Albert Einstein. Marie Curie. Charles Darwin. Isaac Newton. Gregor Mendel. Nicolaus Copernicus. George Washington Carver. Rachel Lloyd. Great scientists all, and each one can be tied to what many people feel is a magnum opus. No, a magnum opus is not a musical piece involving personal firearms; there’s already one for field artillery, a magnum opus literally means “great work” – the biggest thing someone could be known for, despite anything else they’ve created. Not every scientist is going to have that kind of breakthrough where their magnum opus is something that changes the game forever. Sometimes, they just want to make sure they’re productive in their research – after all, that’s what they’re getting paid for. It’s all about the papers, and I’m not trying to be ‘street’ about this either – scientists are judged based on their research output, and that’s based on how much they publish. So when will a scientist’s productivity get to the big one, the game-changer?
According to some research on research (we’re not going deeper than that, relax) by Sinatra et al., it turns out that most scientists are going to have a rather limited impact. While discouraging (and just how do you think I feel!?), let’s face it – they can’t all be game-changers, but I’d like to think that most scientific research paves the way for the eventual big one. How did the researchers come up to this conclusion, among many that I’m going to talk about later?
Their conclusions are based on c*_10 (it’s supposed to be an underscored 10, but I don’t know how to do that on WordPress), which is a scientist’s most-cited paper with 10 years of publishing. That’s usually one of the markers of a successful paper anyway, being cited by many other scientists in their work. Looking at the publishing records of more than 2800 scientists as well as >230,000 physicists (sometimes I think that data science is an extreme sport), the authors found that 5% of their data set had c*_10 > 200, which meant that only a few scientists will ever be cited in more than 200 papers within 10 years. Most (75%) will be between 20-200 papers.
As far as when the c*_10 paper gets published, it seems to happen early in or in the middle of a scientist’s career, so it’s between 0 (their first published work) to 20 years. The authors bring up Nobel laureate (physics, 2004) Frank Wilczek as an example – his paper on asymptotically free gauge theories of strong interactions is the most-cited in his career, and it’s his first paper. No, I don’t know what that is.
They do point out that there doesn’t seem to be some patterns in their publishing that emerge before or after c*_10 – it just happens. They propose two models: ‘R’, which means it’s totally random, and ‘Q’, as in there’s some constant, hidden factor that determines the impact of a paper, bringing it closer to being the magnum opus that c*_10 describes. The R-model doesn’t take into account productivity, or how much a scientist publishes, and how much impact all of their other papers have. If these were both true, all scientists are indeed the same. The Q model, which contains a lot of mathematics that (we presume) model some intangible factors – some personal qualities and abilities , although I suspect luck is involved too – that describe how a scientist can take some project and turn out high-impact work.
It’s their next conclusion that’s interesting: a scientist with a low Q may not release a high-impact paper, even if they double their productivity (read: really crank out papers). Scientists with a high Q are more likely to do it. I bring it up since a lot of universities judge research faculty on their research output (among other things, of course) when those faculty are jockeying for tenure. Of course, this doesn’t mean that one shouldn’t even try to be productive if they’re low-Q; after all, you’ll miss 100% of the shots you don’t take. You still never know if you get the big one that defines your career as a good scientist. You won’t get your name on a theory or a Nobel prize roll, but maybe what you do will help the march of scientific progress move along.
The take-home message for this: it takes more than cranking out papers to be a scientist. They’ve got to be good papers, and good papers can only be carried out by good science. Yes, luck will still be involved, as always. c*_10 is not the biggest issue if with your luck, you managed to get killed by a coconut. To all my fellow scientists, let’s keep our science sound, our methods legit, and every once in a while, look up.
Featured article: Sinatra R, Wang D, Deville P, Song C, Barabási A-L. (2016) “Quantifying the evolution of individual scientific impact.” Science 354, aaf5239. doi: 10.1126/science.aaf5239.
Featured image credit: Wikimedia Commons/Public Domain (CC0, Author: Benjamin Couprie)