Smaller teams of scientists tend to carry out work that is more “disruptive” than that of larger groups, according to a study of more than 65 million papers, patents and software products published between 1954 and 2014. The study, led by James Evans from the University of Chicago, finds that as teams grow from one to 50 members there is a big drop in scientific “disruptiveness”. Large groups instead mostly carry out work that develops on existing research.
To assess whether work can be classed as disruptive, the researchers looked at the citations of their 65 million research outputs. In particular, they examined whether the work cites the references of another paper’s references. Work that cites many references in another paper it cites is considered to be building on and consolidating previous research, while work that refers to few of its reference’s citations is classed as disruptive.
The researchers found that solo authors are 72% more likely to be highly disruptive – in other words in the top 5% of disruptive papers – than teams of five. Teams of 10, meanwhile, are 50% more likely to have a higher impact papers than smaller teams, but these papers were unlikely to be highly disruptive. When validating their work, the researchers found that Nobel-prize-winning papers are among the 2% most disruptive papers, while review articles are among the least disruptive. They also asked academics to propose articles that they thought were disruptive and developed on existing ideas, which again aligned with their analysis.
Evans told Physics World that he thinks smaller teams have more to gain and less to lose from disruption as they cannot compete with larger teams when it comes to building on previous work. This pushes them to explore possibilities that “don’t gel with but could disrupt” current scientific and technological trends. Evans says that clashes in large teams also reduce the number of ideas and steer members towards things they have in common.
Evans adds that the findings and the growth of large teams should concern anyone interested in the long-term viability of science. “Scientific funding agencies and technology financing needs to undertake more risks associated with the higher likelihood of failure and disruptive success that characterise small team research,” he says.
The disruption measure used by the researchers was based on work carried out by Jason Owen-Smith, a sociologist at the University of Michigan, who told Physics World that the findings have practical implications for how society funds and supports science. “It suggests that in an era of ‘big science’ it is also important to identify and support work by smaller teams across diverse settings,” he says, adding that it also signals the importance of research that “may not be directed at any particular near-term goal for the overall progress of science”.
That view is backed up by Pierre Azoulay from the Massachusetts Institute of Technology, who told Physics World that the time is right to end the “collaboration fetish”. “Enough with funding mechanisms that predicate funding on collaboration of this kind or that kind,” says Azoula, who was not involved in the study. “There is nothing wrong with collaboration, but there is no reason for policy makers and funders to think that they can engineer collaboration in ways that are necessarily going to be wonderful.”