The Washington PostDemocracy Dies in Darkness

Half of academic scientists leave the field within 5 years, according to a new study

(Illustration by Christopher Ingraham for the Washington Post)

If you’re trying to understand changes in how scientific knowledge is produced and distributed in recent decades, you could do a lot worse than studying the chart above, which is from a new paper published in the Proceedings of the National Academy of Sciences.

The chart shows what the paper’s authors call the half life, meaning the time it takes for half of researchers active in a given year to abandon the field, for the scientific disciplines of astronomy, ecology and robotics.

Back in the 1960s and early 1970s, researchers in astronomy and ecology had a half life of well over 30 years. Roughly half the cohort of astronomers publishing scientific papers in the 1960s, in other words, would still be actively publishing 30 years later. The median scientist could expect a publishing career spanning several decades.

But since then, academic science careers have been shrinking. In 1990, researchers in the three fields studied could expect to keep at it for 15 to 25 years. By 2000 it was five to 10 years.

Now, the typical researcher active in those fields in 2010 can expect to stop publishing after a period of about five years. “We find dramatic shortening of careers of scientists across all three disciplines,” the paper’s authors write.

That reflects a fundamental shift in how scientific knowledge is produced, the authors say. They derived these findings from a database of articles published in the major journals for astronomy, ecology and robotics. Their final data set included more than 100,000 published authors. For each author the researchers were able to track a rough career trajectory based on when their first and final works appeared in the database.

In the traditional model of academic research, universities produce graduate students who get PhDs and then obtain a tenure-track position that allows them to produce and publish research findings throughout the course of their careers.

But that model has broken down in recent decades. Universities now produce far more PhD graduates than they do tenure-track positions. Some of those PhDs may transition directly into private industry jobs after graduation, but those who remain to pursue a traditional academic career face increasingly stiff odds of success. And many of them appear to be dropping out of the academic path after investing considerable amounts of time and resources in it.

“These are smart and talented people who end up finding ‘good’ jobs,” said Stasa Milojevic, lead author of the PNAS study, “but often after having had to reboot their careers late in their lives (which makes them less competitive in these new positions) and after having had to contend with long years of uncertainty and oftentimes professional disappointment.”

Another factor is the growing size of the typical academic research team, Milojevic and her colleagues write. Research in the sciences tends to be highly specialized, with labor in any given project increasingly likely to be divided among individuals focusing on narrowly defined parts of the process.

“The growth of scientific team sizes is being accompanied by a transition in the organization of scientific work from craft to bureaucratic industrial principles, with increased division of labor and standardization of tasks,” Milojevic and her colleagues write. “The result is a growth of scientists whose function is to support the projects that others are leading.”

Milojevic and her colleagues have found that the percentage of scientists who will contribute to a project as a lead author — meaning the researcher in charge of the study and its publication — has shrunk dramatically. In the 1960s, for instance, about three-quarters of researchers in astronomy and ecology appeared as a lead author on a study at some point in their careers. By 2010 that fraction had dwindled to about 40 percent.

What that means is that the typical scientist is now less likely to spend time pursuing their own research projects and more likely to work in support of a more established scientist overseeing numerous other researchers. Milojevic said it’s similar to the “superstar” phenomenon we see in so many other fields: big, established names gobbling up an ever-larger share of money and talent.

“Other studies, including our previous findings, have shown the operation of rich-get-richer mechanisms in many aspects of science, from the growth of teams (where more productive teams tend to grow much quicker and become ever more productive) to the way citations and recognitions are accrued,” Milojevic said in an email.

While all of this may be creating a more challenging professional environment for individual scientists, it’s not necessarily bad for science itself. “It is almost impossible to do certain types of research as an individual or with only a handful of students,” Milojevic said. “In the vast landscape of modern science the specialization is also inevitable and necessary. So, this new mode of knowledge production is definitely making great contributions to advancing knowledge.”

She cautioned, however, that traditionally it has been “the very small teams which are pushing the frontiers of science and are the incubators of new knowledge.” Whether that type of knowledge will continue to be cultivated depends, in large part, on whether universities are able to update their career pathways to reflect how science is actually being practiced today.