The risk of more creative science

This is a press release from UCLA about some of their own researchers’ work, but it strikes me as interesting: “Pressure to ‘publish or perish’ may discourage innovative research, UCLA study suggests”. The research team used a database of 6.4 million scientific papers to do a citation analysis, in which they examined whether papers make new connections by citing work from different networks of knowledge. Those papers had a high payoff in terms of citations, but they were relatively rare.

Drawing on their analysis of scientific rewards, Foster and his colleagues argue that researchers who confine their work to answering established questions are more likely to have the results published, which is a key to career advancement in academia. Conversely, researchers who ask more original questions and seek to forge new links in the web of knowledge are more likely to stumble on the road to publication, which can make them appear unproductive to their colleagues. If published, however, these innovative research projects are more highly rewarded with citations. And scientists who win awards — especially major ones, like a Nobel Prize — have more of these innovative moves in their research portfolio.
“Published papers that make a novel connection are rare but more highly rewarded,” said Foster, the study’s lead author. “So what accounts for scientists’ disposition to pursue tradition over innovation? Our evidence points to a simple explanation: Innovative research is a gamble whose payoff, on average, does not justify the risk. It’s not a reliable way to accumulate scientific reward.”

Most of my work is interdisciplinary, particularly with connections between fossil hominin analyses and genetics. My papers would probably would fall into the category of “risky” in this kind of analysis. There is a real cost to doing such interdisciplinary work. It takes longer to understand the details of different analytical approaches so that I can discuss and cite them accurately.

Nothing drives me as crazy as talking to geneticists who don’t bother to read the details of papers describing fossil hominins and archaeological patterns. Too often, they try to reduce the complexity of paleoanthropological arguments to simple models. The modern human origins problem of the 1990s was unnecessarily contentious because of a lack of reading. I suppose there was always a fear that grants would not flow to those who admitted, “I cannot produce data with current methods that can test this complex model, but I can test this simple one that no one believes.”

I have been rewarded with citations for the integrative papers that I write, and I know that drawing from different networks of research for my own work has led me to new insights. It’s hard for me to imagine doing work that was not integrative in this way.

Still I know a number of other scientists who publish papers only using a single analytical technique, and several of them are very successful at winning grants to support that kind of research. They probably would not describe their research as “low-risk”, even though their publication strategy is highly conditioned on the need to win the next grant.

Probably it’s fair to say that we are doing two different kinds of science.