areas that are needed when they are needed. Motion, heat and other sensors can track the
17
movements of humans, animals, and vehicles in order to provide more efficient and effective
illumination. This suggests that smart lighting systems can have a much larger impact on our
energy usage than is ordinarily thought, but will require redesign of our homes, offices, and
public spaces.
3.2 Wireless Communication and Mobile Devices
Improvements in wireless technologies enable a broader variety of sensors to be accessed.
Greater amounts of environmental, physiological, traffic and infrastructure-related data can
be collected for big-data analysis and interpretation when the data can be wirelessly sent to
the Internet without expensive and difficult to maintain wires. Environmental data includes
temperature, pressure, and gas content. Physiological data includes those of heart rate, brain
wave, and blood pressure. These and other data can help us better manage traffic, food
supplies, and infrastructure such as factories, buildings, dams, bridges, and power plants. For
traffic, one goal should be to dramatically reduce public and private vehicle breakdowns and
accidents by monitoring vehicles. The importance of managing traffic will further increase if
automated vehicles begin to diffuse (see below).
Improvements in wireless and other technologies will probably make mobile phones or
other personal mobile devices a major collection and control point for many sensors. With
advanced processors, memory, and displays, mobile phones can be used to control home
appliances and collect data about our environment and bodies (see below) simply by placing
the sensors in or attaching them to phones. Such devices can be used to control and program
the thermostat, lighting, and other appliances in the home and test strips for blood, skin, and
saliva can be attached so that checks can be done for flu, insulin, or other medical conditions.
Similarly, electrodes, microscopes, ultrasound, and portable MRIs can conceivably be
connected to these devices.
Improvements in mobile phones and GPS also improve the economic feasibility of bicycle
18
rentals. Bicycles are revolutionizing transportation in many U.S. and European cities and
rental facilities can take this one step further. Such rentals facilitate bicycle usage as a single
mode of transportation and as a supporter of bus and light rail transportation. The latter
become possible because mobile phones and GPS can facilitate sharing and thus reduce
bicycle congestion at storage facilities. Mobile phones can be used to find and rent bicycles
while GPS can be used to track the bicycles in the system and thus move them as bicycles
stack up at a single location.
One reason for optimism about mobile personal devices is that improvements in human-
computer interfaces continue to occur from improvements in various components. In addition
to faster speeds and better spectral efficiencies for wireless systems, which are mostly being
driven by Moore’s Law, improvements in human-computer interfaces are also occurring as
displays, ICs, MEMS, and other components are improved. The cost of LCDs have steadily
dropped and newer types of displays such as organic light emitting diode (OLED),
electrophoretic (e-paper), and new forms of touch displays continue to emerge, many of
which can be produced using cheaper methods such as roll-to-roll printing, OLEDs are much
more flexible than are LCDs and thus can conform to our wrists and other portions of our
bodies. As an aside, displays continue to become more widely available in our societies as
information sources and this will likely continue as their costs and performance are improved.
But improvements in displays are just the first step for continued improvements in human-
computer interfaces. For example, gesture displays and augmented reality benefit from
improvements in camera chips, ICs, displays, and MEMS and they enable us to interact with
our mobile phone and other computing devices in new and exciting ways. While Google
Glasses have received a great deal of attention, it is important to realize that other types of
human-computer interfaces will emerge as firms combine the relevant components in new
and exciting ways.
19
Improvements in MEMS impact on both touch displays and on neural interfaces. For the
former, some types of MEMS enable users to feel texture and perhaps facilitate remote
surgery. Others enable users to find the right place on a display as they feel for the right
buttons or depressions on the display in the dark. For the latter, MEMS-based neural
interfaces are pressed into the skull, they benefit from reductions in scale, and they enable
physically impaired individuals to interact better with computers and thus with the world
around them. As the reductions in scale continue, we can expect better neural interfaces and
thus better lives for physically impaired individuals and possibly the diffusion of neural
interfaces to others who are so inclined. It is important to recognize that improvements in
these interfaces will continue because these MEMS and other components benefit from
reductions in scale and thus improvements similar to those experienced by ICs area probable.
3.3 Health Care
Health care is a unique in that it is experiencing rapid increases in cost even as the cost and
performance of new technologies are being improved. Since the reasons for the rising cost of
health care go far beyond the scope of this paper, we merely focus on the improvements that
are and will likely continue to occur. We have already mentioned how improvements in ICs
and computers have led to dramatic improvements in many kinds of medical equipment.
Here we focus on another impact of Moore’s Law on health care, the combination of
biology and electronics or bio-electronics for short. One form of bio-electronics is bio-
electronic ICs. They are a special type of MEMS that uses bio-compatible materials. They
can be used to analyze blood, urine, sweat, and other biological entities. Like MEMS, most
bio-electronic ICs benefit from reductions in scale and thus dramatic improvements in
performance and cost are probably as their feature sizes are reduced in the future. Smaller
feature sizes enable faster reactions and response time, higher throughput, the analysis of
20
smaller biological materials, and reductions in sample and reagent volume. The creation of
new materials for analyses should also support rapid improvement.
Such developments should enable dramatic improvements in the cost and performance of
MEMS for smart pills and diagnostic equipment. Smart pills contain ICs, camera chips, and
other miniaturized components and they might be used to target cancerous and other
unhealthy cells and thus minimize the impact of treatments on healthy cells. Treatments could
potentially be transported to the unhealthy cells via embedded cameras, magnets, and other
devices. Similarly, point-of care diagnostic equipment can be revolutionized with bio-
electronic ICs. Like the gas chromatograph example cited above, reductions in the cost of
diagnostic equipment will occur as the cost of bio-electronic ICs are achieved and this can
enable cheaper and faster health care monitoring.
Reductions in the scale of bio-electronic ICs can also enable improvements in medical
implants. For example, pixel data from cameras mounted on glasses can be wirelessly sent to
a MEMS-based electrode that is implanted into a person’s optic nerve. Reductions in the size
of the electrodes, which are enabled by the same types of improvements experienced by ICs,
enable better connections to the optic nerve and thus enable improved eyesight. Further
reductions in scale will likely bring 20-20 vision to many people who suffer from macular
degeneration, a disease inflicting about 0.5% of Americans and likely to rise as more people
live longer.
Finally, better physiological data can be collected through a combination of bio-electronic
sensors, wireless transmitters, receivers, and mobile phones. One kind of bio-electronic
sensor is a skin patch. Improvements in the mobility of organic transistors have enabled
greater use of flexible organic materials in skin patches. Thus, just as displays that use
organic materials such as OLEDs are more flexible than are LCDs, skin patches constructed
from organic materials are also more flexible than are silicon. On the other hand, new forms
21
of designs enable conventional silicon-based ICs to be used in these skin patches. One such
design is called an island-bridge design for its use of flexible bridges to connect the IC-based
islands. These patches can inexpensively gather physiological data and transmit the data
wirelessly to one’s mobile phone or other Internet-enabled device. Like displays,
improvements in the performance and cost of skin patches are being made with new
materials, new production equipment such as roll-to roll printing, and increases in scale of the
equipment.
3.4 Transportation
We have already noted that better sensors, ICs, and computers are improving traffic
management. It is probable that these and other improvements will have a much larger impact
on transportation than will improvements in batteries for electric vehicles This is because
batteries have experienced a slow rate of improvement of about 5% a year in energy storage
density29 and if these rates continue, their energy storage densities will not reach the 25 times
higher levels found in gasoline for at least 60 years, so it is very unlikely that battery-
powered cars with the range of existing cars will appear for many decades.
Of course some may argue that hybrid vehicles are sufficient or that these rates of
improvement might increase as scientists and engineers create new materials that have higher
energy and power storage densities; these are certainly a plausible future. We argue, however,
that these are less plausible futures than the ones we describe below. For the former, users
will always prefer a conventional vehicle over a hybrid vehicle since it is much cheaper. For
the latter, other technologies (including capacitors and very different types of batteries) are
experiencing more rapid rates than are batteries and furthermore batteries are not a new
technology; they have been used in vehicles for more than 100 years so acceleration in the
rate of improvement is unlikely.
Instead, we believe that improvements in the sensors and other technologies that were
22
discussed in previous sub-sections suggest more likely future transportation scenarios. First,
most electrical utilities are combining the Internet with the electrical grid to create smart
grids. One outcome of adding intelligence to our well established electrical grid can be the
capability of vehicles to easily find and purchase electricity from a high density of charging
stations in urban and suburban parts of developed countries. Since the cost of distributing
electricity is much lower than that of gasoline, the cost of the charging stations is probably
not as important as licensing large numbers of firms to sell electricity and thus overcoming
the network effects associated with the number of charging stations and electric vehicles.
Overcoming these network effects would enable electric vehicles to be charged while a
vehicle is parked in a parking garage or along a street30 in the future. This would enable the
vehicle to have far smaller storage capacities than are ordinarily thought and to avoid a
vicious cycle of heavier cars requiring more batteries and more batteries leading to heavier
cars.
Second, gradual improvements in the performance and cost of power electronics are
enabling the “electrification” of automobiles, which reduces the weight and thus the
necessary battery capacity of vehicles. This replacement of mechanical controls and drive
trains with electrical ones has already occurred in aircraft and heavy trucks and is now
occurring in automobiles as the cost of power electronics gradually falls. While the rate of
improvement is fairly slow (about 4% per year), announcements by automobile
manufacturers suggest that the extent of necessary improvements are very small. This
suggests that the electrification of vehicles will be largely finished within the next five to ten
years31 and this will reduce the need for large storage capacity in batteries.
Third, the implementation of densely packed systems of rapid charging stations are also
facilitated by the improvements in energy transmission performance that are coming from
improvements in superconductors. Superconductors are widely used in magnetic resonance
23
imaging and are beginning to be used in transformers, cables, fault current limiters, motors,
generators, and energy storage32. We can envision these superconducting transmission lines
providing extensive charging points throughout urban and suburban areas and even on major
highways through perhaps wireless charging. Wireless charging is also getting cheaper
through improvements in power electronics.
Fourth, improvements in cameras, MEMS, lasers, and wireless communication are making
autonomous vehicles economically feasible. With annual improvements rates of 25% to 40%
for many of the sensors, the cost of the controls for autonomous vehicles will probably drop
by 90% in the next ten years thus making autonomous vehicles not much different from
conventional vehicles. The largest benefits from automated vehicles will probably occur
when roads are dedicated to them and thus tightly packed vehicles can travel at high speeds.
Since fuel efficiencies drop as vehicle speeds drop, the use of dedicated roads for autonomous
vehicles can have a dramatic impact on fuel efficiency and road capacity, two common
problems in most urban and suburban settings. While many wonder whether humans will
give up control over their vehicles, we believe that this is the kind of emotional and intuition-
based argument that Daniel Kahneman warned us against. After all, most people have no
problem with allowing someone else to pilot an airplane, train, or ship as they travel. We
believe that autonomous vehicles will become economically feasible before the energy
storage densities of current batteries are doubled, which will probably take as long at the last
doubling (15 years). The bigger question is possibly the legal and governmental policy issues
associated with such a future.
Fifth, two alternatives to batteries, capacitors and flywheels, experienced faster rates of
improvement in energy storage density than did batteries until 2004 (10% for flywheels and
17% for capacitors33) but more recent data is not available and thus they are not shown in
Table 1. This suggests that one of them will probably eventually have higher energy densities
24
than do batteries for electric vehicles. Although capacitors have experienced faster rates of
improvement than have flywheels, flywheels are currently ahead of capacitors and they are
widely used in Formula 1 vehicles, partly because they have higher power densities than do
batteries. One of the reasons for the rapid improvements in the densities for flywheels is the
replacement of steel and glass with carbon fibers. Carbon fibers have higher strength to
weight ratios than do steel or glass and thus can rotate faster than can steel or glass-based
ones34. Rotational velocity is important because the energy storage density of flywheels is a
function of rotation velocity squared. Carbon nanotubes (CNT) have even higher strength-to
weight ratios than do carbon fibers and thus CNT-based flywheels can potentially have even
higher energy storage densities than do carbon-fiber based ones. Some estimates place the
strength-to weight ratios of CNTs at ten times higher than those of carbon fiber. This suggests
CNT-based flywheels can have an energy storage density that is ten times higher than that of
carbon fiber based flywheels and thus batteries.
This discussion highlights some of the problems with an “intuitive and emotional” process
of thought in which the importance of a technology is assessed by the ease of retrieving it
from memory. This encourages us to emphasize old technologies that have been around for a
long time and thus are well known to the media and public. In the case of batteries, this
attachment to an old technology (> 100 years old) occurs even when rates of improvement
are very slow. Furthermore, improving our transportation system by merely replacing one
technology with another is a severe limitation on innovation. It appears highly likely that
more innovative designs for our cities and other aspects of our life will be the future. This
paper’s emphasis on rates of improvement facilitates the creation of a much broader set of
ideas for how technologies can be combined into better transportation systems and into
systems that have a large chance of becoming economically feasible in the near future.
25
3.5 Energy Production
Cellulosic ethanol and solar cells may be the only clean energy technologies that are both
receiving attention from the media and that are experiencing rapid improvements. According
to the Intergovernmental Panel on Climate Change, the cost of electricity from wind turbines
only fell about 2% a year over the last 30 years, and increases actually occurred in the last
two years35. And this is in spite of the fact that wind turbines are still several times more
expensive than are conventional power plants and thus will probably never become a low-
cost source of electricity. For cellulosic ethanol, prices have almost reached the levels of corn
ethanol in the U.S. and the market growth depends more on the amount of ethanol that can be
blended with gasoline than the actual price.
Solar cells also have a high chance of becoming economically feasible in the next 10 to 20
years. Not only has the cost per peak Watt dropped quite rapidly over the last 50 years and
continues to drop36, rapid improvements have been made in the efficiencies of several types
of solar cells that are not yet cheaper than are crystalline silicon. For example, the efficiencies
of multiple junction ones with concentrators were improved from 32 to 44% between 1999
and 2013, those of organic ones from 3 to 11.1% between 2001 and 2012, those of quantum
dots from 3% to 7% between 2010 and 2012, and those of Perovskite cells from 3.5% to
almost 20% between 2009 and 2014.
Furthermore, organic and Perovskite solar cells (along with other types of thin film solar
cells) have potentially much lower costs per area than do crystalline silicon ones because they
use less materials, lower processing temperatures, and simpler processes (e.g., roll-to roll
printing). This suggests that at least one of these types will become cheaper than are
crystalline silicon cells as their efficiencies are improved and their levels of equipment scale
are subsequently increased. Many also believe that Perovskite solar cells have the potential
for efficiencies as high as single crystalline silicon solar cells due to their single crystalline
26
structure. The theoretical efficiencies for Quantum dots are even higher, approaching 80%
since a single quantum dot layer can be optimized for various wavelengths of sunlight. This is
because the optimal wavelength of light depends on the size of the Quantum Dot and thus by
placing many different sizes of Quantum Dots on a single layer, one can theoretically absorb
most wavelengths of light. This suggests that one or more of these new types of solar cells
will likely have much lower costs than will crystalline silicon as their efficiencies are further
improved and the scale of their production equipment reaches that of crystalline silicon. More
generally speaking, the large number of material classes that are experiencing improvements
increases the chances that one of them will become cheaper than conventional sources of
electricity.
Ironically, this long-term view towards solar energy is often ignored by clean energy
advocates and thus is excluded from the clean energy debate in general. Perhaps this is
because the debate is largely an emotional one in which both sides of the climate change
controversy over-state their cases. One side argues we need solar cells now while the other
side argues we don’t ever need them; this prevents a more deliberate and logical discussion
about solar cells and clean energy in general, a debate that requires data on rates of
improvements and knowledge about the details of different clean energy technologies. Many
clean energy advocates promote short-sighted adoption of crystalline silicon through demand-
based subsidies as opposed to long term R&D investments in thin film and other new types of
solar cells even as they criticize fossil fuel promoters for short-term thinking. Demand-based
subsidies encourage firms to produce the current low cost technology in order to get subsidies
and to do so before the subsidies are removed, which they always are. The policy of demand-
based subsidies is one result of ignoring rates of improvement and other aspects of deliberate
and logical arguments.
We are also optimistic about solar cells and possibly nuclear fusion because superconductors
27
are experiencing rapid rates of improvement. Using superconductors for the transmission of
energy can enable electricity to be generated from solar cells near the equator in for example,
North Africa or Mexico and then transmitted to high usage areas such as Europe and the
United States. Improvements in superconductors may also make fusion economically feasible
in the next 10-20 years, as argued by MIT37, since superconductors are the key component in
the main type of nuclear fusion and they are experiencing rapid improvements. However,
rather than fund expensive fusion projects, we should be funding more research on
superconductors: this will lead to the emergence of better superconductors for many
applications including those for fusion and do so cheaper than will directly funding expensive
fusion projects.
4. Discussion
This paper argues that we need to think more effectively about the future of technology and
that doing so requires us to move from what Nobel Laureate Daniel Kahneman calls an
“instinctive and emotional” to a “slower, more deliberative, and more logical” method of
analysis. The future is too important for us to assess the relative importance of technologies
by the ease with which they are retrieved from memory or by letting our judgments and
decisions about technologies be guided directly by feelings of “liking” and “disliking” as
most of us do, according to Kahneman’s research. It is easy for us to believe that certain
technologies are important because the media regularly discusses them or to like or dislike
certain technologies just based on feelings about them.
This paper argues that an important part of a “slower, more deliberative, and more logical.”
method of analyzing technologies is better data on rates of improvement and a better
understanding of their drivers. Technologies must provide some level of performance and
price for specific applications before they will begin to diffuse. Technologies that experience
28
faster rates of improvement are more likely to become economically feasible in the near
future than are other technologies. They are also more likely to become economically feasible
for an increasing number of applications and thus diffuse faster than other technologies.
This paper summarizes a number of technologies that are experiencing rapid rates of
improvements and it describes some plausible futures if these rates continue. We believe that
these futures are far more plausible and quite different from ones that are discussed by the
media, particularly with respect to clean energy. The discussion about clean energy is far too
often driven by the biases identified by Daniel Kahneman and others.
However, in describing these futures, we are not arguing that these are the only
technologies that are experiencing rapid improvements or that these “systems” are the only
ones that can be constructed from these technologies. Instead, we admit that our list of
technologies is far from complete and argue for more data collection, analysis, and
interpretation. There are many possible futures and the better data we have on technologies
experiencing rapid rates of improvement, their rates of improvement for multiple dimensions
of performance and cost, and the possible systems that can be constructed from these
technologies, the better our understanding of the possible futures of technology will be.
We also believe that a more informed debate about our technological future can help us
solve global problems such as sustainability and urban congestion and implement better R&D
policies. For the former, we can think of rapidly improving technologies as a kind of tool
chest that can be used to solve global problems. Not only does the current performance and
cost of these technologies provide us with useful tools here and now, their rapid rates of
improvement mean that better tools continue to emerge and we should be thinking about how
these better tools can help us solve global problems. This point was made in the discussion of
transportation where rapid improvements in ICs, sensors, the Internet, and superconductors
will probably have a larger impact on our transportation systems than will improvements in
29
batteries. This suggests that policy makers, firms, universities, and even students should be
thinking about a wider range of solutions than are currently considered and also thinking
about the policies and strategies needed to implement these solutions. These technologies will
probably require a different set of policies and strategies than will the technology that is most
often focused: electric or hybrid vehicles with the same range as conventional vehicles.
In terms of R&D policy, particularly for long-term research, one goal should be to fund
those technologies with rapid rates of improvement or with the potential for rapid rates of
improvement, since these technologies will have a larger impact on our world than will other
technologies. This paper provides us with both a list of technologies that are experiencing
rapid improvements and a method of identifying those technologies that are or will likely
experience rapid improvements. Since data on rates of improvement are not always available,
particularly for technologies that very new, understanding the reasons for rapid rates of
improvements can help us identify those technologies with the potential for rapid
improvements.
This paper argues that rapid improvements are driven by a subset of two drivers: 1) creating
new materials (and often their associated processes) to better exploit their underlying physical
phenomena; and 2) geometric scaling. Some technologies directly experience improvements
through these two mechanisms while those consisting of higher-level “systems” indirectly
experience them through improvements in specific “components.” First, creating new
materials (and processes for them) can lead to rapid improvements in performance and cost
when new classes of materials are continuously being created. Second, technologies that
benefit from reductions in scale (e.g., integrated circuits) have experienced much more rapid
improvements than have technologies that benefit from increases in scale (e.g., engines).
Understanding these drivers of rapid improvements can help us understand those
technologies that will likely experience rapid improvements and thus when they might
30
become economically feasible.
Finally, we believe that universities should play a more active role in collecting,
disseminating, and interpreting data on technological progress. In particular, social scientists
should take the lead in this effort and work closely with engineers and hard scientists to
collect better data on rates of improvement and the extent of improvements that are needed
before technologies become economically feasible. Doing so can enable universities to
propose better solutions to global problems than they are currently proposing. Furthermore,
we also believe that students can greatly benefit from being a participant in this data
collection and interpretation because this participation can help budding entrepreneurs
propose better products and services and more generally speaking help students think more
effectively about the future. This is important because they have the most at stake. We need to
give them the tools to think about and design their future, because the future is really their
future, not ours.
31
Table 1 Technologies with Recent Rapid Rates of ImprovementTechnologyDomain
Sub-Technology Dimensions of measure Time Period
Improvement Rate Per Year
Energy Trans-formation
Light Emitting Diodes (LEDs)
Luminosity per Watt, red 1965-2005 16.8%Lumens per Dollar, white 2000-2010 40.5%
Organic LEDs Luminosity/Watt, green 1987-2005 29%GaAs Lasers Power density 1987-2007 30%
Cost/Watt 1987-2007 31%Liquid Crystal Displays
Square meters per dollar 2001-2011 11.0%
Quantum DotDisplays
External Efficiency, red 1998-2009 36.0%
Solar Cells Peak Watt Per Dollar 1977-2013 13.7%Efficiency, Organic 2001-2012 11.4%Efficiency, Quantum Dot 2010-2013 42.1%Efficiency, Perovskite 2009-2013 46.5%
EnergyTransmission
Super-conductors
Current-length per dollar 2004-2010 115%Current x length - BSSCO 1987-2008 32.5% Current x length - YBCO 2002-2011 53.3%
InformationTrans-formation
Microprocessor Integrated Circuits
Number of transistors per chip/die
1971-2011 38%
Power ICs Current Density 1993-2012 16.1%Camera chips Pixels per dollar 1983-2013 48.7%
Light sensitivity 1986-2008 18%MEMS for Artificial Eye
Number of Electrodes 2002-2009 45.6%
MEMS Printing Drops per second 1985-2009 61%Organic Transistors Mobility 1984-2007 94%Single Walled Carbon Nano-tube Transistors
1/Purity 1999-2011 32.1%Density 2006-2011 357%
Super-conducting Josephson Junctions
1/Clock period 1990-2010 20.3%1/Bit energy 1990-2010 19.8%Qubit Lifetimes 1999-2012 142%Number of bits/Qubit lifetime
2005-2013 137%
32
Table 1 Technologies with Recent Rapid Rates of Improvement (continued)TechnologyDomain
Sub-Technology Dimensions of measure Time Period
Improvement Rate Per Year
InformationTrans-formation
Photonics Data Capacity per Chip 1983-2011 39.0%Digital Computers Instructions per unit time 1947-2009 36%
Instructions per kw-hour 1947-2009 52%Quantum Computers Number of Qubits 2002-2012 107%
Information Storage
Magnetic Storage Bits per unit cost, disks 1956-2007 39%Bits per unit area, disks 1956-2007 43%Bits per unit cost, tape 1994-2011 33%Bits per unit area, tape 1994-2011 34%
Flash Memory Storage Capacity 2001-2013 47%Resistive RAM Storage Capacity 2006-2013 272%Ferro-electric RAM Storage Capacity 2001-2009 37%Magneto RAM Storage Capacity 2002-2011 58%Phase Change RAM Storage Capacity 2004-2012 63%
Information Transmission
Last Mile Wireline Bits per second 1982-2010 48.7%Wireless, Cellular Bits per second 1996-2013 79.1%Wireless, WLAN 1995-2010 58.4%
Materials Trans-formation
Carbon Nanotubes 1/Minimum Theoretical Energy for Production
1999-2008 86.3%
Biological Trans-formation
DNA Sequencing per unit cost 2001-2013 146%Synthesizing per unit cost 2002-2010 84.3%
Cellulosic Ethanol Output per cost 2001-2012 13.9%
WLAN: Wireless Local Area Network; RAM: random access memory; MEMS: microelectronic
mechanical systems. Sources: 38
33
34
35
36
1 Daniel Kahneman, Thinking Fast and Slow, 20112 Renewable Energy Sources and Climate Change Mitigation: Special Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. 2013. Tarascon, J. 2009. Batteries for Transportation Now and In the Future, presented at Energy 2050, Stockholm, Sweden, October 19-20.3 Griliches Z 1957. Hybrid Corn: An Exploration in the Economics of Technological Change, Econometrica 25 (4), 501-522. Mansfield E 1968. The Economics of Technological Change, NY: W.W. Norton & Company Inc. 4 One of the author’s slide share accounts http://www.slideshare.net/Funk98/presentations has more than 50 presentations on specific technologies by both the author and students from a course devoted to when new technologies become economically feasible. Also see an Amazon eBook by the authors, Exponential Change: What drives it? What does it tell us about the future? 5 Funk J 2013a. What Drives Exponential Improvements? California Management Review Spring. Funk 2013b. Technology Change and the Rise of New Industries, Stanford University Press. Funk J and Magee C. 2014. Exponential Change: What drives it? What does it tell us about he future? Amazon Kindle eBook.6 Funk, 2013a, Funk, 2013b7 Chandler, A 1994, Scale and Scope: The Dynamics of Industrial Capitalism, Boston: Belknap. 8 Axelrod L., Daze R. and Wickham, H. 1968. The large plant concept, Chemical Engineering
Progress 64(7): 179 Service R 2013. Battle for the Barrel. Science 339: 1374-1379. 10 Funk, 2013b11 Roll-to-Roll Manufacturing of Flexible Displays, Hewlett Packard, Phicott12 Source: http://www.electroiq.com/articles/sst/2012/02/led-manufacturing-highlights-from-strategies-in-light-day-2.html 13 ICKnowledge,2009. www.icknowledge.com. last accessed 12/7/2009 14 Clark Ngyuen, August and September 2011 Berkeley lectures15 Sanket Goel, Presentation at National University of Singapore, Bio-electronics: Changes and Opportunities, 3 March, 201116 Ryan D, Rahimi M, Lund J, Mehta R, Parviz BA, 2007. Toward nanoscale genome sequencing. Trends in Biotechnology 25(9):385-9. 17 The SSD Review, April 29, 2013. http://thessdreview.com/daily-news/latest-buzz/flash-memory-to-be-based-on-2d-materials-a- single-atom-thick/ http://edition.cnn.com/2013/04/29/tech/graphene-miracle-material/index.html?hpt=hp_c3 18 Flamm, K 1988. Creating the computer: government, industry, and high technology, Brookings
Institution: Washington DC.19 Smith R 1988. A Historical Overview of Computer Architecture, IEEE Annals of the History of Computing 10(4): 277-303. 20 Koh, H. and Magee, C. 2006. A functional approach for studying technological progress: Application to information technologies, Technological Forecasting and Social Change 73: 1061-1083. 21 Preil M 2012. Optimum Dose for EUV: Technical vs. Economic Drivers, Future Fab International (41): April 26. http://www.future-fab.com/documents.asp?d_ID=4926. Cortada J.
2003. The Digital Hand: How Computers Changed the Work of American Manufacturing, Transportation, and Retail Industries, NY: Oxford University Press.
22 iSuppli 2013. http://www.isuppli.com/teardowns/pages/headlines.aspx, last accessed April, 26, 2013
23 Kalender W. 2006. X-ray computed tomography, Physics in Medicine and Biology 51: 29-53. Trajtenberg, M. 1990. Economic Analysis of Product Innovation - The Case of CT Scanners. Cambridge, Mass: Harvard University Press.
24 MacKie-Mason, J. and Varian, H. 1994. Economic FAQs about the Internet, American Economic Association 8(3): 75-96. Koh, H. and Magee, C. 2006..
25 Garrard, G. 1998. Cellular Communications: Global Market Development, Artech House. Gonzalez M, 2010. Embedded Multicore Processing for Mobile Communication Systems, accessed May 16, 2012. ruhr-uni-bochum.de/integriertesysteme/emuco/files/hipeac_trends_future.pdf.
26 Franklin A 2013. Electronics: The road to carbon nanotube transistors, Nature 498, 443–444. Shulaker et al 2013. Carbon Nanotube Computer, Nature 501: 526-530.27 Mayer-Schonberger V and Cukier K 2013. Big Data: A Revolution That Will Transform How We Live, Work, and Think, NY: Houghton Mifflin28 http://www.idc.com/prodserv/FourPillars/bigData/index.jsp29 Tarascon, J. 2009. Batteries for Transportation Now and In the Future, presented at Energy 2050, Stockholm, Sweden, October 19-20. Subsequent improvements are also consistent with an annual rate of 5%.30 Huber P 2011. Broadband Electricity and the Free-Market Path to Electric Cars, Manhattan Institute for Policy Research.31 http://cesa-automotive-electronics.blogspot.sg/2012/09/dual-voltage-power-supply-system-with.html 32 richard.grisel.free.fr/ICEM2012/TUTORIALS/TUT5.pdf 33 . Koh, H. and Magee, C. 2008. A functional approach for studying technological progress: Extension to energy technology, Technological Forecasting and Social Change 75: 735-758.34 Renewable and Sustainable Energy Reviews 11(2007): 235-258 35 Renewable Energy Sources and Climate Change Mitigation: Special Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. 201336 The economist claims 21.4% over the last ten years but only 13.7% over the last 36 years. This data is consistent with the data from Gregory Nemet (2006. Beyond the learning curve: factors influencing cost reductions in photovoltaics, Energy Policy 34: 3218 - 3232), who claims 16% between 1957 and 2003.37 http://nextbigfuture.com/2013/05/10-tesla-superconductors-could-enable.html38 Azevedo I, Morgan G, Morgan F, 2009. The Transition to Solid State Lighting, Proceedings of the IEEE 97: 481-510. http://www.globalsmtseasia.com/index.php?option=com_content&view=article&id=4228:printed-electronics-for-flexible-solid-state-lighting&catid=109&Itemid=115. Lee, C 2005. OLED 1 – Introduction, http://wenku.baidu.com/view/783fa93283c4bb4cf7ecd196.html Sheats, J, H Antoniadis, M Hueschen, W Leonard, J Miller, R Moon, D Roitman, A Stockinget, Organic Electroluminescent Devices, Science 273 (1996): 884-888. Martinson R 2007. Industrial markets beckon for high-power diode lasers, Optics, October: 26-27.ww.nlight.net/nlight-files/file/articles/OLE%2010.2007_Industrial%20markets...pdfTelevision Making: Cracking Up, Economist, January 21st, 2012, p. 66; Kwak J 2010. PhD Thesis. Cited in Changhee Lee, Seoul National Universityhttp://www.andrew.cmu.edu/org/nanotechnology-forum/Forum_7/Presentation/CH_Lee.pdfhttp://www.economist.com/news/21566414-alternative-energy-will-no-longer-be-alternative-sunny-uplands. National Energy Research Labs, http://en.wikipedia.org/wiki/File:PVeff(rev131204)a.jpg. Perovskite-based solar cells, Hodes G, science 342, 317 (2013, Oct): 317-318. Selvamanickam V 2011. Second-generation HTS Wire for Wind Energy Applications, Symposium on Superconducting Devices for Wind Energy, 25 February, Barcelona, Spain. Shiohara T, Yoshizumi M, Takagi Y, Izumi T 2013. Future prospects of high Tc superconductors-coated conductors and their applications, Physica C: Superconductivity 484, 15 January 2013: 1-5. en.wikipedia.org/wiki/Moore's_law, last accessed on January 30, 2013. ISSCC, 2013. International Solid State Circuits Conference, ISSCC Trendshttp://www.embedded.com/design/components-and-packaging/4371098/New-power-semiconductor-technologies-challenge-assembly-and-system-setupshttp://www.future-fab.com/documents.asp?d_ID=4926Suzuki T, 2010, Challenges of Image-Sensor Development, Int’l Solid State Sensors
Conference. Stasiak J, Richards S, and Angelos S 2009. Hewlett Packard's Inkjet MEMSTechnology, http://144.206.159.178/ft/CONF/16431771/16431793.pdf; Proc. of Society ofSPIE: 7318. Shaw J and Seidler P 2001. Organic electronics: Introduction, IBM Journal of R&D; 45(1): 3-9. Dong H, Wang C, Hu W 2010. High Performance Organic Semiconductorsfor Field-Effect Transistor, Chemical Communications 46: 5211-5222. Franklin A 2013.Electronics: The road to carbon nanotube transistors, Nature 498, 443–444. www.istec.or.jp/web21/pdf/12_Winter/E15.pdf; Devoret M and Schoeldopf R 2013. Superconducting Circuits for Quantum Information, Science 339, March 8: 1169-1174. EvansP et al 2011. 1.12 Tb/s superchannel coherent PM-QPSK InP transmitter photonic integratedcircuit (PIC) Optics Express 19(26): B154-B158. Computers: Implications of Historical Trends in the Electrical Efficiency of Computing, July-September 2011 (vol. 33 no. 3), pp.46-54, J Koomey, S Berard, M Sanchez, H Wong; http://nextbigfuture.com/2013/05/ dwave-512-qubit-quantum-computer-faster.html;http://lib.stanford.edu/files/pasig-jan2012/11B7%20Francis%20PASIG_2011_Francis_final.pdf; Yoon Y 2010. Nano-Tribology of Discrete Track Recording Media, Unpublished PhD Dissertation, University of California, San Diego. Brown R 2011. Staying with Copper in a Fiber World, Rural Telecom, March-April: 22-26. International Solid State Circuits Conference 2013 (http://isscc.org/doc/2013/2013_Trends.pdf). Minimum Exergy Requirements for the Manufacturing of Carbon Nanotubes, Timothy G. Gutowski, John Y. H. Liow, Dusan P. Sekulic, IEEE, International Symposium on Sustainable Systems and Technologies, Washington D.C., May 16-19, 2010. NHGRI: http://www.genome.gov/sequencingcosts/; http://singularityhub.com/2012/09/17/new-software-makes-synthesizing-dna-as-easy-as-using-an-ipad/; Service R 2013. Battle for the Barrel. Science 339: 1374-1379.