Disruptive ideas: Small teams score over big ones

Mom en ts of great scientific disruption aren’t necessarily glamorous. A world-changing idea might spring up when someone is sitting in an office or scrawling notes, or maybe riffing with a colleague. Big projects leave more vivid impressions— landing on the moon, testing atom bombs, or, more recently, finding the Higgs boson and detecting Einstein’s predicted gravitational waves using a pair of sprawling but stunningly precise detectors. The paper announcing the Higgs boson boasted 5,000 authors, the one on gravitational wave detection, 1,011 authors. So why bother with small projects at all?

A new analysis suggests the papers that spawn whole new branches of science most often springfrom lone researchers or small groups, and the more innovative and disruptive the findings, the smaller the teams. Big teams are still needed for some kinds of tasks, but the most successful of them either test an existing hypothesis or conduct some feat of applied science, such as turning the principle of nuclear fission into abomb. The small team of three people who published the team size analysis in last week’s Nature took advantage of amassive database containing 65 million documented advances — scientific papers, patents and software products.

How do you measure disruption? Dashun Wang of Northwestern University said they looked for a telltale pattern in the way a paper was cited in subsequent work. He and his colleagues reasoned that the more innovative papers would be cited without the older studies that they themselves cited.

That would suggest these papers weren’ t continuing along known branches, but had started new ones. Less disruptive papers, by contrast, would be cited along with a string of previous work. As an example, they point to a paper introducing a new phenomenon called self-organised criticality, which was cited without the trail of previous citations — and indeed it launched a new branch of research. The disruptive papers also get cited longer into the future — even if they have lower impact at first.

“We realised as team size increases, the character of the work they produce systematically changed ,” said Wang. When they tracked individual scientists, they found the same person had more disruptive impacts when part of smaller teams than when part of larger ones. This comes at a time when small science needs some respect. As measured by the initial burst of early citations, the highest-impact papers have switched from those with individual authors to teams, said Laszlo Barabasi, a network scientist at Northeastern University.

The moon shot and the Manhattan Project are often held up as banners for the power of big science, but this may be misguided. Historian Alex Wellerstein, quoted in Scientific American, put it this way: “Almost every week someone, somewhere calls for a new Manhattan Project. Apparently we need aManhattan Project for cancer, for AIDS, for health, for solar power… for cybersecurity, for nutritional supplements and most literally for protecting the island of Manhattan from sea level rise.”

The problem, as Wellerstein saw it, was that these new calls for a new Manhattan Project were aimed at answering scientific questions while the original Manhattan Project was aimed at applying a new scientific idea, nuclear fission, to bombs.

The science behind it was found by pairs and individuals, not making big explosions, but doing the mundane-looking task of thinking.