Do the Deaths of Top Scientists Make Way for New Growth?
New ideas advance in science not just because they are true, but because their opponents die, physicist Max Planck wrote in 1948. He was referring to a fundamental theory that, at the time, provoked a nasty feud, yet today is taught in nearly every high school physics classroom. The belief that science advances one funeral at a time is the kind of folk mythology in which any researcher might indulge in a discouraging moment, says Kevin Zollman, a philosopher of science at Carnegie Mellon University.
“It’s very comforting to imagine there’s some evildoer behind the scenes,” Zollman says wryly, “when your paper gets rejected.”
Older scientists aren’t notably worse at accepting revolutionary ideas compared to younger colleagues, research has found. But a paper published in August in the American Economic Review suggests there may be subtler ways in which the top dogs have a discouraging effect on new entrants. According to the paper, which draws on decades of data on more than 12,000 elite biology researchers, when a superstar scientist dies their field sees a small burst of activity in the form of fresh publications. What’s more, the authors of the new papers, which are more likely than usual to be highly cited, are typically newcomers who have never published in this subfield before.
The results imply that the deaths of important scientists may open up opportunities for fresh ideas, reaffirming Planck’s statement. But they also suggest that science is reassuringly robust; instead of fields getting into a rut, or even falling apart when a star dies, they continue to evolve.
The work that led to the new study began around 15 years ago, when the economists Joshua Graff Zivin of the University of California, San Diego and Pierre Azoulay of the Massachusetts Institute of Technology began investigating what happens to people who have published papers with biology superstars after the stars die. The researchers defined these elite scientists by criteria such as how highly they were cited, how well they were funded, how many patents they had been awarded, and whether they were members of the National Academy of Sciences. The researchers expected that after a death, the star’s collaborators might jockey for position in the field. Or they might nominate a particular member of the group as the new leader.
But looking through the subsequent publications of each collaborator, the economists were surprised. “We did not find that,” says Graff Zivin. “We saw that everyone who ever wrote with the star published less and less important work after the star had passed.”
Perhaps the star was the source of the innovative ideas in the group, then. But that left the economists wondering about the state of the star’s subfield as a whole. Did the death kill their little corner of biology as well? Or did other people who’d never worked with the star come in and pick up the slack?
For nearly a decade, the two researchers and their colleague Christian Fons-Rosen, an economist at the University of California, Merced, collected detailed data about scientists’ careers and considered the problem. A crucial breakthrough came when they devised a way to define subfields in biology using keywords in PubMed, the National Library of Medicine’s literature database. Papers using the same keywords represented the community of biologists looking at a particular research problem, whose members nevertheless might not all be collaborating or writing papers together. This allowed the economists to look at stars’ collaborators and other researchers separately. Then, the economists identified 452 stars who died prematurely while they were still active between 1975 and 2003 and asked what happened to the stars’ subfields afterwards.
In the first two years after a star’s death, publications in their subfields increased modestly. But as the years passed, breaking the numbers down by author showed a startling change: Papers by newcomers grew by 8.6 percent annually on average. At the same time, papers published by collaborators took a nosedive, decreasing by about 20 percent a year. After five years, growth from newcomers was so substantial, it made up for the deficit from the collaborators.
In other words, large swaths of these fields had essentially been turned over.
One possible interpretation: The new arrivals could be upending orthodoxy after the death of dogmatic elders. It’s an attractive narrative — who doesn’t like a tale of crusty arrogance defeated by fresh creativity? But Azoulay cautions: “We don’t really know whether they are coming in and turning against the superstar.”
The economists didn’t contact any of the newcomers to ask about their motivations or their timing in entering the field. Instead, the economists tried to get more detail about the nature of these changes indirectly by looking at how the newcomers cited other research in their papers. Intriguingly, newcomers referenced more work outside the subfield than usual, and they were also much less likely to cite work from the star who had died. “It’s still recognizably the same subfield,” says Azoulay, “but they are injecting it with different ideas.” The new papers also tended to be more highly cited than other work the newcomers had done in the past, suggesting scientific peers found the work useful. But Azoulay, Graff Zivin, and Fons-Rosen found no sign that these new papers were particularly disruptive.
In fact, Azoulay thinks these changes are probably not revolutionary — not paradigm shifts, as philosopher of science Thomas Kuhn dubbed transformative discoveries in his landmark 1962 book “The Structure of Scientific Revolutions.” Rather, the newcomers’ work is probably more like what Kuhn called the progress of “normal” science: the gradual morphing of consensus. What the numbers could be showing, then, is a detailed view of a scientific field’s natural evolution, paper by paper, over the course of decades.
Still, changes do come in the wake of a noteworthy death. “It remains significant that even the trajectory of normal science exhibits a dependence on the presence of individual scientist stars,” said Dean Simonton, who studies the psychology of science and scientists at the University of California, Davis. The stars did not seem to be obviously excluding others — only a tiny fraction were on journal boards or decision-making panels for grants, which would make them explicit gatekeepers when it came to scientific funding and publishing of new ideas.
But that would hardly be necessary, points out Aaron Clauset, a physicist and complexity scientist at the University of Colorado Boulder, who has studied the career trajectories of academics. “A superstar scientist doesn’t have to exert power themselves in order to be powerful,” he says. Many members of a scientific community are invested in the success of a star’s ideas, which are accepted because they’ve allowed research to advance. While the star lives, attention will be focused on them; ideas that don’t mesh with theirs are perhaps unlikely to get easy acceptance.
One driver of the pattern may be the exertion needed to get a new idea into a tight-knit field, suggests Zollman, the philosopher of science. “Scientists usually have way more ideas than they have time to work on,” he says. They might ask themselves, he adds, as they decide what to do next: “Is this so controversial that’s it going to take a lot of energy to get traction?” And outsiders may not be sure if their new approaches are useful and worth sharing.
But growing research suggests science benefits when a variety of methods are in play. “It’s really important to a given scientific field that there be more than one methodology available at a time,” Zollman says. The findings in the American Economic Review paper are phrased carefully, he adds, but they suggest “that there’s a danger when you have scientific superstars that they might needlessly reduce the diversity of methodologies in a field in a way that might be hampering scientific progress.”
Numbers from the National Institutes of Health (NIH) show that scientists are older than they used to be when they receive their first big research grants — in 1980 the average age was around 36, while in 2016 it was 42. The age researchers start their own labs has also been rising, as young scientists spend more time in the holding pattern of postdoctoral positions or leave academia for other pursuits.
A 2016 study from the NIH and the National Science Foundation (NSF) asked why older scientists received so much more grant money and found that younger scientists simply applied less often. These younger scientists may not have thought they had a decent enough chance to go for it.
The new study is silent on exactly what might have been going on in the heads of the scientists, established and otherwise, who created the patterns over the 30-year period covered by the data. It’s impossible to say whether newcomers knew about the death and thought it was a good time to act or had no inkling of the star’s recent passing. It’s also not clear if some of the patterns may have come from journal editors who, dealing with a dearth of papers from the stars and their collaborators, started to admit new voices to fill the space. And the study doesn’t explore the possibility that the impetus might have been indirect, perhaps from a colleague who knew about the death and encouraged others to put forward new ideas.
Without more detail about why these patterns emerge, it’s hard to say what they mean for science. Still, perhaps one takeaway from the new research is that stars, who are often lauded for good reasons, may have more of a restraining effect than they know, says Zollman. The social structure of science, in which those with powerful ideas rise to the top and then stay there, may make a field less of a strict meritocracy than we usually imagine.
“Scientists love to talk about science and the social processes that go along with it because we just feel very viscerally that these social processes do shape the direction and taste and texture of science,” says Clauset. The paper confirms that feeling. Still, he goes on, the fact that fields neither fall to pieces nor experience cataclysmic shifts after a star departs is, in some ways, a testament to science’s ability to evolve slowly in the right direction, with an infusion of new voices gradually changing the status quo.
“In the long run,” Clauset suggests, “science figures it out.”
Veronique Greenwood is a science writer whose work has appeared in The New York Times, The Atlantic, National Geographic, and Aeon, among many others.