11

CROSSCUTTING THEMES
AND COMMONALITIES

Chapters 1 through 10 addressed ten individual functions—addressing problems in novel ways and
technology areas. This chapter pulls together some solving challenges that people did not even know
common themes that cut across them. Of course, they had. In the first half of the twentieth century, for
the scientific issues at play are different because example, polio afflicted thousands of people world-
the science is different. However, there are impor­ wide. The iron lung was invented in the 1930s to
tant similarities in how people and institutions make help polio victims breathe, and over the next twenty-
progress that are often lost when each field is con- five years, improvements were made to the iron
sidered in isolation. lung.1 But the groundbreaking Salk vaccine in 1955
brought an entirely different way to defend people
against polio. Within a few years, use of the iron lung
The Value and Risk of dropped to nearly zero.
Technological Progress
Manufacturing provides another example. For
Takeaway Innovation that emerges too fast threat- decades, large-scale manufacturing has relied on
ens the legitimate interests of those who might be the idea of an assembly line to fabricate essentially
negatively affected by such innovation, while inno- identical models of the same product. Workers were
vation that moves too slowly increases the likelihood originally all human. Then robots began replacing
that a nation will lose first-mover advantages. them, performing many assembly-line tasks more
rapidly and accurately while reducing production
New technologies typically bring two types of ben- costs. In the past two decades, a complemen-
efits. First, they can enhance or improve existing tary fabrication paradigm has emerged: custom,
processes. Second, they can enable entirely new on-demand manufacturing of products in small

128
quantities using 3-D printing, or what’s known as Furthermore, it may well be that only upon delivery
additive manufacturing. This new paradigm enables of new products do other risks become apparent,
production that is far more localized and custom- with innovators facing issues of ethics and equity,
ized, though it does not yield the economies of privacy, and increased challenges to health, safety,
scale that mass production offers. and security—all risks that could lead to an erosion
of trust in their services or capabilities.
Technological progress also brings risks—risks of
moving too fast or too slowly. Innovation that emerges
too fast threatens to disrupt the often-delicate bal- The Central Importance of Ideas and
ance that has been established among many national, Human Talent in Science and Technology
organizational, and personal interests. As we are
seeing today with AI, the rush to deploy new capa- Takeaway Human talent plays a central role
bilities may give short shrift to issues such as safety, in generating the ideas for innovation, it can be
security, employment, values, ethics, societal impact, found all over the world, and it cannot be manu-
and geopolitics. On the other hand, innovation that is factured at will.
too slow increases the likelihood that a nation will lose
the technical, economic, and national security advan- From time to time, lone scientists working on their
tages that often accrue to first movers in a field. In own achieve breakthroughs on very difficult prob-
both cases, policy measures are often needed to steer lems. But it is far more common that successful sci-
outcomes in a more optimal direction. ence and technology (S&T) innovations are a result of
a well-functioning collaborative effort that can bring to
The road from scientific discovery to useful appli- bear a broad range of cognitive styles and disciplinary
cation is often rockier than expected as well, with expertise.2
would-be innovators finding that the realization
of the benefits promised to investors and cus- Scientific progress obviously benefits from new
tomers actually entail greater costs, deliver fewer ideas. New ideas are created every day by talented
capabilities, and take more time than anticipated. Americans, but Americans do not have a monopoly

FIGURE 11.1 Number of patents per year

70,000

60,000

50,000

40,000

30,000

20,000

10,000

0
1996 2001 2006 2011 2016 2021

US China

Source: World Intellectual Property Organization, “WIPO IP Statistics Data Center.”

11 Crosscutting Themes and Commonalities 129

 on the creation of new ideas. As one metric, con- In fact, the same point—that ideas move much
sider that China’s production of patents reached more effectively when conveyed through people
parity (100 percent of US patent production) around than through papers—underscores how the United
2015 and (as of 2018) surpasses that of the United States actively benefits from foreign scientific input.
States (see figure 11.1). Even allowing for the pos- Skilled immigrants support American innovation
sibility that historically, Chinese patents might be today. For example, immigrant college graduates
of generally lower quality than those of the United receive patents at double the rate of native-born
States, the trend line is clear. Americans.5 In part, this is explained by a higher
proportion of immigrant students pursuing STEM
Other nations are also investing heavily in research education in the United States. Technology compa-
and development (R&D), which is a critical source nies in the United States also rely on immigration
of new ideas. According to the National Science visas to bring foreign scientists and engineers to
Board, most of the world’s R&D expenditures occur work in the United States.
in a few countries, with the United States account-
ing for 27 percent of global R&D in 2019, followed Conversely, the United States suffers when foreign
by China (22 percent), Japan (7 percent), Germany scientific inputs are curtailed. For instance, a Harvard
(6 percent), and South Korea (4 percent).3 At the Business School study found that pro-migration
same time, the concentration of R&D expenditures changes to immigration policy significantly increase
continues to shift from the United States and Europe innovation within a country—as measured by the
to countries in East, Southeast, and South Asia. production of patents—while changes that dis-
This trend is consistent with the observation that an courage immigration lead to significant declines in
increasing share of the world’s patents are shifting patent production.6
over to Asia—particularly to China.4
Human talent capable of creating ideas in S&T
How can the United States take advantage of ideas cannot be manufactured at will. It must be domes-
produced in other countries? One obvious way is to tically nurtured or otherwise imported from abroad.
read the scientific and technical literature produced by Today, both paths to growing the requisite talent
scientists abroad, and that does happen in abundance. base to sustain and grow US innovation face seri-
But it is well known that people are a much more effec- ous and rising challenges. Test scores clearly show
tive information transfer mechanism than papers. declining performance in STEM subjects in K–12
education, both in absolute terms and in compari-
For example, informal interviews with Stanford fac- son with other countries.7
ulty across most of the technology areas yielded
two important points regarding the value of direct, Regarding US STEM education, the US Department
in-person interactions with foreign scientific col- of Defense noted in 2021 that improving the capacity
leagues. First, these interactions enable them to and resilience of the defense industrial base requires
learn things they could not learn simply from read- more workers trained in the skilled trades and in
ing papers published by the same people, as papers STEM.8 Yet bias against careers in the industrial
often do not capture vital “tacit knowledge” that trades among parents and educators has shrunk the
enables researchers to build upon the work of others. pool of potential workers, and adverse demographic
Second, they are able to develop a much better trends have led to an aging-out of a skilled work-
understanding of the scope and nature of progress force with irreplaceable knowledge. The Defense
made and not made by their foreign colleagues— Department also noted the dearth of trained soft-
an understanding that would not result simply from ware engineers working on classified projects was in
reading the literature. part because they must be US citizens.

130 STANFORD EMERGING TECHNOLOGY REVIEW

At the same time, US immigration policies dis- A report from Korn Ferry Consulting asserts that
courage or prevent foreign S&T talent from work- “human capital is the greatest value creator available to
ing here. Current policies are facilitating a shift of organizations,”13 also finding that every dollar invested
skilled immigration and associated multinational in human capital adds $11.39 to GDP. The same holds
R&D investment toward other countries. Tightening true for advances in science and technology more gen-
immigration also can prevent companies from hiring erally: the most important ingredient for S&T advances
enough skilled workers to operate their R&D facili- is human talent because human talent is the goose that
ties, increasing their incentives to relocate abroad.9 lays the golden eggs of ideas and innovation.

A significant portion of academic researchers are None of these comments are intended to suggest that
PhD students and professors who have immigrated concerns about foreign appropriation of American
to the United States to seek better educational and intellectual efforts are unfounded. But the fact that
research opportunities. It is crucial to establish a American S&T efforts are deeply connected to those
better pathway to permanent residence upon grad- of the rest of the world is overall an accelerator of
uation for PhD students on student visas so that the those efforts rather than a brake on them. Using an axe
United States does not lose highly trained workers. to impose blanket restrictions on engaging foreign sci-
The United States and universities invest heavily in entists when a surgical scalpel is needed to curb only
the education of STEM graduate students, and it the issues that warrant serious concern is a sure way to
would be wise to find a path to allow these scientists reduce the effectiveness of US scientific efforts.
and engineers to work and live in the country per-
manently. Furthermore, if ambitious goals in build-
ing up the semiconductor industry, biotechnology, The Changing Role of Government
or decarbonization are to be met, then increased regarding Technological Innovation
investment is needed in the labor force. These indus-
tries hire highly trained workers who have advanced Takeaway The US government is no longer
degrees. Research funding supports not only the the primary driver of technological innovation or
scientific outcomes but also an essential method for funder of R&D.
training highly skilled engineers and scientists.
Many technological advancements—such as satel-
Finally, it is important to realize that the global talent lites and access to space, the development of jet
challenge is not just about China. US allies and part- engines, and the emergence of the semiconductor
ners also compete for technology talent from around industry in Silicon Valley—have their roots in US
the world. For example, Canada has always had an government financial support and advocacy. But in
immigrant-friendly policy that attracts foreign-born many fields today, the US government is no longer
graduates of US universities—nearly forty thousand the primary driver or funder of R&D.
such individuals were recruited to Canada from 2017
to 2021.10 More recently, Canada introduced its Tech Private companies have taken up much of the slack.
Talent recruiting program in June 2023.11 This pro- For example, while the US government once used
gram targets tech workers in the United States who its own rockets to launch satellites, it now often
hold US H-1B nonimmigrant visas, providing them does so by contracting with companies that provide
more favorable terms. A 2022 survey of almost access to space as a service. These companies, how-
1,500 global leaders hiring tech professionals in the ever, may be under the jurisdiction of nations or con-
United States, United Kingdom, Germany, France, trolled by senior executives whose interests are not
the Netherlands, and Sweden found that more than aligned with those of the users of their services. For
one-third were searching globally.12 example, for a several-year period in the mid-2010s,

11 Crosscutting Themes and Commonalities 131

 the United States was entirely dependent on Russia of individual actors. Export controls may delay this
to transport American astronauts to the International spread in some cases for a limited time, but the
Space Station. More recently, the Starlink satellite overall trend is toward decentralized access.
communications network has been an essential part
of Ukrainian battlefield communications; however, This trend has several implications:
the CEO of Starlink curtailed Ukrainian access on a
number of occasions in ways that affected Ukrainian ° Greater policy complexity results from more
battlefield strategy.14 Such concerns are most serious actors. Other actors, both state and nonstate,
when there is only one or a small number of private- will have capabilities to challenge US interests
sector providers of the services in question. that they did not have before.

Many US officials recognize the growing role of a ° Technological advantages will diminish. The
United States may have the most technologically
handful of private actors behind influential innova-
advanced capabilities, but even the more rudi-
tions in technology. They believe that supporting
mentary instantiations of these capabilities avail-
closer public-private sector cooperation, as well as
able to other actors eliminate monopolies and
informed government regulation of emerging tech-
narrow the relative advantages once enjoyed by
nologies, is a pressing imperative. Even if the gov-
the United States.
ernment cannot rely on its own capabilities to remain
at the forefront of technological innovation, it still has
an important role to play in funding and promoting
° Winning isn’t winning anymore. The old par-
adigm of “winning” a technological race to
R&D, facilitating the broad adoption of key innova-
achieve gains that last decades and are hard to
tions and standards, and convening coalitions of like-
replicate—traveling to the moon, developing the
minded actors both domestically and internationally.
atomic bomb—will be replaced by the paradigm
of constant competition.
A Trend toward Increasing Access to
New Technologies Worldwide ° More diversity in bureaucracy and ethics has
consequences. Actors less subject to bureau-
Takeaway National monopolies on technology cratic and ethical constraints will be able to
are increasingly difficult to maintain. Even inno- exploit technology more nimbly and adapt more
vations that are exclusively American born (an rapidly to conditions on the ground.
increasingly rare occurrence) are unlikely to remain
in the exclusive control of American actors for On the other hand, for physical technologies whose
long periods. effectiveness depends on deployments in large
quantities, geography still plays a role. Natural
Access to technologies such as synthetic biology, resources such as rare-earth metals are geographi-
robotics, space, and blockchain often spread from cally constrained, and production facilities for physi-
rich nation states and large corporations to less cal artifacts still matter.
wealthy nations, smaller corporations, universities,
and individuals. Even innovations that are American It may be possible to extend periods of American
born—an increasingly rare occurrence—are unlikely monopoly on certain technologies, but not indefi-
to remain in the exclusive control of American actors nitely. Such extensions can have valuable short-term
for long. Many emerging technologies exhibit a benefits, not least because they buy time for US
long-term trend of riding a declining cost-curve over policy­makers to better anticipate a world of democ-
time, making them accessible to an ever-larger set ratization. But all too often buying time becomes an

132 STANFORD EMERGING TECHNOLOGY REVIEW

end unto itself, and actions to craft a better policy— in technology A can be used to improve the per-
such as improving targeted immigration reform and formance of technology B, while improvements in B
sustainability—are not taken. help C— then C and B together can help improve A.

To be sure, there are probably exceptions to this For example:

democratization trend. One is the first appearance
of an emerging technology. At such a point in time, ° AI contributes to advances in synthetic biology
the democratization process has not yet begun, at in addressing the protein-folding problem, pre-
least not in full force, and it may indeed be that the dicting protein shape from the DNA sequence of
technology in question will be characterized by the base pairs.
dominance of a few key actors.
° AI helps to screen many candidate compounds to
For example, the training of large language models predict the ones most likely to exhibit desirable
(LLMs) from scratch is a new capability still only properties for materials science.
belonging to a few large companies, and it is com-
pletely out of reach even for large coalitions of top ° Materials science is central to the identification of
research universities. On the other hand, research new semiconductors that may be useful in devel-
is already underway to build applications based on oping more energy-efficient chips, which in turn
foundational models at much lower cost. In many can reduce the cost of training AI models.
cases, the approach taken is to fine-tune a foun-
dational model that has already been trained from ° New materials are important in space research for
scratch, thus building on previous efforts. Therefore, the construction of spacecraft and satellites.
it may well be that this particular exception is appar-
ent only because of the quirk of timing, and LLMs ° New materials are needed to enable the develop-
will also be further democratized as time goes on. ment of neural probes that can send and receive
electrical signals in neural tissue.
In other cases, the impact of emerging technologies
depends on the scale of deployment. Universities ° Energy technologies help to improve the perfor-
may be able to develop sustainable energy tech- mance of robotics and spacecraft.
nologies such as better batteries, but they lack
the infrastructure to manufacture them at scale, an ° Synthetic biology can build organisms that pro-
enterprise that requires enormous investments from duce certain specialized materials.
the private sector.
° Cheaper semiconductors have driven down the
cost curve of DNA sequencing, which itself is a
Synergies between Different Technologies
fundamental technology for synthetic biology.15

Takeaway The synergies between different

technologies are large and growing as advances in Some of the S&T areas in this review—AI, synthetic
one technology often support advances in other biology, materials science, and energy—have a
technologies. foundational flavor impacting a variety of problem
or technology domains.16 Others are better char-
The technologies described in this report span a acterized as technology applications—space and
broad range, but most have in common a synergis- robotics, for example—which focus on solving spe-
tic relationship to other technologies. Improvements cific problems through an artful blend of a number

11 Crosscutting Themes and Commonalities 133

 of technologies. But these are differences of degree, understanding of a real-world phenomenon—what
not of kind, and even more foundational technolo- some call “use-inspired basic research.” One of the
gies can benefit from advances in different technol- most famous examples is the work of the French
ogy applications. chemist Louis Pasteur on milk spoilage, an applied
problem that required advancements in fundamen-
tal biological science regarding bacterial processes.
Nonlinear Paths from Research
to Useful Application
In other cases, deep fundamental scientific knowl-
edge may be necessary in technology areas adja-
Takeaway The traditional “linear” model of R&D,
cent to the primary problem of interest. New drugs
in which basic research leads to applied research,
that are effective against cancer tumors do no good
which then leads to development and prototyping,
if they cannot be delivered in lethal concentration
which finally leads to novel and useful products or
to the tumor. Therefore, research on drug delivery
services, is only one model for how societies obtain
mechanisms is as important here as the develop-
value for investments in technology innovation.
ment of new anticancer compounds.

The traditional “linear” model of R&D in science

and technology starts with basic research leading to The Relationship of Political Regime Type
applied research, which then leads to development to Technological Progress
and prototyping, which finally leads to marketable
products. Takeaway Democracies provide greater free-
dom for scientific exploration, while authoritarian
° Basic research is activity aimed at fundamental regimes can direct sustained funding and focus to
scientific understanding without any particular technologies they believe are most important.
applications in mind.
Technological innovation occurs in both democra-
° Applied research is activity to deepen this scien- cies and autocracies, but different regime types face
tific understanding with an application area or different advantages and challenges. True democra-
specific problem in mind. cies enjoy the rule of law and a free flow of ideas and
people, as well as the ability of individuals to pursue
° Development is activity that builds on scientific research goals of their own choosing. Perhaps most
understanding to construct engineering proto-
important, because failure in a democracy does not
types and proofs-of-concept.
lead to persecution or necessarily result in profes-
sional ostracism, individuals are freer to experiment
This model of scientific development has a long his- and explore.
tory, but it is by no means the only path. Many ana-
lysts argue this model is so unrepresentative of how Authoritarian regimes are more aptly characterized
scientific development actually proceeds that it can by the rule of the state, or the whim of a single
be harmful to the scientific enterprise. “supreme leader.” This can lead to the constrained
flow of ideas, coercion to limit individual freedom
Other models are less linear in nature; they acknowl- of action and thought, and top-down direction to
edge and even exalt the need for feedback between explore only topics of interest to the state. In this
the various activities. For example, some problems environment, failure may carry very high conse-
or application areas are so challenging that they quences for individuals. Under such circumstances,
entail obtaining a deeper fundamental scientific it would be understandable if individual scientists

134 STANFORD EMERGING TECHNOLOGY REVIEW

limit themselves to studying “safe topics” for which of change is hard even for leading researchers to
failure to make progress is unlikely. anticipate.

Yet it must be noted that authoritarian regimes have Taken as a whole, technological progress exhibits a
the advantage of being able to direct funding and variety of patterns. Some technologies have demon-
public attention to problems that they believe are strated consistent progress for extended periods. For
important. They can sustain that focus for long peri- example, semiconductor technology is characterized
ods of time more independently of short-term con- by Moore’s law, an exponential reduction in the cost
siderations such as profit or politics. To the extent of semiconductors over time. Solar cells and LED effi-
that technology-based solutions are known, author- ciency have followed similar cost reduction curves.
itarian leaders can exploit that knowledge to imple-
ment those solutions, regardless of any downsides. Other technologies have demonstrated much more
uneven rates of progress. These technologies see
Successful innovation requires both an exploration long periods of incremental development and
of the relevant space of possible solutions, to elim- refinement that are punctuated by short bursts of
inate paths that will not lead to viable solutions, radical innovation. In some cases, the bursts are
and an exploitation of viable solutions to focus the result of some particular breakthrough—exam-
resources on specific problems deemed important ples might be the emergence of the personal com-
to the regime.17 Competitors such as China have puter in the 1980s or the World Wide Web in the
taken advantage of US scientific exploration in many early 1990s. In other cases, the bursts are due to
domains through means both legal and illegal and the simultaneous availability and maturity of several
have gone on to exploit that knowledge through a key technologies that are required to make signifi-
variety of commercial and military efforts. cant progress in some other technological domain.
Here, an example might be electric cars, where
Attempts to obtain some of the benefits of a more battery technology, lightweight materials, sensors,
centralized direction to the technology policy efforts and computing power have come together to make
of the United States have been described as steps in such cars more economically feasible.
the direction of adopting an industrial policy. Critics
often argue that such efforts unduly interfere in a
When punctuated progress characterizes a technol-
free market and that picking “winners and losers”
ogy, forward projections of progress based on past
leads to inefficiencies. Advocates argue that only
rates may well be misleading. Successful forecasting
through such action will the United States be able
depends on familiarity with a wide variety of technol-
to offset some of the advantages that authoritarian
ogies precisely because it is hard to predict which
nations would otherwise enjoy over it. The public
specific technologies will prove critical. Indeed,
policy problem is acknowledging some truth in both
even experts in a given field can be surprised by
perspectives and seeking an appropriate balance
the rapidity of progress, as has happened in the last
consistent with both American values and economic
year with artificial intelligence and applications such
competitiveness.
as ChatGPT and large language models. Geoffrey
Hinton, a 2018 Turing Award winner for his work on
Punctuated Technological Progress artificial intelligence, said, “I have suddenly switched
my views on whether these things are going to be
Takeaway Technology often progresses in fits more intelligent than us. I think they’re very close to
and starts, long periods of incremental results are it now and they will be much more intelligent than
followed by sudden breakthroughs, and the speed us in the future.”18 This sentiment is shared by other

11 Crosscutting Themes and Commonalities 135

 experts in LLMs, who have told us they too have been Scientific proof-of-concept is only the first step.
stunned by the speed of advances in their own field. Engineering feasibility must also be demonstrated,
which includes considerations of cost and ease of
While there is broad recognition that we may be at use. Take the case of the technical success of early
the cusp of a moment of radical technological change attempts to build supercomputers out of supercon-
across a number of fields (AI, synthetic biology, nuclear ducting components that required liquid helium to
fusion), the precise contours, speed, and implications cool them. Technical feasibility was demonstrated,
of this moment are much harder to ascertain. but the liquid helium requirement would make these
computers difficult to deploy and use in practice;
alternative computing technologies appeared to
Nontechnological Influences on
offer comparable performance at lower cost.
Technological Innovation
In other cases, engineering feasibility can be demon-
Takeaway Technology applications in society
strated but cost considerations must first be resolved.
require scientific proof-of-concept, engineer-
For example, when carbon fibers were first being
ing feasibility, economic viability, and societal
investigated in the laboratory, they cost $10 million
acceptability.
per pound—clearly infeasible for large-scale use.19 A
substantial amount of work was required to reduce
Technology plays a key role in supporting and advanc- the cost by what is today a factor of a million.
ing national interests, and headlines in the news often
tout scientific breakthroughs that offer opportunities In still other cases, it may prove too difficult to
to solve societal problems and to improve the quality develop a manufacturing process to build a device
of life. But to play a valuable role, any given technol- based on Q, or the materials used to demonstrate Q
ogy application must demonstrate not only technical are too expensive or rare to support large-scale pro-
feasibility but also economic viability. It must also be duction. Less expensive or cumbersome alternatives
acceptable to the relevant constituencies, including to devices based on Q may be available, thus reduc-
the public at large. People and organizations must be ing the marketplace viability of Q-based devices.
able to adapt to its use, despite the disruption it may
cause. The requirements and burdens imposed by Societal acceptability matters as well. In Europe,
law, policy, and regulation must be compatible with though much less so in the United States, genetically
widespread adoption and use of the application. modified organisms as food are highly controver-
sial, and concerns over their safety have prevented
There is often a large gap between a demonstration the uptake of GMO foods consumed widely in the
of scientific feasibility—let’s call it Q—and a prod- United States. The psychology of individuals and
uct or a service based on Q that is useful to soci- cultural practices and beliefs of a community or soci-
ety. Press reports of scientific breakthroughs often ety also contribute to the adoption and use of any
give the impression that useful exploitations of these given technology application. The essential point
breakthroughs are just around the corner. That is here is that technology in society is not just about
almost never true. Scientific feasibility is a necessary the technology.
prerequisite, but it may well be that engineering or
economic feasibility does not follow. After scientific Lastly, given that some technological demonstra-
proof-of-concept is achieved, engineering feasibil- tions of scientific feasibility do not advance to the
ity must be demonstrated, which includes consider- marketplace and become “orphaned,” an impor­
ations of cost and ease of use. tant public policy question is how to manage them.

136 STANFORD EMERGING TECHNOLOGY REVIEW

For example, a start-up company may be estab- The federal government also operates a large number
lished to commercialize Q. If the company fails in of laboratories and federally funded research and
the market­place for economic reasons, a competitor development centers. For example, the Department
or another nation with a different cost structure may of Energy operates seventeen national laborato-
be able to make it economically viable—and that ries for conducting research and development that
competitor’s interests may not align with those of serve the department’s core missions in energy, sci-
the United States. What if the competitor is a bad ence, national security, and environmental steward-
actor and simply buys the now-defunct start-up, ship.22 These labs specialize in particularly difficult
thereby acquiring the rights to the intellectual prop- problems that fall beyond the capabilities of pri-
erty underlying Q? vate industry or individual universities. (The fusion
breakthrough described in chapter 6 was conducted
at the Lawrence Livermore National Laboratory.)
The Role of Universities in Tech Innovation Nor is the Department of Energy unique in having
mission-driven laboratories that actually conduct
Takeaway US universities play a pivotal role research and development work; the Department of
in the innovation ecosystem that is increasingly Defense, NASA, the National Institutes of Health, and
at risk. the Department of Commerce are just a few of the
departments with their own mission-driven labs.
The US infrastructure for funding and conduct-
ing R&D (to which innovation is closely related) is But the role of universities is unique and pivotal
broad and deep. For example, the private sector is in the innovation ecosystem, and this role is often
the second-largest supporter of R&D in the United underappreciated. For example, in contrast to
States (the first is the federal government).20 Entities research done in mission-driven federal laboratories,
such as Bell Laboratories, IBM’s Thomas J. Watson the scope and breadth of research conducted at
Research Center, and Xerox PARC once performed major research universities tends to be much larger
substantial amounts of basic scientific research. than that conducted at federal labs simply because
Today, their present-day equivalents focus most of the various foci of university research are not con-
their R&D efforts on process and product devel- strained by particular missions.
opment closely related to the bottom line of their
parent companies. This focus has two important In contrast to the private sector, university research
implications. First, companies tend to focus their is almost always open, enabling would-be innova-
efforts on research with foreseeable commercial tors to take advantage of it. Open research pro-
application, not frontier or fundamental research motes transparency, accountability, collaboration,
where the connection between breakthroughs reduced duplication, and wider impact. By sharing
and application may not be apparent and where study details, data, and results openly, researchers
it may take years, if not decades, for a technology allow others to verify, replicate, and build on their
to mature. Yet companies depend on nonindustrial work more easily. This accelerates discovery as more
research. For example, more than 80 percent of the minds can work on problems in a collaborative way,
algorithms used today (not just in AI but in all kinds with reduced redundant efforts. Published open
of information technology) originated from sources research also reaches more people, magnifying its
other than industrial research.21 Second, corporate educational and societal benefits.
R&D outputs tend to be restricted and proprietary
to preserve any market-competitive edge that they The role of universities in building the national
may afford to the company that paid for them. economy has been recognized since 1862 with the

11 Crosscutting Themes and Commonalities 137

 passage of the Morrill Act establishing land-grant uni- policy ambiguity can deter collaborations and create
versities. Government-supported university research obstacles for non-US researchers wishing to contrib-
made a critical difference in World War II on technol- ute to work in the United States. This is particularly
ogies such as radar, proximity fusing, computers— concerning as international cooperation could expe-
and the atomic bomb. University research has since dite progress in emerging fields like nanomaterials,
generated knowledge whose exploitation creates where countries like Korea are making significant
new industries and jobs, spurs economic growth, strides, especially in biomedical applications. These
and supports a high standard of living, while achiev- policy issues, widely recognized among the research
ing national goals for defense, health, and energy.23 community, underscore the urgent need for clarifi-
University research has been a rich source of new cation and reform to advance research and promote
ideas, particularly for the longer term, and uni- effective international collaborations.
versities are the primary source of graduates with
advanced S&T skills.
The Structure of Research and
Universities have the mission of pursuing high-risk Development Funding and the
research that may not pay off in commercial or soci- Valley of Death
etal applications for a long time, if ever.24 For exam-
ple, research in number theory—a branch of pure Takeaway Sustaining American innovation
mathematics—was undertaken for decades before requires long-term government R&D investments
it became foundational to modern cryptography. In with clear strategies and sustained priorities,
the 1960s, academic research on perceptrons sought not wild swings from year to year, which is
to develop a computational basis for understanding increasingly common.
the activity of the human brain. Although this line of
research was abandoned after a decade or so, it ulti- Budget is one obvious aspect of government fund-
mately gave rise to the work in AI on deep learning ing for R&D. But three other aspects deserve as least
several decades later. The term “mRNA vaccines” as much attention. First, government has an impor­
entered the public lexicon in 2021 when COVID-19 tant role in funding important research with long
vaccines were released. Yet development of these horizons, as industry is not generally structured to
vaccines was built on university research with a thirty- support long-term R&D efforts. Such government-
year history.25 Magnetic resonance imaging (MRI) funded research should generally be regarded as
was first discovered in university studies in the 1940s, precompetitive in nature.
but it took another three decades of research, much
of it university based, for the first medical MRI imag- Second, wide swings in funding from year to year—
ers to emerge. increasingly common in government funding—are
antithetical to a systematic R&D effort. In a free market
Finally, the increasingly blurred distinction between economy, talented scientists can choose where and
fundamental research and export-controlled re- in what domains to work, and they have a natural
search is creating challenges in academia in fostering aversion to work environments that do not provide
international collaboration, particularly in fields such stability. Therefore, wide swings in funding have the
as the semiconductor industry, nanotechnology, AI, effect of driving away the scientific talent that can
and neuroscience. Some researchers are concerned best find employment in that field elsewhere.
that fundamental research could now be considered
export controlled and may steer clear of foreign Third, the so-called valley of death remains a sig-
collaborations out of an abundance of caution. This nificant problem. This refers to the period after

138 STANFORD EMERGING TECHNOLOGY REVIEW

research has demonstrated the engineering fea- than before. In some cases, the innovation never
sibility of a particular innovation—a step beyond scales beyond the initial stages, regardless of its
scientific feasibility—but before the innovation technical sophistication or desirability.
achieves adoption on a scale large enough to
establish the viability of a business model based on These points suggest bridge funding may help in
the innovation. establishing commercial viability. The sticking point,
however, is the difficulty of distinguishing between
When an innovation is first offered to customers, it real innovations that would be truly valuable if only
is expected to provide new functional capability. Its they could get through the valley of death and ersatz
cost matters as well, as the new functional capability innovation look-alikes for which valley-of-death con-
may or may not be worth the price of adopting it. At cerns are merely a smokescreen to cover up their
one cost, a potential customer might choose not to genuine inadequacies and problems in the face of
acquire it, while at a lower cost, that customer may market realities.
well do so.
A firm’s failure to pass through the valley of death
If the initial cost is too high, customers will be scarce may also have competitive implications internation-
and the firm producing or providing the innovation ally. Such a firm is ripe for acquisition by foreign
is likely to fail commercially if it does not receive competitors with deeper pockets who may be will-
funding from a source not related to production to ing to invest in innovative products that have not
stay afloat. But it often happens that the per-unit yet reached market viability. Chinese investors, for
production cost will decrease as the total cumula- example, were successful in acquiring Atop Tech,
tive volume of production increases. Known as the a firm with an automated designer capable of pro-
learning curve in manufacturing, this phenomenon ducing high-end microchips, after it went bankrupt
is primarily due to the efficiencies and knowledge in 2017. This transaction failed to elicit any reac-
gained from repetitive production processes. Such tion from the Committee on Foreign Investment
cost reduction is particularly important when, as is in the United States (CFIUS), despite its mandate
true for energy production, significant societal ben- to review and, if necessary, block certain transac-
efits accrue and new technologies are deployed tions involving foreign investment that may impact
at scale. US national security. The Foreign Investment Risk
Review Modernization Act of 2018 was enacted in
Research funding generally disappears after feasibil- part to improve the ability of CFIUS to review just
ity has been demonstrated. For a firm then to get such transactions.
through the valley of death, it must either demon-
strate the viability of its business model to inves- A new funding model, known as focused research
tors who believe in the promise of the innovation organizations (FROs), seeks to fill the gap inherent
or attract enough customers on its own to sustain in the valley of death. The FRO provides funding to
it. True commercial viability is unlikely to start until assemble scientists and engineers with the required
the per-unit cost has dropped to levels affordable by expertise to rapidly prototype and test materials and
most would-be customers. technologies for their applications. One initiative to
support FROs was launched in 2021, Convergent
While in the valley of death, it is typical that no party Research, a nonprofit organization with the mission
is willing to invest the minimum level for product or of incubating and funding new FROs. In March 2023,
manufacturing refinement to continue, and projects it received $50 million in philanthropic donations to
often have to stop or progress much more slowly launch two new FROs.26

11 Crosscutting Themes and Commonalities 139

 Cybersecurity important currency in which academic R&D trades.
Additionally, draft working papers are often incom-
plete, inconsistent, or downright wrong and are not
Takeaway Researchers working in highly com-
in any sense defensible—premature disclosure of
petitive environments who neglect cybersecurity
such papers as though they did in fact reflect com-
place their research progress at risk.
pleted work is a nightmare of any scientist.

Cybersecurity refers to technologies, processes, Many laboratories rely on computers to control or

and policies that help to protect computer systems, supervise data collection from various instruments.
networks, and the information contained therein Compromising these computers through a cyber­
from malicious activities undertaken by adversaries attack could cripple data collection efforts or cor-
or unscrupulous competitors. It is often believed rupt the data being collected. The instruments in
that cybersecurity is an issue that primarily affects question could also be damaged by hacking the
private-sector businesses and government, but the controlling computers.
world of academic R&D faces a variety of cyber
threats as well. Technical safeguards are available for most cyber-
security problems, of which the above are just a
One important cybersecurity interest is in ensuring sample. But especially in academic laboratories,
the integrity of data. Scientific experiments pro- maintaining and operating such safeguards con-
duce data, and if important data are deleted or sistently calls for a serious management effort to
destroyed, scientific progress can be significantly impose the necessary discipline on all those work-
retarded. A possibly more worrisome scenario is ing in those labs. Such discipline often conflicts with
that the data are altered in hard-to-detect ways that informal laboratory cultures that stress collegiality,
subtly and invisibly skew the subsequent work based openness, and flexibility.
on that data, possibly putting scientific investigators
on the wrong track and wasting significant effort. A second cyber-related threat to the R&D enterprise
Adversaries or competitors seeking to delay scien- is selective targeting of key personnel working on
tific work have significant incentives to engage in important research projects. It is a matter of public
activities that could compromise data in this manner. record that a number of Iranian nuclear scientists
have been killed since 2007, reportedly because
Similar comments apply to the computer programs they were associated with Iran’s nuclear program.27
used to analyze data. If a computer program is mali- But assassination is not the only form of targeting.
ciously altered in a subtle way, it may be a long time Much less violent forms of targeting could involve
before the alteration is noticed. Once such an alter- what might be broadly termed harassment, which
ation is noticed, all previous analyses performed using often originates in or is perpetrated through cyber-
that program are inevitably called into question. space. For example, compromising the personal
life of a principal investigator (e.g., draining bank
A second cybersecurity interest is in ensuring the accounts, interfering with the investigator’s personal
confidentiality of various work products, such as finances, threatening the investigator’s family) can all
datasets and working papers. Datasets may have be accomplished through the internet. Dealing with
been collected under promises of confidentiality or such matters will inevitably reduce the work effec-
nondisclosure agreements, and unauthorized access tiveness of the individuals targeted in this manner.
to such datasets quite possibly violates such prom- Calling into question the investigator’s professional
ises or agreements. Premature disclosure of work- conduct and ethics is another approach that could
ing papers can compromise claims of priority, an have comparable effectiveness.

140 STANFORD EMERGING TECHNOLOGY REVIEW

NOTES

1. Science Museum, “The Iron Lung,” October 14, 2018, https:// 15. James M. Heather and Benjamin Chain, “The Sequence of
www.sciencemuseum.org.uk/objects-and-stories/medicine/iron Sequencers: The History of Sequencing DNA,” Genomics 107, no. 1
-lung. (January 2016): 1–8, https://doi.org/10.1016/j.ygeno.2015.11.003.
2. Keith Sawyer, Group Genius: The Creative Power of Collabora- 16. Jean-Lou Chameau, William F. Ballhaus, and Herbert Lin,
tion (New York: Basic Books, 2007). eds., Emerging and Readily Available Technologies and National
3. Amy Burke, Abigail Okrent, and Katherine Hale, The State of Security: A Framework for Addressing Ethical, Legal, and Societal
U.S. Science and Engineering 2022 (Arlington, VA: National Sci- Issues (Washington, DC: National Academies Press, 2014), https://
ence Board, National Science Foundation, 2022), https://ncses doi.org/10.17226/18512.
.nsf.gov/pubs/nsb20221. 17. James G. March, “Exploration and Exploitation in Organiza-
4. World Intellectual Property Organization, “World Intellectual tional Learning,” Organization Science 2, no. 1 (1991): 71–87, http://
Property Indicators: Filings for Patents, Trademarks, Industrial www.jstor.org/stable/2634940.
Designs Reach Record Heights in 2018,” October 16, 2019, https:// 18. Will Douglas Heaven, “Geoffrey Hinton Tells Us Why He’s
www.wipo.int/pressroom/en/articles/2019/article_0012.html. Now Scared of the Tech He Helped Build,” MIT Technology
5. Jennifer Hunt and Marjolaine Gauthier-Loiselle, “How Much Review, May 2, 2023, https://www.technologyreview.com/2023
Does Immigration Boost Innovation?,” American Economic Jour- /05/02/1072528/geoffrey-hinton-google-why-scared-ai.
nal: Macroeconomics 2, no. 2 (April 2010): 31–56, https://doi 19. American Chemical, “High Performance Carbon Fibers: National
.org/10.1257/mac.2.2.31. Historic Chemical Landmark,” September 17, 2003, https://www
6. Dany Bahar et al., Talent Flows and the Geography of Knowl- .acs.org/education/whatischemistry/landmarks/carbonfibers.html.
edge Production: Causal Evidence from Multinational Firms, Har- The original publication is by Roger Bacon, “Growth, Structure, and
vard Business School, Technology and Operations Management Properties of Graphite Whiskers,” Journal of Applied Physics 31,
Unit, Working Paper no. 22-047, last modified December 23, 2022, no. 2 (February 1960): 283–90, https://doi.org/10.1063/1.1735559.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4005693. 20. Congressional Research Service, Federal Research and Devel-
7. Amy Burke, Abigail Okrent, and Katherine Hale, “U.S. and opment (R&D) Funding: FY2022, January 2022, https://crsreports
Global STEM Education and Labor Force,” in The State of U.S. .congress.gov/product/pdf/R/R46869.
Science and Engineering 2022 (Arlington, VA: National Science 21. Neil C. Thompson, Shuning Ge, and Yash M. Sherry, “Building
Board, National Science Foundation, 2022), https://ncses.nsf.gov the Algorithm Commons: Who Discovered the Algorithms That
/pubs/nsb20221/u-s-and-global-stem-education-and-labor-force. Underpin Computing in the Modern Enterprise?,” Global Strategy
8. US Department of Defense, Industrial Capabilities Report Journal 11, no. 1 (2021): 17–33, https://doi.org/10.1002/gsj.1393.
to Congress: 2020 Annual Report, January 2021, https://www 22. National Laboratories,“What We Do,” accessed September 19,
.businessdefense.gov/docs/resources/USA002573-20_ICR_2020 2023, https://nationallabs.org/our-labs/what-we-do.
_Web.pdf. 23. National Research Council, Research Universities and the
9. Britta Glennon and David R. Dollar, “What’s Behind the Global- Future of America: Ten Breakthrough Actions Vital to Our Nation’s
ization of R&D?,” April 26, 2021, in Dollar & Sense, podcast pro- Prosperity and Security (Washington, DC: National Academies
duced by Brookings Institution, MP3 Audio, 25:02, https://www Press, 2012), https://doi.org/10.17226/13396.
.brookings.edu/articles/whats-behind-the-globalization-of-rd. 24. National Academies of Sciences, Engineering, and Medicine,
10. Cecilia Esterline, “Previously Unreported Data: The U.S. Lost Information Technology Innovation: Resurgence, Confluence, and
45,000 College Grads to Canada’s High-Skill Visa from 2017 to Continuing Impact (Washington, DC: National Academies Press,
2021,” Niskanen Center, March 14, 2023, https://www.niskanen 2020), https://doi.org/10.17226/25961.
center.org/previously-unreported-data-the-u-s-lost-45000-college 25. Elie Dolgin, “The Tangled History of mRNA Vaccines,” Nature,
-grads-to-canadas-high-skill-visa-from-2017-to-2021. October 22, 2021, https://www.nature.com/articles/d41586-021
11. Immigration, Refugees, and Citizenship Canada, “Canada’s -02483-w.
Tech Talent Strategy,” Government of Canada, last modified 26. Alex Knapp, “Why Billionaires Ken Griffing and Eric Schmidt
June 27, 2023, https://www.canada.ca/en/immigration-refugees Are Spending $50 Million on a New Kind of Scientific Research,”
-citizenship/news/2023/06/canadas-tech-talent-strategy.html. Forbes, March 17, 2023, https://www.forbes.com/sites/alexknapp
12. Clare McDonald, “Quarter of UK Businesses Looking /2023/03/17/why-billionaires-ken-griffin-and-eric-schmidt-are
Abroad for Tech Talent,” Computer Weekly, December 14, 2022, -spending-50-million-on-a-new-kind-of-scientific-research.
https://www.computerweekly.com/news/252528353/Quarter-of 27. Mehdi Jedinia, “History of Assassinations of Iran’s Top Nuclear
-UK-businesses-looking-abroad-for-tech-talent. Scientists,” Voice of America, December 03, 2020, https://www
13. Michael Distefano et al., The Trillion-Dollar Difference, Korn .voanews.com/a/extremism-watch_history-assassinations-irans
Ferry Institute, 2016, https://www.kornferry.com/content/dam -top-nuclear-scientists/6199135.html.
/kornferry/docs/article-migration/Korn-Ferry-Institute_The-trillion
-dollar-difference.pdf.
14. Adam Satariano et al., “Elon Musk’s Unmatched Power in the
Stars,” New York Times, July 28, 2023, https://www.nytimes.com
/interactive/2023/07/28/business/starlink.html.

11 Crosscutting Themes and Commonalities 141