The End of the Line

By G. Pascal Zachary

Innovation is almost universally desired but almost always misunderstood. Confusion abounds over such basic tasks as how to describe how innovation works and even what counts as innovation. If culture conditions innovation, as surely it must, then can some national and subnational cultures possess more innovative capacity than others? How much does geography matter? Is innovation in digital electronics fundamentally different than in, say, energy, transportation, or biopharma? Is it possible to speak of social and technological innovation in the same breath? What does it mean to go from “imitation to innovation,” as South Korea’s national champions have done, and yet still insist that science and the discovery of new knowledge decisively contribute to technological advance and human well-being?

Fortunately for innovators, whether situated in universities, industry, or civil society, advances across a range of fields can occur without ever answering any of these questions. More urgent than a fundamental understanding of innovation in all its marvelous forms is public understanding of and appreciation for the relationship among public funding, government policies, and innovation outcomes. Whereas scholars take for granted that government is often the handmaiden of vital innovations, the case for the centrality of public funds is today battered and beaten by conservative and corporatist critiques that insist government spending on research and development too often amounts to a form of glorified welfare for scientists and engineers isolated from markets.

Concerned by what they consider to be weak outcomes from publicly funded science and engineering, Venkatesh Narayanamurti and Toluwalogo Odumosu have produced a necessary new book on the politics of research, Cycles of Invention and Discovery: Rethinking the Endless Frontier. In their wide-ranging, well-documented, and deeply informed analysis, the two scholars of innovation effectively demolish the so-called “linear model.” In this conceptual framework, technological innovation begins with basic research—often in a scientific laboratory—and moves to applied research and engineering, followed by diffusion of the innovation. In debunking this model, the authors draw on evidence from the history of science and technology as well from detailed accounts of private-sector innovation. Arguing that government funders and policy makers remain devoted to a false, unidirectional understanding of the flow among science, engineering, and innovation, they deliver the bracing conclusion that federal research policy—and some significant funded research—”has become so divorced from actual practice that in many cases it is now an impediment to the research process.”

The authors maintain that rather than science serving as nourishment for engineering, some forms of research, whether done by scientists, engineers, or even use-inspired amateurs, “proceed interactively.” Citing seminal papers by the historian Edwin Layton in the 1970s on “technology as knowledge,” they persuasively argue that many crucial inventions in the past “reached relatively advanced stages of development before detailed scientific explanations about how the technologies worked emerged.”

Some of their most illuminating examples come from the co-evolution of physics and electrical engineering in the twentieth century. For instance, a multidisciplinary team at Bell Labs, the research arm of AT&T during the company’s decades as a telephone monopoly, created in 1947 the integrated circuit, which became the building block for digital computers and the massive semiconductor industry. In another case, the legendary applied mathematician Claude Shannon, pursuing ways to expand AT&T’s capacity, made a seminal breakthrough in information theory that had wide conceptual and practical applications. In charting the effect of advances in physics, electrical engineering, and applied math on digital innovation, the authors argue that the linear model is clearly inaccurate, because the “boundaries are porous” between physics and electrical engineering and “research trajectories in either field can intersect and bisect each other.”

Cynicism in the United States about the value of government sponsorship for innovation was not always as high as it is today. Beginning in 1940, the federal government began aggressively promoting science and engineering. Significant outcomes came in waves during World War II and in the decades following the war. Computers and information processors, vaccines and novel medical therapies, jet engines and space travel, and of course nuclear weapons were some of the results of sustained spending by government on research. Distinctions between “basic” science and “applied” engineering became political terms of art, useful for justifying large appropriations of public money, the authors write, but at best “a very partial and incomplete picture of how the science and technology enterprise functions.”

Narayanamurti and Odumosu, who unabashedly declare they wish to “hasten” the “demise” of the pure/applied distinction, want to blame Vannevar Bush, the electrical engineer who served as the first presidential science adviser, for successfully promoting this distinction in his highly influential 1945 report, Science, the Endless Frontier. But as the historian Ronald Kline argues in an article on the public rhetoric of US scientists and engineers from 1880 to 1945, which Narayanamurti and Odumosu draw on heavily, leading engineering societies in the 1920s and 1930s advocated loudly for the linear model and the pure/applied distinction in research, going so far as to insist, as Kline writes, that “applied science itself will dry up unless we maintain sources of pure science.” (Ronald Kline’s latest book, The Cybernetics Moment, is reviewed by David Auerbach in this issue.)

Just as Bush was not the only leader who made a pragmatic decision to present science as the source of new knowledge and technology as the application of this knowledge by engineers, attacks on the linear model are not new. Since the economist Richard Nelson published his influential The Moon and the Ghetto in 1977, scholars and policy makers have fretted that producing more science did not automatically, or even ever, result in practical solutions to urgent social problems, such as improved health, education, and environmental quality. But rejecting the linear model, though amply justified conceptually, is not the same as identifying a replacement. Science and engineering contribute to innovation, in different ways and to different degrees, depending on the situation. The making of the atomic bomb, for instance, relied on theoretical physics to a degree that suggests, as Narayanamurti and Odumosu themselves concede, that the linear model “is at times correct” in “accounting” for a specific innovation.

How science and engineering interrelate remains intensely debated by intellectuals, policy makers, and practitioners. Deceptively simple questions continue to provoke anxieties over how and why technological change occurs and how such change can be accelerated or managed. Who should fund the science and engineering that leads to innovation? Who should own and distribute the fruits of this research? Do some innovations arise only from private investment and market forces, while others arise only from public funding and “use inspired” research? And though human values and social institutions are undeniably influenced by techno-scientific advance, do values and social forces simultaneously help to shape or construct the demand for specific innovations and innovations as artifacts themselves?

All of these questions arise from the central tension between the innovations people think they want and the innovations they actually get. Resolving this tension is no easy task, although there are models for generating more socially desired innovations. Narayanamurti and Odumosu invoke the many public goods produced by Bell Labs. They cite in particular the breakthrough work in the 1940s that created the transistor, which laid the basis for digital electronics and reflected deep collaboration among engineers and scientists, technicians and inventors.

Bell Labs was an outlier, of course, the beneficiary of both the largesse of its monopoly parent and explicit mandates from the federal government that the lab’s inventions in computing and information be widely and speedily disseminated to promote competition in emerging industries. Today, few industrial labs exist, and none of the high-fliers in the digital economy—companies such as Google, Facebook, Apple, and Amazon—maintain such centralized labs. Instead, these innovative companies ask their researchers, whatever their pedigree, to stick closely to product development and think deeply about what markets will support, not what science sustains. The authors don’t have an index entry for Google, but I surmise that they would approve of the company’s “moonshot” efforts in driverless cars, sensors, language translation, and virtual-reality glasses, for instance. Whereas such research would seem to lack immediate market applications, the work ought to benefit Google’s competitive position—and the health of society.

Narayanamurti and Odumosu are most persuasive when they argue that a new language is needed to describe and sustain innovation with public funds, and that even though the linear model is intellectually bankrupt, it is now a zombie framework that sows confusion and retards reform of publicly funded research. “Nomenclature is important,” they write. “We should immediately drop the use of the terms basic research and applied research and instead talk about ‘research’ with the understanding that it encompasses both invention and discovery.” In short, “development,” which the authors describe as “a scheduled activity with a well-defined outcome in a specified time frame,” must be viewed as part of a single research domain.

That the authors fail to build a persuasive case for their preferred conceptual understanding of research should not diminish the achievements of Cycles of Invention and Discovery. They do a great service by cogently arguing that the riddle of the politics of innovation is subject to rational analysis and that researchers, wherever they fall on the spectrum of science, engineering, discovery, and invention, must make good more often on their promise to deliver what humans say they want and need, and do so in an economic and expeditious way. Few can disagree with these high-minded aims.