What if key elements of science policy are based on patterns of discovery that no longer exist?
That's the question behind a paper (abstract available here) released Monday by the National Bureau of Economic Research. The paper -- by Benjamin Jones, associate professor of management at Northwestern University -- argues that science has changed in key ways. Specifically, it argues that the age at which researchers are able to make breakthroughs has advanced, and that scientists are parts of increasingly larger teams, encouraging narrow specialization. Yet, he argues, science policy (or a lot of it) continues to assume the possibility if not desirability of breakthroughs by a lone young investigator.
Much of the paper focuses on the greater difficulty of making key contributions -- solo -- early in one's career. Jones cites, for example, the growth in the number of journal articles. In 2006, for example, there were 941,000 journal articles published, 90 percent of them in science and engineering, and these articles cited 4,372,000 unique journal articles. With the publication rate growing by 5.5 percent a year, someone able to read only a certain number of articles a year is seeing his or her "fraction of extant knowledge" decreasing by the same percentage.
He then reviews a variety of measures that show the twin trends of an aging and more group-oriented scientist. On age, he notes that:
- During the 20th century, the average age at which researchers made the accomplishments that were later honored with Nobel Prizes in physics, chemistry, medicine and economics increased by 5.83 years.
- During the 20th century, the average age at which researchers made "great" technological achievements rose by 4.86 years, while the age of those achieving a first patent -- for more average inventors -- went up by 6.57 years.
In exploring the data more fully, Jones finds that the gains are not the result of people living longer, but generally of a decline in "great achievement" in scientists' 20s and 30s. "Peak productivity has increased by about 8 years, with the effect coming entirely from a collapse in productivity at young ages," Jones writes.
Then Jones looks at the teamwork trends, focusing on data for journal articles (since 1955) and patents issued (since 1975). He finds here that the mean size of teams is increasing (across both categories) by 15-20 percent per decade. The growth is nearly universal across scientific fields (including the social sciences). By 2005, he finds that more than 80 percent of science and engineering publications, more than 50 percent of social sciences publications and more than 60 percent of patents had multiple authors or inventors.
While Jones notes that some might view the question of collaborating or working individually as a choice, he argues that the most influential science is increasingly done in teams. Between 1995 and 2005, he writes, group papers had twice as many citations as those authored by individuals. And comparing "home run papers" -- those with at least 100 citations -- team papers are more than four times as likely to produce such impact.
A series of changes may be needed in science policy to reflect these shifts, Jones argues. He notes some efforts already under way at the National Institutes of Health and other funding agencies to reward interdisciplinary research and work by young scholars.
But he also argues for changes in the mindset about science, as expressed in the way discoveries are honored. "Celebrated achievements historically often carry the scientist’s name – Euclidean geometry, Newton’s laws of motion, Mendelian inheritance, and the Heisenberg uncertainty principle, to name a few," he writes. Those who award prizes and whose opinions confer prestige should consider that "shifting toward high status and/or financial reward 'team prizes' for particular innovations could help undo the incentive challenges that individual rewards impose," Jones writes.