NAE Perspectives offer practitioners, scholars, and policy leaders a platform to comment on developments and issues relating to engineering. 

Venkatesh Narayanamurti is Benjamin Peirce Professor of Technology and Public Policy, Engineering and Applied Sciences, and Physics Emeritus at Harvard University. He was Founding Dean of the Harvard School of Engineering and Applied Sciences.

Over the past decade there have been consistent alarm signals about US leadership in science and technology.[1] Arguments often boil down to the need for additional funding for R&D. In this perspective, I reflect not on the well-justified need for such additional funding, but for more effective funding, especially in technoscientific research, a key engine of innovation.

Beyond Vannevar Bush

In Cycles of Invention and Discovery: Rethinking the Endless Frontier (Harvard University Press, 2016), Tolu Odumosu and I argued that the American linear model of innovation codified after World War II—that scientific R&D is the pacesetter of engineering R&D—is faulty.

Epitomized by the great industrial R&D laboratories of the past century—places like GE, Bell Labs, IBM, Dupont, Xerox PARC—path-breaking scientific discoveries and engineering inventions go hand-in-hand, in a dance in which sometimes one leads, sometimes the other, and sometimes both occur almost simultaneously. Examples from the past abound: the transistor, the laser, and the fundamental principles of communication and information technology. Future examples can be anticipated: quantum information science and engineering, microelectronics and computing for machine learning and artificial intelligence, energy technology innovation to address climate change, and synthetic biology and biologically inspired engineering.

The Technoscientific Method

In my most recent book, with Jeff Tsao of Sandia National Laboratories, The Genesis of Technoscientific Revolutions: Rethinking the Nature and Nurture of Research (Harvard University Press, 2021), we dive more deeply into the dance between scientific and engineering R&D.

We make a central distinction between, on the one hand, science and technology (denoted S and T, respectively) as repositories of knowledge; and, on the other hand, the scientific and engineering methods (denoted Ṡ and Ṫ) as dynamic processes by which new science and technology are created. The scientific method—the processes by which new science is created—depends on existing science and technology, so one might write Ṡ = f (S, T); likewise, the engineering method—the processes by which new technology is created—also depends on existing science and technology, so one might write in a symmetric manner Ṫ = g (S, T).

There is thus a virtuous cycle between science and technology, mediated by what we call the technoscientific method: the combination of the scientific and engineering methods, neither leading but each strengthening the other. The engineering invention of the steam engine famously catalyzed the scientific discovery of the principles of thermodynamics; the scientific discovery of the principle of mass-energy equivalence famously catalyzed the engineering invention of the atomic bomb; and the engineering invention of the transistor and the scientific discovery of the transistor effect catalyzed each other nearly simultaneously.

Technoscientific Research ≠ Technoscientific Development

In Genesis we make another central distinction: between Ṡ and Ṫ that consolidates and extends conventional wisdom, and Ṡ and Ṫ that surprises and overturns conventional wisdom. The first we associate with development (D) and the “schedulable” meta goal of practical utility; the second we associate with research (R) and the “unschedulable” meta goal of learning and surprise. Together, they lead to a “punctuated equilibrium,” in which the consolidation of conventional wisdom is punctuated by surprise to conventional wisdom—analogous to how, in evolutionary biology, gradual change in existing species is punctuated by the sudden creation of new species.

Importantly, the distinction between R and D is independent of the distinction between Ṡ and Ṫ. Both Ṡ and Ṫ can have research or development flavors: there is new science that surprises and disrupts, there is new science that consolidates and extends; there is also new technology that surprises and disrupts and there is new technology that consolidates and extends. Of the distinctions, that between R and D is far more important: an organization that emphasizes research and surprise to conventional wisdom must be funded very differently from an organization that emphasizes development and consolidation of conventional wisdom.

Question Finding ≠ Answer Finding

Another central distinction we make is that, among the various processes associated with the technoscientific method, there are those with a “question finding” nature and those with an “answer finding” nature. Both are important.

Finding answers to known questions is important, as epitomized by DARPA’s famous Heilmeier catechism.[2] But finding new questions is also important, as articulated by Einstein: “The formulation of a problem is often more essential than its solution, which may be merely a matter of mathematical or experimental skill. To raise new questions, new possibilities, to regard old questions from a new angle, requires creative imagination,”[3] a hallmark of both scientific and engineering breakthroughs.

In related terminology, reductionism, in which new answers are imagined by reduction into constituent parts, is important; but constructivism, in which new questions are imagined by integration of constituent parts, is also important.

Indeed, it is the intricate dance between answer finding and question finding, between reductionism and constructivism, that reuses and connects knowledge domains in new ways and that creates the “seamless web of knowledge” whose power was articulated 2 decades ago by Phil Anderson.[4] This dance is at work everywhere in the seamless web of knowledge. As also articulated by Anderson: the integration of constituent parts into larger systems almost always leads to emergent phenomena that could not have been predicted solely by reductionist thinking, but requires constructivist, “more is different” thinking.[5]

Materials science and device engineering has been a rich source of such emergent phenomena—from high-temperature superconductivity in oxides to fractional quantization in high electron mobility transistors. Likewise, computational hardware and software has been a rich source of such emergent phenomena—from the iPhone’s “App Store” to artificial intelligence (AI) in the cloud.

Nurturing Effective Technoscientific Research
Let me come back now to my call, at the beginning of this perspective, for more effective technoscientific research. Technoscientific research is fragile, and its organization, funding, and governance must be carefully nurtured in a manner that is aligned with its nature.

Because technoscientific research relies on a virtuous cycle of Ṡ and Ṫ, research funding agencies must not artificially separate Ṡ from Ṫ but instead encourage holistic synergy and symmetry in their evolution. Recent legislation in the US Congress that reforms the charter of the National Science Foundation and recognizes this deeper union between Ṡ and Ṫ is a hopeful sign. [6]I even argue that the legislation could go further and propose to change the agency’s name to the National Research Foundation to eliminate entirely funding-enforced stovepiping between Ṡ and Ṫ. Because technoscientific research aims to surprise, its impact, often enormous, cannot be anticipated—both in terms of when the impact will occur and whom the impact will benefit. Thus, research must not be held, as it increasingly is, accountable to short-term private impact (confined to the organization that funds or performs the research), but rather to long-term public impact (extending beyond the organization that funds or performs the research).

Because technoscientific research aims to find new questions just as much as new answers, a questioning attitude is critical and highlights the need to nurture informed contrarians who have a penchant for challenging the status quo and stretching boundaries, be they scientific, technological, or cultural. But question finding, just like surprise, can be neither scheduled nor “projectized.” Thus, we must shift from funding projects, which is increasingly the norm, to funding people—not just as a mantra, but as fundamental to effective technoscientific research.

Moving Forward

The industrial labs of the last century may be physically of the distant past, but some of the lessons learned from that period about pushing forward the research frontier are timeless. I hope that as we learn more about the social construction of technoscientific research and research organizations, we will put that knowledge on firmer ground, and then go further.

The opportunities are enormous. The emerging revolutions in AI/biological and quantum computing are just two examples. To be at the frontier in these and other areas will require more effective funding: funding that engages rather than short-circuits the virtuous cycle of Ṡ and Ṫ; that seeks surprise to, not just consolidation of, conventional wisdom; and that nurtures people to engage in the constructivist question finding that often crosses disciplinary boundaries, including the linking of the hard sciences and engineering with the computational and social sciences that future computing may require. These public-goods risks are truly worth taking.

[1] American Academy of Arts & Sciences. 2020. The Perils of Complacency: America at a Tipping Point in Science & Engineering. Cambridge MA.

[2] https://www.darpa.mil/work-with-us/heilmeier-catechism

[3] Einstein A, Infeld L. 1938. The Evolution of Physics. Cambridge UK: Cambridge University Press.

[4] Anderson PW. 2001. Science: A ‘dappled world’ or a ‘seamless web’? Studies in History and Philosophy of Modern Physics 32(3):487–94.

[5] Anderson PW. 1972. More is different. Science 177(4047):393–96.

[6] S.1260 of 117th Congress, United States Innovation and Competition Act of 2021 (https://www.congress.gov/bill/117th-congress/senate-bill/1260). 

Disclaimer

The views expressed in this perspective are those of the author and not necessarily of the author’s organization, the National Academy of Engineering (NAE), or the National Academies of Sciences, Engineering, and Medicine (the National Academies). This perspective is intended to help inform and stimulate discussion. It is not a report of the NAE or the National Academies. © National Academy of Sciences. All rights reserved.