The Perils of Innovating on the Edge

Arati Prabhakar describes a new venture: forming a private-sector entity modeled on the Defense Advanced Research Projects Agency to sponsor research breakthroughs. But wait, you say. DARPA is high risk, oriented to technology breakthroughs. The private sector likes to limit risk, and focus on incremental technology advances under a “stage gate” approach that weeds out technologies based on their potential for near-term profitability. So what’s a DARPA model doing in the private sector? Won’t it be a fish without water? After all, DARPA has been around for 62 years and nobody has ever tried to replicate it in the private sector. So what are you up to, Dr. Prabhakar?

The nonprofit organization she has helped set up, called Actuate, aims to provide seed support for breakthroughs in four areas: data and information that can be trusted in an era where data privacy is ending, information-based coaching for better population public health, applying social science for economic opportunities for a broader base of the population, and mitigating climate change. Answers would come from a blend of data analytics, modeling complex systems, and, through interactive devices, personalized interactions at scale. Actuate wants what it calls “solutions R&D,” using these information tools to solve challenges, as DARPA does, that are “barely feasible” but still feasible. It is seeking areas for social returns with important innovation opportunities that are completely outside problems addressed today by government R&D agencies.

Can Actuate round up the seed money to get its projects going? Venture capital holds much of society’s risk capital, but it seeks large, short-term returns, not societal solutions, so is unlikely to help. But are there other sources of money out there—are there relevant examples? Deeply worried about future funding for basic science, Marc Kastner left his position as a science dean at the Massachusetts Institute of Technology to organize the Science Philanthropy Alliance to encourage successful entrepreneurs to become Medici’s of science, with some significant successes. Prime Coalition in Boston was set up to fund high-risk, clean energy start-ups in “barely feasible” areas such as direct carbon capture that would never get venture funding. Applying charitable tax laws, it has built a fund from philanthropists and foundations with a portfolio of early-stage companies. So, there are examples. Philanthropy is a growing, vastly tax-advantaged and largely unaccountable force in society. Robert Reich in a new book argues that although it has placed little focus on our growing economic inequality, philanthropy could be centered on longer term innovations that build democratic experimentalism. That power, not market capital, if harnessed, could support Actuate.

So the funding may not be impossible to raise, and economic as well as societal benefits may eventually flow. DARPA’s high-risk projects did stimulate massive private-sector investments once DARPA had shown the way. That’s how much of Silicon Valley evolved once Stanford’s engineering dean, Frederick Terman, marked off the boundaries for an industrial park.

But what about information system solutions for societal challenges? Jay Forrester was a computing pioneer who developed magnetic core memory and stood up Whirlwind, the early breakthrough real-time computer. He left his computers in 1956 to become a business school professor and, applying his computer systems experience, he created the field of system dynamics. He aimed his system at solving urban problems in a period of failing cities and social unrest, spurring a continuing debate on the feasibility of modeling broad social problems. The results were thin, although the computer game Sim City was influenced by his ideas. Prabhakar and team are betting that a new generation of far more sophisticated information tools can take up this light sword in four specific societal areas. It’s a laudable goal—may the force be with you.

William B. Bonvillian

Arati Prabhakar says the fundamental premise behind Actuate, the new philosophically motivated venture she cofounded, is to adopt the vision and model of the Defense Advanced Research Projects Agency and its successes in creating wonderous technology from basic research for addressing societal problems. The focus is on addressing health care challenges, inequality in opportunity, and global climate change—enabled by a trustable data infrastructure.

One hurdle will be the fundamental difference between national security and societal problems. National security is socially accepted as a public good. Societal problems such as inequality, health, and global climate change are not accepted as a public good problem yet in the United States. The current pandemic and its economic effects show that the health of the nation is a public good, and that the current system is not sustainable even with all our best technology. This difference in the nature of the good changes the dynamics of what is or is not possible with a DARPA-like structure.

Unquestioned public good status of national defense allows the Department of Defense, which houses DARPA, to define a global strategy that can be implemented top-down weighing social and global security interests and that will be supported. In that context, DARPA as an R&D arm focuses on serving the defense of the country, with emphasis on force amplification (projection of power), maximizing information collection and survival (intelligence), and the ability for dynamic command and control of resources and automation of processes toward efficient “outcomes” with minimal human casualties to win wars. This triad of needs in light of the desired outcome requires investigations into science, medicine, and engineering problems. These goals evoked the individual “curiosity” of the scientists and engineers that was tapped in a publicly driven public-private partnership. Spillover from these defense technologies enabled industry to pursue the same trajectory in automation, information intelligence, reducing human resource needs, enhancing productivity, and reducing uncertainty in a relatively cheap energy world.

Actuate’s “solution-based” approach essentially adopts the same features of the defense strategy: force amplification (mental and physical power), trustable data infrastructure discovery of social and science problems, remotely helping people to manage health conditions, genetic manipulation for diseases (intelligence), and enhanced artificial intelligence for command and control of resources and infrastructures. The additional constraints are ethical design and operations with constraints on privacy to get desired outcomes to societal problems. To achieve these goals, Actuate proposes applying a similar technological strategy as DOD to solve them.

The lack of public good status for robust health, inequity, and climate change poses challenges that do not just parallel those of national security. We are then forced to ask: If these societal problems are not treated as public good problems, how does this new privately led public-private partnership effectively solve a public good problem? Do we agree that goals we have chosen are the right ones to provide a space for very differently directed curiosity driven by barely possible science and engineering?” The question from a design perspective is to understand the nature of the problem to see whether it is a private good or a public good with social goals, as that would decide the institutional infrastructure and R&D to produce a solution-based approach for this effort to be successful.

Eswaran Subrahmanian

Arati Prabhakar identifies the cracks in the nation’s current innovation ecosystem and the power of solutions-focused R&D in repairing them. She pinpoints the need for creativity from outside private and government sectors and highlights evidence as a critical place from which new solutions to our complex societal problems should be launched. In essence, she has laid a road map for a new approach to solutions-oriented, multisector R&D, one that carefully articulates the realm of the barely feasible in which nonprofits must operate—a space perhaps allowing for more creativity and risk, but always working under limited resources. There are several additions that serve to augment this road map:

First, as the human-centered design firm IDEO describes, feasibility is only a part of the sweet spot of innovation. There are two other areas that are just as important: desirability and viability. The pairing of “barely feasible” and “deeply desirable” can help us understand where solutions R&D priorities should be placed—where there is a community need that is unmet. Compensating behaviors demonstrate that people are often the creators of their own solutions, but what if there is a critical infrastructural need not being met that can funnel into our solutions R&D approach?

Second, part of the solutions R&D road map should be to make visible the inherent complexities of the innovation ecosystem. The invisibility of science allows for incongruities such as GPS being used to navigate to a rally against big government, even though big government created GPS technology. We cannot continue to uplift a paradigm where science is made to feel like magic. Making scientific innovation visible and transparent would help to heal the deep mistrust in government, scientific institutions, and science itself. Part of reasserting the relationship between science and society will be one in which the efforts and cost of innovation—the money, time, material, and human resources—are not unseen. The invisibility of scientific innovation is intrinsically a danger to democracy.

People often see innovation as a wasteful activity, but it takes a lot of failures to find a potential solution. To move solutions R&D to the point of creating systemic advances, every next stage of the problem requires significant resources and the scaffolding of complex infrastructure. Problematically, the culture of innovation and entrepreneurship asks nonprofits to be fully end-to-end, responsible for all that falls under the implementation pipeline. Stopping at a prototype is not enough. The modality must instead be, as Prabhakar notes, one that is additive—resourced, financed, and taken to scale without the requirement of nonprofits bearing the full burden. Ensuring a culture in which solutions R&D can grow transparently and visibly across all sectors, pairing the barely feasible with the deeply desirable, and acknowledging the resource intensity of innovation, should become critically important road map features as we create a new solutions R&D ecosystem.

Shannon Dosemagen

Arati Prabhakar advocates solutions R&D: a new approach to innovation that promises to address the biggest problems of our day by looking for answers “in the realm of the barely feasible.” It sounds good. But her vision is too focused on technology and too wedded to the status quo to bring about transformative change.

The greatest opportunities for progress lie not in the realm of the barely feasible, but in the realm of the perfectly doable and the realm of the scarcely imaginable. Operating effectively within those realms requires calling in the expertise of individuals and communities traditionally excluded from innovation ecosystems.

This past winter, I taught a course on “Innovation and Social Justice.” Students’ capstone project was to propose an innovation that would address the problem of homelessness. New technology was on their minds when they asked guest speaker Liz Hersh, director of the Office of Homeless Services for Philadelphia, what it would take to solve the city’s homelessness problem. Hersh’s answer: money. She and her staff knew how to get people into housing. They just needed several million more dollars to be able to achieve it.

This is the realm of the perfectly doable. There is no need for a new app or better algorithms. The people closest to the problem know how to solve it. What they need is for their solutions to be recognized as efficacious and innovative. Even more importantly, they need their solutions to be funded even when they involve no seductive new technologies.

After speaking with Hersh, my students were incredulous that a problem in the realm of the perfectly doable was not being solved. One inescapable reason is a pervasive narrative that homelessness is caused by people’s laziness or poor decisions. For public resources to be devoted to the problem, we would have to stop blaming individuals and acknowledge structural factors such as gentrification, discrimination, and low wages. We would have to agree as a society that housing is a human right.

This is the realm of the scarcely imaginable. Achieving economic justice requires reexamining deeply ingrained but poorly conceived ideas about who has worth. Moving the needle on climate change and other environmental crises requires redefining notions of sufficiency and comfort to bring our consumption into line with planetary carrying capacity. Shooting for the feasible—even if only barely feasible—evades these necessary imaginative leaps. It supposes that new technologies can work within existing structures to create solutions, when in fact true solutions demand restructuring the way we live and think.

As in the case of homelessness, our ability to do the perfectly doable is too often undermined by our unwillingness to imagine the scarcely imaginable. Those well served by existing social structures, including professionals embedded in existing innovation ecosystems, have little need or perspective to question their foundational assumptions. Victims of structural injustices have unique insight into intractable problems and ample incentive to cultivate imaginative responses. Valuing their knowledge of the perfectly doable and their perspectives on the scarcely imaginable—that is, being thoughtful and inclusive about who should be involved in innovation, something that Prabhakar never addresses—is essential to any project that aims to solve society’s most important problems.

Gwen Ottinger

Arati Prabhakar has targeted the most important issues on the “endless frontier” of research. I am entirely in sympathy with her project. But there is a problem: corporate leaders are not currently trustworthy repositories of the knowledge generated by scientists and engineers. Vannevar Bush was perhaps more justified in his confidence in the use of the discoveries he promoted.

The creation of a method for protecting privacy while making personal data available for analysis is a laudable goal. But can we trust the giant corporations built on the exploitation of personal data to adopt the technology without perverting its application? We have abundant evidence of their inability to take reasonable precautions in the face of the exploitation of their services by malevolent actors. I am doubtful that they will change without the imposition of regulations they may well fight off for years to come.

The Boeing 737 MAX scandal was shocking, not just because hasty work and poor information led to disaster, but also because a well-established principle was violated. The airplane seems to have been shipped with two levels of safety, offering as an option certain supplemental equipment that might have saved lives. This appears to contradict the airline industry’s long-standing commitment never to offer safety equipment as a paid option. The violation of this principle suggests an inability of the management of one of the largest US corporations to act prudently, even where many years of traditional practice point the way.

Is this an isolated case? Unfortunately, it is not. The managers of several major German automobile companies fraudulently engineered their cars to frustrate regulators concerned with the impact of air pollution on public health. Oxycontin has taken the sometimes doubtful veracity of the pharmaceutical industry to apocalyptic heights.

Is this situation new? Not entirely. The lead and tobacco industries offer a dismal precedent, and Exxon is behind much climate change denial. Those industries were fighting against science for survival, even while harming their customers. But it is astonishing to find well-established industries with no urgent reason to cheat engaged in the same sort of chicanery on a huge scale.

We rely on economic leaders to transfer technology from the laboratory to the world in which we live. In recent years, some of those we trusted most have become unreliable allies of science and technology. I do not know the solution to this predicament, and I do not expect Prabhakar’s new organization to find it for me. There has been much recent talk about the role of ethics on the frontiers of knowledge, but it seems to me that a politics of technology is more urgently needed.

Andrew Feenberg

Arati Prabhakar argues that tackling contemporary social problems requires a new approach: solutions R&D. Yet by her own account, this is a well-established model. The Defense Advanced Research Projects Agency “has been doing solutions R&D well for six decades,” she writes. This approach even characterized the World War II Office of Scientific Research and Development under the leadership of Vannevar Bush.

Indeed, there is nothing quite so old in our technoscientific culture as calls for something new. Our obsession with the new is what makes us modern (from the Latin modo, “just now”). It dates back to Francis Bacon’s original vision of an R&D agency in his 1626 story “New Atlantis.” In it, a technocratic elite devised wonderful goodies to make life for the citizens of that utopian island long, healthy, happy, and full of pleasurable conveniences.

This was truly a new vision in Bacon’s time. Given the pervasive sickness, poverty, and hardships for most people in the seventeenth century, a turn toward technological innovation made a great deal of sense. Continuing to double down on yet more and ever faster innovation in the developed world in the twenty-first century, however, is not so clear cut. A genuinely new approach to innovation now would consider threshold phenomena such as diminishing returns and counterproductivity.

In other words, we need to be mindful of the ways in which solutions R&D can drift into “solutionism,” a dogmatic insistence on seeking a tech-fix to all problems. Without casting aspersions on the importance of R&D, we can wonder if, for example, apps based on big data and machine learning are really the best approach to diabetes and public health. Maybe it is better to promote living wages, eliminate food deserts, and improve school nutrition.

Prabhakar is right to flag the ethical dimensions of innovation, but she doesn’t give that sufficient weight. Her solutions R&D team needs an in-house Thoreau, who wrote in Walden about how “with consummate skill” we set our trap to catch comfort and independence only to get our own leg caught in it. The ethics of solutions R&D can’t just be about how to make innovations safer or more equitably distributed. It needs to also be about whether a tech-fix is appropriate in the first place. Without that, we risk getting more skillful at falling into our own traps.

When I think about what Vannevar Bush might make of today’s world, I bet he would marvel at just how much more stuff we have, witnessed by a gross domestic product (the value of all goods and services produced each year) that is eight times bigger than in his day. Then he would be surprised to discover how much more we seem to need. Invention, perhaps, is the mother of necessity.

Adam Briggle

2020 may go down as the most challenging year in a century. Yet the last century is also distinguished as being a time of development and successful deployment of solutions rooted in science and technology. In her must-read article, “In the Realm of the Barely Feasible,” Arati Prabhakar describes her plans to deploy the best scientific evidence to address some of the most intractable societal problems.

Prabhakar begins with a salute to Vannevar Bush, who produced the landmark report Science, the Endless Frontier. Undeniably, Bush was the catalyst that ignited a US scientific and technology revolution in the mid-20th century. But despite his immense intellectual prowess, Bush may not have been equipped to tackle the really hard problems the nation now faces. Prabhakar shows an affinity for hard issues: in her article she asserts, “And by hard problems, I mean hard.” Yet the solutions to hard problems need to go beyond hard science. Views of the importance of “hard” and “soft” science are changing. Some years ago, the widely regarded researcher and author Jared Diamond suggested that disciplines that perform experiments using manipulation of variables are the “easy” sciences. Those that study broader social and ecological variables in natural environments are actually the hard sciences. Writing in Population Health: Behavioral and Social Sciences Insights, Nancy Adler and Aric Prather explain that beyond the traditional hard sciences, systematic studies that take on the complexities of the human experience are the “harder sciences.”

Bush held a fairly narrow view of hard science. As an engineer, he concentrated on linear mechanistic models. This led to remarkable advances, but the models were of limited value for understanding complex systems that often characterize the biological, behavioral, and social sciences. To be fair, relatively simple linear models can be informative. For example, efforts to control infectious disease have benefitted greatly from understanding the invading pathogen, modeling the immune response, and purifying antigens that provoke neutralizing antibodies. Witness the recent success in developing coronavirus vaccines.

But there are countless cases in biomedical science where targeting a biomarker has no effect or even harmful effects on human health. In type 2 diabetes, simplistic models that concentrate only on lowering blood sugar have had underwhelming effects on diabetic complications, including mortality. Climate science, studies of social mobility, and public health require a much wider range of soft-science investigative strategies. Bush was not secretive about his disrespect for the “softer sciences.” One of his first moves when he came to the Carnegie Institute of Science was to get rid of the archaeology program. In addition, he reduced funding for the major journal that concentrated on the history of science and its cultural influence.

Our current challenges require an integration of the lessons from the Vannevar Bush era with the recognition that big problems require broad approaches. Population health, social mobility, and climate change epitomize complexity science. Understanding these complex nonlinear systems requires a harder science that builds on methods and models from a range of disciplines. It is encouraging that Prabhakar and Actuate are willing and ready to mix hard and soft science approaches to apply harder science solutions to our most challenging societal problems.

Robert M. Kaplan

A fine diagnosis, but what about the cure?

In the world of research and innovation policy, few institutions consistently attract positive international attention. The Defense Advanced Research Projects Agency is among this select group, widely celebrated for its model of public-private collaboration in the promotion of innovation.

Arati Prabhakar is a former director of DARPA, so we shouldn’t be surprised that her diagnosis of the challenges facing the nation is both sharp and convincing. As she suggests, we need to understand familiar problems in new ways. We need new methods, new participants, new incentives. We need an approach that can tackle pressing societal challenges. We need creative experimentation. Public institutions should play an important role. And let us indeed include “ethics from the start.”

Prabhakar’s diagnosis is hard to fault. My doubts begin when we move to the innovation cure. Prabhakar is careful to note that we do not have “an instant solution to all the problems of the world.” Panaceas, as she writes, are not around the corner—and any innovation solution can be only a step in the right direction.

But take Prabhakar’s example of “data and information we can trust” as a challenge. How might we potentially go about tackling this? Transparent and accountable governance structures seem essential. And education could be crucial too: how can we help young people (and ourselves) cope with relentless pressure from social media? If we are to bring in ethics from the start, we might especially try to engage disadvantaged, hard to reach, and vulnerable groups in a reflective dialogue. The list of possible responses is both long and sobering, all the more so given the global nature of the problems. However, the indicative solution that Prabhakar offers is much more limited: a data-sharing system that can allow statistical analysis while maintaining privacy.

Taking a second example from the article, the gap between problem and solution is even more apparent when it comes to “robust population health.” The difficulties of the US public health system are well-stated in terms of both spending and outcomes. Still, offering smartphone-based coaching systems as an individual solution to a huge societal challenge risks minimizing the deeper issues that Prabhakar so clearly presents. If the public health system is diagnosed as “weakened,” shouldn’t the suggested cure directly address the underlying ailment?

One could say that every little innovation helps. But the danger is that we overplay the degree to which marketable innovations can solve large social, political, and environmental problems. And we end up looking in the wrong places for the answers.

It is good to be optimistic. And we need solutions that, partial as they are, can take us forward. By all means, let us build steering mechanisms and brakes into technology. But let us be sure that innovation is part of the cure and not another manifestation of the challenges we face.

Alan Irwin