Where disruptive ideas come from
Illustration: Caresse Haaser, Rebecca Zisser/Axios
Science is increasingly specialized, big and so information-saturated that experts can struggle to keep up in their own fields, fueling debate about whether the massive investments are being matched with novel findings and solutions.
What's new: Generating disruptive ideas, according to research published this week, can be encouraged in two seemingly contrasting ways — by forming small teams and leveraging crowds of workers. Each addresses a common enemy of innovation: how we interfere with each other's thinking and creativity.
Big idea 1: Small teams of scientists are more likely than large ones to produce disruptive ideas, according to a new analysis of more than 65 million scientific papers, patents and computer code published between 1954 and 2014.
- A study was deemed more disruptive if those that come along later and cite it don't include the research the original study drew on — a sign that the work was original and took the field in a new direction.
- "In every time period and every field, we find that large teams have an aversion to risk. They basically pick up and build on yesterday’s hits," says study co-author James Evans, a sociologist at the University of Chicago, who argues a venture capital-minded approach to science that supports small teams may be needed.
- Against a backdrop of "big science," it's "a cautionary tale" that we may be losing some of what small teams contribute, says Jeff Nickerson of Stevens Institute of Technology, who wasn't involved in the study.
- Evans and his two collaborators found that small teams, possibly because they are freed of multiple competing ideas and views, are more likely to build on older ideas or, if there are less than about 10 people, those from different areas of science.
Big idea 2: It's difficult for humans to produce analogies that drive some of the disruption Evans observes in small teams, in part because we fixate on the surface-level details of a problem, says Niki Kittur, a computer scientist at Carnegie Mellon University. He wants to create tools to vastly increase the number of diverse ideas people encounter, especially as they become awash in specialized information.
- "Scaling up serendipity:" In a recent paper, Kittur and his colleagues outlined a process for taking the ability of finding analogies out of one individual's mind and distributing it among many people — and machines.
- The steps: Groups of humans strip a problem down to the function or purpose of what they seek to design. An AI could then comb papers, patents, videos, legal briefings, the internet in order to find a common purpose in research in disparate and distant fields, an insurmountable task for any human. Experts are called back in to select and apply those analogies to try to solve the original problem.
- Nickerson, who studies how groups of people approach and solve problems, adds that AI could make us more productive — "a big maybe" — but not if it misses the connections between fields and over time that Evans says is behind the disruption seen in small teams.
The bottom line: "We’ve eaten up low hanging fruits," Evans says. "But there are a host of possible fields, methods and problems that may have been less valuable when we constructed the fields we inherit now. We basically haven’t even begun to climb those trees."