For Researchers, Risk Is a Vanishing Luxury

A couple of years ago, as science was deep in a rut of flat federal financing, Roberta B. Ness, vice president for innovation at the University of Texas School of Public Health, toured the country, urging scientists to take on innovative, high-risk research. Such work, though prone to failure, can overturn whole scientific paradigms. Expand assumptions, she said. Change points of view.

Science has always balanced traditional lines of investigation with radical, divergent hypotheses. Follow her methods, she told them, and they could become risk takers themselves.

Yet during each question-and-answer session, Dr. Ness, a professor of public health with an M.D., faced a battery of questions, mostly from early-career scientists in the back of the hall. The replies were so similar, they melded into a chorus.

Your ideas are encouraging, they’d begin. But I can’t do this. If I were to propose really innovative ideas, I’d never get a promotion. I’d never get tenure. I’m completely blocked. Is something being done about that?

Each time, Dr. Ness found herself hard-pressed to respond. This was a systemic problem. How could anyone start to fix it?

A basic mission of the American research university is eroding, Dr. Ness now says in her latest book, The Creativity Crisis. The U.S. government chose universities as the home for high-risk research after World War II, due to their tolerance for science that might fail or produce elusive returns.

Since then, though, the American scientific system has evolved to where, at almost every level — university hiring and promotion; publishing; the awarding of government grants — predictability is prized over boldness. Just as it’s become more difficult to express provocative or risky views on campuses, the research conducted there has grown more conservative.

It’s a trend that began decades ago, as universities came to rely on, and shape, federal grants from agencies like the National Institutes of Health and the National Science Foundation, which provide a majority of the country’s research dollars. The flat financing of the past decade has only made it worse.

Universities increasingly judge faculty members on not just their research or teaching, but also their ability to pull in dollars. They hire scientists who pay their entire salaries through grants, an employment deal that breeds incrementalism. They use metrics biased toward short-term productivity. To guide their hiring, they use peer evaluations, which, research has shown, are naturally hostile to radical ideas.

“The lack of risk taking and associated conservatism is one of the most dispiriting aspects of modern university life,” said Andrew F. Read, a professor and director of the Center for Infectious Disease Dynamics at Pennsylvania State University. “I don’t see too many people in leadership roles worrying about it.”

The American scientific system is far from failing, said Thomas R. Cech, a Nobel laureate and professor of biochemistry at the University of Colorado at Boulder. Crisis rhetoric is overblown. It still produces remarkable discoveries. But that’s solely because scientists, by their nature, push the verge. “That’s what’s saving the system,” he said. “The incentives are pretty much all in the opposite direction.”

As World War II drew to a close, the United States faced a crucial decision about its scientific future. University scientists, supported by federal money, had been essential to victory, developing radar and the atomic bomb, among many advances. If the country was to cement its world leadership, it would need to continue financing scientists, argued Vannevar Bush, the head of the government’s wartime research and development, in an influential report to President Harry S. Truman, “Science — The Endless Frontier.”

Where to house that research? Bush had a clear answer: universities, medical schools, and stand-alone research institutes. “These institutions,” he wrote, “provide the environment which is most conducive to the creation of new scientific knowledge and least under pressure for immediate, tangible results.” Such basic research would naturally entail risk.

This was truly a different country in 1945. Only 10 to 15 research universities existed; now there are more than 100. The United States spent just $31 million on university science in 1940, which, even when adjusted for inflation, is less than 1 percent of the $67 billion now spent on university research and development.

At first, the government had to recruit scientists, who had previously turned to foundations or endowments, to apply for financing. But universities and scientists adapted quickly, building out sophisticated lobbies, according to Paula E. Stephan, an economics professor at Georgia State University who studies the structure of American science policies.

Indeed, if a contract existed between the government and universities to reward high-risk research, it was the universities who broke it, orchestrating a system that, by the 1960s, had increased the proportion of a grant that would go to the university for its overhead costs; pushed graduate students toward financing their education through research grants, in addition to fellowships; and allowed faculty members to pay parts of their own salaries with federal money.

That structure, if not fed by continual growth, would begin to break down, Ms. Stephan said.

This is not to say that all scientists should pursue esoteric theories and odd hypotheses. “Risky research is not the same as excellent research,” Ms. Stephan said. “As a physicist told me years ago, there’s a real need for ditch diggers, in a nonpejorative sense. We need people to keep working on areas where we’ve made progress and extending it.”

It’s a bedrock theory of science that researchers must balance their pursuit of traditional and novel work. Most experiments follow known traditions, and rightly so: Shared methods building on a known tradition make science work. Advances often spring out of irregularities popping up in the next sensible experiment. And nearly every sociological study of science has shown that genuine breakthroughs come from only 10 to 15 percent of the population. Scientists, like the rest of us, are normally distributed.

What the system needs to do is be certain that the innovative 10 percent can survive. Many researchers wonder whether famed scientists of the past, if they were transported, early in their careers, to the present, would survive. Sidney Farber, the pioneer of chemotherapy, would have been forced out when his first round of treatments exacerbated leukemia, Dr. Ness said. Roger D. Kornberg, a Nobel laureate, has testified to Congress in recent years that his own prize-winning research, on how DNA is transcribed in complex cells, would not have passed muster.

Those are the people science can’t afford to lose.

“We should think of it as getting our portfolio right,” Ms. Stephan said. “Right now we’re overinvested in what we call safe bonds.”

The productivity clock is always ticking in modern science.

From the moment a budding scientist begins her graduate studies, the limited time horizons of career development prevail. Say you’re a biologist who wants to land a postdoc in a top lab. How? Do something central to what biologists are already thinking or doing, often building directly off your mentor’s work. Land that postdoc, and then you have a few years — enough time to place a relatively innovative paper in a prestige journal, perhaps, but not enough to risk creating a new way of investigation.

It was that conservative culture that prompted one neuroscience postdoc at Duke University, Jean-François Gariépy, to very publicly quit academe, a decision he announced on Facebook several months ago.

“By creating a highly competitive environment that relies on the selection of researchers based on their ‘scientific productivity,’ as it is referred to, we have populated the scientific community with what I like to call ‘chickens with no head,’” he wrote, “that is, researchers who can produce multiple scientific articles per year, none of which with any particularly important impact on our understanding of the world.”

After the postdoc comes the hiring committee. Every department wants to hire great talent. But few committees would dream of hiring scientists who had pursued a line of risky research that failed. Faculty members instead favor established fields they know and can easily assess.

Hiring committees can also be tempted to rely on misleading metrics to winnow the hundreds of applications on their desks. The metrics are typically based on citation counts and can be biased against researchers pursuing high-risk ideas.

Indeed, a forthcoming analysis, by Michael J. Kurtz and Edwin A. Henneken at the Harvard-Smithsonian Center for Astrophysics, will show that, at the time junior faculty members in the sciences are hired, their citations are practically worthless in predicting their ultimate success, as judged by prizes and long-term citation counts.

Hired into an assistant-professor job, a scientist finds the tenure countdown gets louder. Within a few years, departments expect researchers to begin winning external grants. A scientist who doesn’t begin bringing in money will almost never receive tenure. Promotion decisions will often be delayed until a researcher has renewed a grant successfully.

That risky idea you’ve harbored since graduate school? It is likely to become an “after tenure” project, said Jacob G. Foster, a physicist turned sociologist and assistant professor at the University of California at Los Angeles.

At one major private research university on the East Coast, the promotion bar has been set so high that it’s killing morale, said one junior scientist, who spoke on the condition of anonymity since he is still pursuing tenure. The university won’t provide bridge funds to support junior scientists, despite a sizable endowment, as earning a grant is vital to winning tenure, he said.

Rather than pursuing a high-risk hypothesis that may take years to prove, it’s far better to master one method and turn out a series of similar papers. “These developments increase anxiety while stifling creativity and innovation,” he said.

If a scientist does secure tenure, he or she may be 40, and this will be the first time the person has had reasonable time to explore research that may not bring immediate returns. What habits have been formed? “Have you trained the ability to do something more out of yourself?” Mr. Foster said.

It’s particularly tough in biomedical labs, with their large support staffs. “If you propose a crazy thing for your next NIH grant,” Mr. Foster said, “there could be a number of people out of a job.”

Few take such risks. Instead, entrepreneurial scientists balance their own portfolios, financing riskier projects on the side through play-it-safe grants. Everyone hedges.

“That might be the most insidious part of it,” said Mr. Read, the Penn State researcher. “Everyone is evidently staying well clear of serious risk. If everyone lives, no one is climbing dangerous mountains.”

Evidence of just how much researchers are playing it safe has grown in the past few years, as social scientists have found new ways to study risky research, Ms. Stephan said. “We’re beginning to find that there’s a bias against novelty.”

Several years ago, work by Pierre Azoulay, an associate professor of economics at the Massachusetts Institute of Technology, and several collaborators compared scientists financed by the NIH’s traditional project-based grants with investigators supported by the Howard Hughes Medical Institute, which finances the researcher, not the project.

The scientists were all top performers, but, along with publishing higher-impact results, the Howard Hughes scientists, given freedom, shifted their research, the Azoulay team wrote, “in ways that suggest the program induces them to explore novel lines of inquiry.” In other words, the traditional system seemed biased toward conservatism.

Research forthcoming in Management Science, by Kevin J. Boudreau, a visiting assistant professor at the Harvard Business School, and three colleagues looks at peer review, the fundamental mechanism used, in various guises, to screen grant applications, papers, and hiring. Soliciting 142 scientists from a research-intensive medical school to evaluate 150 different grant proposals, they found participants systematically scored novel proposals lower.

So incentives may deter high-risk research, but that may not mean the actual work has changed in recent years. Is this just how science has always been practiced? Intrigued by that question, Mr. Foster, the UCLA sociologist, and three colleagues have found a new way to judge the balance between tradition and innovation in science. And it suggests that in recent years science has become more hidebound.

Their method, put simply, involved scraping the abstracts of millions of biomedical papers published from 1983 to 2008. Examining papers that dealt with biochemistry, they built a network, with each chemical compound as a node in it; when two molecules were mentioned in the same abstract, they added a link between the corresponding nodes. By 2008 the network had 181,078 nodes and more than 10 million links.

Analyzing such networks in papers published late this year in the American Sociological Review and the Proceedings of the National Academy of Sciences, the team found scientists increasingly exploring existing links, rather than venturing out in a rapidly expanding terra incognita. From 1987 to 2008 that preference increased as much as sixfold.

The conclusion carries caveats. Risky research is more likely to fail, and scientists don’t often publish failures. The findings may not extend to other disciplines. Also, it’s possible that scientists have found all the important biochemical combinations. Maybe the clusters of investigation are it; the unexplored potential links don’t correspond to the structure of the natural world.

Mr. Foster is the first to insist on the importance of tackling the next logical experiment. But if the first objective of science is raw discovery, then the current system is inefficient.

“It’s a question of values,” Mr. Foster said. “Is this what we want? Yes or no. If not, how do we go somewhere else?”

When it comes to a system as successful as American science, revolutionary change is rare. But on the margins, advocates for increased risk have scored a few victories.

Those innovations start at the financing agencies. Though under pressure to demonstrate productivity to Congress, the NIH and the NSF have grown worried about conservatism among the researchers they back. The NIH now has four modest grant programs aimed at supporting high-risk research, including one that copies the Howard Hughes model. The NSF also has a program to foster high-risk research, begun in 2009, but the agency’s own officers have been reluctant to use it.

The agencies have been proactive in other ways. The NIH has reformed its policies to better account for scientific practice today. It requires biographical sketches in grant applications that paint a researcher’s background and framing, not just a list of publications. And one of its institutes is experimenting with grants that last for five years by default, allowing more freedom to fail before useful evidence is needed to win renewal.

At the universities, Dr. Ness and others have pushed for smaller reforms. As she worked on her book, Dr. Ness was wrapping up her tenure as dean of the University of Texas School of Public Health, and she took several steps then to promote risk, she said. She and her colleagues conducted an “innovation search” for early-career, interdisciplinary scientists who had created “surprise in the service of health and prosperity.”

Sorting through 225 applications, they hired four researchers deemed to be conducting frame-shifting work in, for example, environmental modeling or the interaction between genes and behavior. She also began a modest internal grant program that gave scientists up to $25,000 each to explore creative research.

She swiped that idea from Max Michael III, dean of the School of Public Health at the University of Alabama at Birmingham, who for eight years has accepted high-risk ideas from his faculty members written, literally, on the back of No. 10 envelopes. No budget or references — just an idea.

So far, for about a $250,000 investment by the university, members of his faculty have secured $4 million in external grants based on work he helped initially finance, he said. They’ve developed a smartphone app for more accurately judging body-mass index and an online survey measuring health issues among Muslim women in the United States.

“This is something you can do at a dinner table,” Mr. Michael said, “before you have your cocktail.”

Some universities, especially wealthy ones, have instituted similar internal financing. For several years, Princeton University’s School of Engineering and Applied Science has given out $200,000 grants for risky research. And thanks to a $100-million gift from Gordon and Betty Moore this past summer, the California Institute of Technology will begin supporting many of its graduate students without federal financing, allowing the students more choice in the experiments they pursue, rather than relying on grants funneled through faculty members.

Still, few universities have thought about changing their requirement that tenure candidates carry substantial grant financing. But if this system is pushing innovators aside, then universities may need to soften that requirement, said Ronald J. Daniels, president of the Johns Hopkins University, in a commentary published this year in the Proceedings of the National Academy of Sciences.

“If we truly believe that process is leaning away from young scientists, one remedy is to relax the custom,” he wrote.

A few other changes could be useful, Mr. Foster added. Universities can create large departments, which allow more risk on individual hires; in a large department, one dud hire is unlikely to sink its reputation or mean there’s no one available to teach its essential courses. They can rein in their growing fascination with short-term tallies of productivity. And universities, along with the federal agencies, can tweak peer review so that it relies less on consensus. The Bill & Melinda Gates Foundation, for example, allows its reviewers to champion a proposal despite peer objections.

“Instead of avoiding irritating anyone,” Mr. Foster said, “all you have to do is excite someone.”

Seven years ago, a panel of scientific luminaries led by Colorado’s Mr. Cech issued a stern warning to universities, in what is known as the “Arise report.” Update policies, the report urged, so hiring decisions don’t always favor scientists who develop sure bets. Stop penalizing scientists who find negative results despite sound experimental plans. Reward quality over quantity when it comes to publications.

Since then, little has changed in what universities reward. “I have not seen any major response to our recommendations in that area,” Mr. Cech said.

Mr. Cech knows a few things about supporting high-risk research — he ran the Howard Hughes Medical Institute for a decade. But he doesn’t see much hope for university science to increase its support for high-risk work anytime soon.

The most innovative models, he said, are institutions unaffiliated with universities, with support from philanthropists or technology companies, like the huge endowment behind the Stowers Institute for Medical Research or the money Alphabet, Google’s parent company, is putting into Calico, its anti-aging venture.

Indeed, now that he’s back at a research university, Mr. Cech said he would be reluctant to give people tenure if “they can’t pull their weight on funding.” His institution has not changed its policies to tolerate more risk from its professors. “And, in spite of what we said in ‘Arise,’” he said, “I’m not sure they should.” Boulder simply can’t afford such a change unless the American research system rewarded it.

Science is a balance. It will never be perfectly efficient, as in Mr. Foster’s model. But every day — because of the incentives universities and funding agencies create — sees an invisible loss as bold ideas are postponed, ignored, or never even conceived.

Even simple shifts could help, Mr. Cech added. Take the shorthand they used at Howard Hughes, inspired by a tool in genetics: the deletion test. They didn’t just look for productivity. They looked for paths not taken. “If this person was deleted from the scientific landscape, what difference would that make?” Mr. Cech said. “That puts things in a very different discussion.”

Paul Voosen is a senior reporter covering the sciences. Write him at paul.voosen@chronicle.com; follow him on Twitter @voooos; or see past work at voosen.me.