Improving the Endless Frontier Act
Proposed legislation to catalyze technological innovation is unlikely to succeed. Here’s why—and how the United States can maintain its leadership in the development and commercialization of important new technologies.
In an increasingly competitive global economy, America’s leadership in technological innovation is no longer assured. The Endless Frontier Act, with bipartisan sponsors in both the House and Senate, is a new legislative initiative aimed at helping the United States stay ahead of foreign competitors. Key provisions of the bill would create a new directorate at the National Science Foundation to support fundamental research at universities to help catalyze the next generation of valuable new technologies, with $100 billion in expenditures authorized over five years.
The challenge is real. Does this legislation move the nation in the right direction?
The authors of the bill, Senators Charles E. Schumer (D-NY) and Todd Young (R-IN), and Representatives Ro Khanna (D-CA) and Mike Gallagher (R-WI), have recognized that government investments in advanced technologies are good for the United States. This is exactly the right place to start. Consider a few of the technologies that federal research and development agencies and the engineering teams they fund have created: advanced jet aircraft, atomic energy, mass production of antibiotics, digital computers, the internet, the Global Positioning System, and biotechnology. In addition, the government has served as a valuable early customer for industry-developed technologies, including integrated circuits, solar cells, and communications satellites. The contributions of these government investments to the US economy and national security have been enormous. Economists beginning with Robert Solow in the 1950s have shown that a major portion of the nation’s economic growth over the past century is the result of technological advancement.
Moreover, since World War II the United States has understood how to effectively create and use world-changing technologies. The country has a robust national innovation system. It includes federal funding of university researchers to expand the frontiers of knowledge; government technology development organizations such as the Defense Advanced Research Projects Agency (DARPA) and the National Aeronautics and Space Administration that fund engineering teams to create and demonstrate prototypes; companies that design, demonstrate, and commercialize these new government-funded technologies as well as technologies developed in a wide array of industry laboratories; and a rich mix of venture capitalists and other services that support start-up companies. There are frequently debates over whose interests federal R&D should serve and what national needs it should address, but there is no question that Americans know how to create valuable new technologies.
But the Endless Frontier Act’s approach to enhancing the nation’s innovation system is unlikely to succeed. The act would create a Technology Directorate at NSF (and in the process change the agency’s name to the National Science and Technology Foundation, NSTF), and would give the new directorate three goals: strengthen US leadership in critical technologies through fundamental research in key technology focus areas; improve education in key technical areas; and, consistent with the operations of NSF, foster the economic and societal impact of federally funded R&D through accelerated translation of fundamental technology advances into processes and products. The new directorate’s awards would go to institutions of higher education, other not-for-profit entities, and consortia led by institutions of higher education.
Is giving additional money to universities for fundamental research—basic and use-inspired research, in NSF’s definition—really the most effective way to develop and commercialize new technologies?
Fundamental research in universities and government laboratories is of course important, and some valuable start-up companies have come directly out of university research. But the idea that university research can and should be the key driver of US technology development and commercialization is a limited and outdated model.
We know how to do this
In the first place, most radical new technologies in recent decades have come from solutions-driven engineering work funded by agencies such as DARPA and NASA or from companies driven by market opportunities. On the government side, DARPA has played a decisive role in developing advanced materials, the personal computer, the internet, and more recently advanced prosthetics and RNA vaccines. Several government organizations, in particular the Naval Research Laboratory, were central to the development of the Global Positioning System. Government technology agencies draw on important fundamental research funded by NSF, and they themselves also fund additional university research in support of their technological goals—but as part of larger R&D programs that fund a wide range of R&D performers, including companies that often go on to commercialize these new technologies. University discoveries contribute to such work, but university research alone did not create most of the radical new technologies of our time.
Although the incentives and capabilities at universities and at NSF itself are well-aligned with basic research and open publication, universities are not equipped to undertake large applied engineering projects, much less to translate the resulting new technologies into products and processes. Putting universities in the lead on technology development misunderstands both their role and their capabilities in the innovation system. It would also risk diluting the valuable basic research role that universities and NSF play.
The bill also falls prey to the allure of radical technical progress as the sole source of technological and economic leadership. Breakthrough technologies do indeed help a country create new industries, but keeping existing industries healthy and competitive also requires steady incremental improvements in technology. The Endless Frontier Act says little about this second type of technical advancement, but examples include applied work on manufacturing technology funded by companies, the Defense Department, the Energy Department, and the National Institute of Standards and Technology. Today, the manufacturing institutes of the Manufacturing USA program are an important part of this essential work. Universities play a valuable role in these institutes, but the institutes succeed because small companies, large companies, government laboratories, and universities work together to improve manufacturing processes and products. Universities are players, not leaders, in this work.
Scan, don’t predict
It is tempting to say that the United States should focus its R&D resources on highly visible technologies such as artificial intelligence and quantum computing, particularly since other countries have major initiatives in these areas. The Endless Frontier Act uses this approach of focusing on key technology areas; the bill even provides an initial list. But here, too, the legislation does not capture important lessons about the strengths of the US innovation system.
US efforts in the 1990s to identify “critical technologies” did not succeed, partly because it is hard to predict which technologies will be most valuable in the future and partly because decisions about R&D funding priorities inevitably become political, as groups and leaders vie to have their favorites supported. Often this type of process results in a broad list that pleases everyone but is largely useless as a guide to policy.
The US government actually has a better way to identify and fund promising new areas of research and technology. National leaders set overall priorities while researchers and agency experts scan for new scientific and technical opportunities and propose new R&D directions. Then agency leaders and, for big initiatives, the White House and Congress vet these ideas and decide which to support. The result is a flexible federal system that identifies new opportunities, reviews them, and creates a diverse and high-quality portfolio of R&D programs.
The top-down portion—statements of overall national priorities—consists of annual White House memoranda on presidential R&D priorities (including one for fiscal year 2021), agency planning documents, and congressional laws. The bottom-up portion is a remarkable American strength. Instead of looking only at current technologies, researchers and agency technical experts constantly scan for the next big things in their fields and propose new initiatives. Sometimes agency directors directly evaluate these ideas and decide which to support, as was the case with the National Nanotechnology Initiative. Sometimes NSF workshops and National Academies meetings test and refine new R&D proposals before policy leaders consider them, as with Academies reports that help set priorities in chemistry, space sciences, and other fields. Some of the resulting investments are large, such as multiagency initiatives in high-performance computing and nanotechnology, while many others are smaller or even experimental “seedling” projects. By funding a wide range of existing and new R&D areas—funding an overall R&D portfolio—federal agencies do not just develop today’s technologies; they also begin investing in the technologies of the future. Confining technology development support to a relative short list of predetermined areas that can only be updated every four years or so seems sure to result in a system far less dynamic than the current one.
Here is one example: by the 1990s, the original versions of artificial intelligence, such as expert systems, had plateaued. Expert systems were expensive to create and could solve only a limited number of problems. But Daphne Koller of Stanford University and other leaders in the field convinced government agencies that growing computer power and vast new data might one day allow new forms of AI, and DARPA and other government agencies, including some in Canada, placed important bets on new but still unproven ideas. The result was the current revolution in machine learning and modern AI. Again, this is the lesson: constantly scan for new possibilities and create future technologies as well as improve existing ones. Do not simply stick to a list of current high-profile technologies.
Let NSF be NSF
NSF is a well-run and successful agency with broad bipartisan support. It is understandable that the authors of the Endless Frontier Act selected it as the home for these new technology development responsibilities. But this new mission would not fit well with NSF’s existing basic research culture, and it would be a shame to damage the existing agency and its programs in an effort to turn NSF into something that it’s never been.
It also might not be possible. NSF has almost no experience or expertise in managing technology development programs, with the notable exception of its work in the 1980s and early 1990s to help develop advanced computing and the NSFNET, which eventually formed a backbone of today’s internet. The bill discusses “moving technology from laboratory to market,” but its implicit theory of commercialization is incomplete. For example, the bill would fund universities to train “scientists and engineers who are interested in research and commercialization.” But what kind of training? Would it be short-term mentoring, similar to NSF’s existing Innovation Corps? Or would it be substantial and comprehensive, like the two-year fellowships and laboratory access provided by the highly regarded Activate organization to help young technical entrepreneurs, which began at Lawrence Berkeley National Laboratory? Beyond questions of how to recruit and equip academic entrepreneurs, the bill seems to ignore the fact that most commercialization—getting new technologies into the marketplace—is done by existing companies or nonacademic entrepreneurs. Yet the bill says almost nothing about industry or even how universities will in fact create and demonstrate complex technologies that companies will then want to commercialize. How would the new NSTF succeed without meaningful industry participation?
NSF might figure all this out, of course, but this would be a new role with a steep learning curve and possible harm to existing programs. Meanwhile, as we’ve said, other agencies and programs have the necessary track record and expertise.
The bill would also create unnecessary management confusion and operational problems at NSF. The new Technology Directorate would be in NSF but semi-independent of it, raising questions about who is really in charge. The bill speaks of DARPA-style program managers, who operate independently of peer review panels, but then the legislation also says that the directorate may use peer review. So, how is the directorate supposed to pick and supervise projects? Also, what is the relationship between the new Technology Directorate and NSF’s existing directorates for engineering and computing? Then the lab-to-market section emphasizes training academic scientists and engineers who are interested in commercialization—but as we’ve emphasized, most commercialization in the United States is done in the private sector.
One particularly muddled part of the bill would create a Board of Advisors for the new Technology Directorate. What is its role exactly? How much is the directorate’s leadership supposed to listen to the board’s advice? Is the board supposed to help set overall policy for the new directorate? Eight of the 12 members will be appointed by congressional leaders. The authors of the bill may feel that oversight by congressionally appointed advisors will reassure Congress that the money is well spent. However, to have congressionally appointed advisors oversee an executive branch agency is unusual—and in fact a violation of the separation of powers. And why does this advisory board need a substantial staff and exemption from the Federal Advisory Committee Act that governs the behavior of federal advisory committees, as stipulated in the legislation? At a minimum, the presence of an advisory board with so many congressionally appointed members will cause management confusion. It could also thoroughly politicize the directorate’s agenda from day one. A better approach is to rely on the traditional, proven approach: executive branch officials in charge, with Congress providing supervision through oversight, authorizations, and appropriations.
Try this instead
To sum up, serious questions arise about the ability of NSF (or NSTF) and its university researchers to create significant new technologies and then help commercialize them. Is there a better way to help the United States maintain leadership in the development and commercialization of important new technologies? We believe the answer is yes, and here we offer, for the sake of discussion, a three-part alternative strategy.
1. Boost DARPA funding and create DARPA-type entities in other federal agencies. A first step is to significantly increase funding for DARPA (especially the offices that develop fundamental technologies, rather than military systems). But Congress may also wish to create other DARPA-like entities associated with other mission agencies, as it did when it created the Advanced Research Projects Agency-Energy (ARPA-E) at the Department of Energy in 2009. In its short history ARPA-E has already shown that this model can be adapted to new missions and can create valuable new technologies. One can imagine other versions associated with the Departments of Commerce, Agriculture, and Transportation, among others. DARPA itself has recently received additional funding for work on artificial intelligence, one of the priority technologies in the Endless Frontier Act. To the extent that AI has dual-use implications, this makes sense. But it would be a mistake to expect DARPA to diverge from its focus on its defense mission; let it continue to explore promising new ideas, not just existing technologies, and let it continue to use its program manager system to identify and then manage promising new R&D. But for applications clearly outside the defense realm, Congress should create other capabilities, as it did with ARPA-E.
2. Pilot a new Technology Frontier Projects Program at NSF. Second, NSF should undertake a policy experiment to test whether universities are indeed able to play a new role in national technology development. This effort would be less-far-reaching than the one in the Endless Frontier Act, but it would create a learning environment in which the more revolutionary concepts in the bill, such as universities planning and carrying out complex engineering projects, could be put to the test. Under this idea, Congress would authorize and appropriate funds for what might be called a Technology Frontier Projects Program, which would be administered by NSF’s existing directorates. The projects would be different from most university research. They would be solutions-focused (that is, aimed at building and demonstrating significant new technology capabilities); engineering-focused and multidisciplinary; and operate under NSF contracts (not grants), with deliverables and milestones. Project teams would be allowed to hire full-time professional engineers, as well as use students and postdocs. The teams would include industry participation, but would focus on long-term, high-risk breakthrough technologies and not industry’s short-term needs. In addition to encouraging project leaders to work with existing companies, NSF should fund a new set of two-year fellowships for would-be academic entrepreneurs who worked on these projects—fellowships similar to Activate’s. NSF should also consider reorienting some of its Small Business Innovation Research funds toward helping small firms develop products based on this university R&D. This policy experiment might rightfully justify changing the name of NSF to the National Science and Technology Foundation. But such a program should start small and ramp up only as it demonstrates success.
3. Expand support for Manufacturing USA institutes. Congress should continue and expand federal funding for Manufacturing USA’s manufacturing institutes. These manufacturing institutes do focus on the shorter-term and medium-term needs of manufacturing companies and thus play an important role in keeping American companies technologically strong. The manufacturing institutes are still relatively new, and evaluations are needed to learn lessons and assess impacts. However, they need to become an important part of the US national innovation system.
The Endless Frontier Act is an important first effort to bolster the development of new technologies that will be vital to the nation’s economic growth and national security. Two parts of the bill that we have not covered here call for a Regional Technology Hub Program to help ensure widespread distribution of economic benefits, and for a national strategy and annual report on economic security, science, research, and innovation. Both deserve consideration. But if the goal of the bill is to produce “new technologies to ensure the leadership of the United States in the industries of the future” rather than simply to support additional university research, then the DARPA-type model of technology development should be carefully considered. Enough is known about the strengths and weaknesses of the nation’s innovation system to guide the development of effective national investments in America’s future. Further discussion inside and outside of Congress will strengthen all parts of this bill and help the United States stay technologically strong.