Success is just a bolt from the blue

Citations don't always reflect the quality or impact of a researcher's work.

5 January 2017

COMMENT

Tetra Images / Alamy Stock Photo

I sometimes fantasize about winning the Nobel Prize. Despite knowing the staggering odds against such an honour, I imagine how my latest idea will somehow change the world and I’ll be recognised with the ultimate scientific prize. I don’t think I’m alone. I’m sure many scientists are as idealistic (or blinkered) about the potential impact of their work.

A recent paper in Science by Roberta Sinatra from the Central European University Budapest and colleagues from across the US and Europe sustains my dream. It examined the publication records of more than 230,000 physicists from 1893 to 2010 and found that the highest impact paper for Nobel Laureates could happen at any time in their career. Both the first and the last paper in a physicist’s career are equally likely to be their best.

Optimistic scientists like myself will interpret this as meaning that our best work is yet to come. There’s nothing like a scientific result to sustain our unscientific fantasies. But, of course, the findings could also mean that our best is already behind us.

An important caveat is that the highest impact paper was defined as that with the most citations, but previous work in Nature by John Ioannidis of Stanford University and colleagues has found that a scientist’s most highly cited papers often do not include what they consider to be their best work. This is because highly cited papers can often be summary papers that draw ideas together in one place, rather than original papers that contain an exciting new theory.

Sinatra and colleagues also measured an individual’s scientific quality based on how researchers lift the impact of the papers they co-author. They found high-profile co-authors helped boost their collaborators’ papers. For example, if Steven Hawking was my co-author, I could expect him to provide clever insights that would boost the quality of the paper and hence the number of colleagues who read and cited it. This boosting is multiplicative, so it would lift a poor paper up to average and a good paper up to excellent.

Overall, the authors found that a scientists’ quality remains remarkably constant through their career. This suggests that to become a researcher who has a significant impact on your field isn’t just a matter of training and hard work. You’ve either got it or you don’t. Perhaps this comes from an innate ability or the influence of an inspirational teacher or parent. Scientific experience seems to count for very little.

This rather deflating result suggests scientists, who go through years of training and mentoring, can’t improve their impact during a career. Of course scientific training is still important, and improves other factors such as whether experiments are correctly conducted and reported.

Research into how science functions is often surprising and this latest paper is no exception. Perhaps science is similar to fields such as music and sport where talent is necessary for success and emerges from an early age, think of Maradona in football and Mozart in music. Perhaps I should stop fantasizing about that Nobel Prize.

Dr Adrian Barnett is a statistician who works in meta-research at the Queensland University of Technology, Brisbane. He tweets @aidybarnett