Growing inequality in the United States was a major media and policy topic well before President Obama gave a speech about the issue on 4 December 2013. People generally think of it in economic terms: the widening income gap between the 1% and the rest of us, especially those in Mitt Romney’s infamous 47%. New research, however, finds that inequality is growing not only in the economy but in science, too. Just as a relatively small number of zillionaires are scarfing up more and more of the nation’s earnings, a top echelon of scientific “stars” are scoring a growing share of scientific publications.
This picture of growing distance between scholarly “haves” and “have nots” emerges from a study by Ajay K. Agrawal of the University of Toronto (U of T), John McHale of Queen’s University, and Alexander Oettl of the Georgia Institute of Technology, which examined decades of research output by everyone publishing in a particular scientific field. Entitled “Collaboration, Stars, and the Changing Organization of Science: Evidence from Evolutionary Biology,” the study defines stars as researchers with publication records that place them among the top 10% of producers in their field.
Who are those well-connected researchers? The authors suggest that they are top professors with large numbers of former grad students and postdocs located at institutions far and wide, but still closely aligned with their former mentor’s scientific interests.
Evolutionary biology is a small enough scientific specialty to permit the tracking of the entire research output over a period of decades. The authors offer no evidence that the trends they identify extend to other fields, but the causes they cite seem universal, so there’s every reason to suppose that similar trends are happening across all of science.
The analysis revealed an apparent contradiction that the authors initially found “puzzling.” The concentration of research, though “increasing at the individual level,” is simultaneously “declining at the department level,” they write. Between 1980 and 2000, “the fraction of citation-weighted publications produced by the top 20% of departments falls from approximately 75% to 60% but over the same period rises for the top 20% of individual scientists from 70% to 80%.” Apart from these “seemingly contradictory trends,” the authors also note a “broadening [of] the base of science” as more lower-ranked institutions are represented in publications.
What’s going on? How can top-producing scientists account for a greater share of publications at that same time that their departments account for a smaller share? “Collaboration offers a possible explanation,” the authors write. Star scientists—indeed all scientists—are working with more collaborators than in the past, the study shows. What’s more, collaborators are located farther apart from each other—in both miles and ranking for institutional prestige—than ever before.
Twin changes spur trend
Two forces appear to be driving the growth in long-distance collaborations, the authors argue. First, the “burden of knowledge”—the total amount of available scientific information—grows relentlessly so that any given individual, no matter how stellar, can master only a dwindling share of all that is known. As specialization increases, so too does the value—indeed the necessity—of working with experts in fields and subfields outside one’s own.
Even as specialization drives scientists apart intellectually, ever-cheaper and more convenient communication technologies—e-mail, text messaging, videoconferencing, Skype, instant file sharing, and more—vastly reduce the cost and difficulty of working with other scientists, regardless of their physical location. Back when an excellent researcher could keep up with pretty much everything happening in his or her field, working alone or with an assistant or two made sense. In the days when communicating at a distance meant paying for expensive long-distance calls or waiting for the mail to arrive, the best people to work with were often those who were close at hand, including departmental colleagues. With both of these conditions now drastically changed, the relative “returns to collaboration”—including with people far away—have risen sharply, the authors note. Significantly, the researchers who have “gain[ed] disproportionately from improvements in collaboration technology” turn out to be those with “a larger set of potential collaborators to choose from … .”
Who are those well-connected researchers? The authors suggest that they are top professors with large numbers of former grad students and postdocs located at institutions far and wide, but still closely aligned with their former mentor’s scientific interests. Evidence shows that “a large fraction of collaborations occur between scientists who … previously” worked at the same location, suggesting that at least some of them had teacher-student relationships, the authors add. As if to illustrate this point, Oettl earned his master’s degree at Queen’s University under McHale and his Ph.D. at U of T under Agrawal.
Stars in motion
With scientific stars now “increasingly central” to “collaboration and overall output,” the authors see a “dramatically changing organization of scientific activity” in the field they studied. The new conditions encourage collaborative teams centered on big-name scientists who “specializ[e] in leading major research initiatives, identifying key research questions, and writing grant applications, while their collaborators at lower ranked-institutions run experiments, collect and analyze data, and work together with all collaborators to interpret and write their results.”
A third factor?
Here’s another factor, which the authors did not consider but which also seems likely to increase the prominence of stars and their collaboration networks: the extremely tight job market caused by a large surplus of aspirants for relatively few faculty openings. In a system that values researchers’ prestige (which in practical terms is a surrogate for their ability to achieve the funding and publications that underlie prestige), the students and postdocs of star professors have a big advantage over the protégés of less prestigious and productive mentors in winning the few available jobs.
Almost a decade, ago, we reported on the dynamics of the star system as it operates in the hiring of young faculty members at research institutions. Hundreds of applications routinely arrive for any announced opening. “Amidst mountains of resumes,” we wrote then, “hiring committees of competing research universities vie for the same small group of elite candidates. These few stars each garner multiple interviews and job offers while many of their fellow applicants count themselves lucky to receive a form rejection letter. Sometimes, in fact, departments that fail to land their top choice leave tenure-track positions unfilled rather than hire from a slightly lower echelon.”
The key to all this is universities’ desperate desire to hire faculty members who can quickly begin landing the grants that justify the university’s large investment—in facilities and in new hires’ start-up packages. Such packages now routinely cost a million dollars or more, according to a remark by the dean of a not-quite-top-drawer institution at a recent high-level policy conference. Meanwhile, funding rates are at record lows, sharply increasing universities’ financial risk.
Star applicants can demonstrate—via extensive and prestigious publication records facilitated by relationships with senior stars—that they have a better chance than most of winning funding. So, in addition to the factors mentioned by Agrawal, McHale, and Oettl, today’s brutally competitive funding scene appears to have increased the importance of picking—and being—a sure winner.