Scientists and inventors collaborate across greater distances than ever, but those far-flung teams are less likely to produce breakthroughs. A study of 20 million research papers and 4 million patents found that remote teams consistently disrupt their fields less often than teams working in the same city.
The work is by Yiling Lin and Lingfei Wu of the University of Pittsburgh and Carl Benedikt Frey of the University of Oxford. It was published online in Nature on November 29, 2023.
The study tackles a puzzle. There are more researchers than ever, more connected than ever, yet economists have found that new ideas are getting harder to find.
Measuring disruption across half a century
The team compared on-site teams, with every member in the same city, against remote teams spread across two or more cities. Papers from 1960 to 2020 came from the Microsoft Academic Graph, and patents from 1976 to 2020 from the US Patent and Trademark Office’s PatentsView platform.
Each paper and patent got a disruption score. If later work cites a paper but ignores the older work it built on, the paper counts as disruptive; if later work cites both, it counts as developing an existing line. Watson and Crick’s 1953 DNA paper scores among the most disruptive.
In the preprint version of the analysis, the average distance between co-authors grew from under 100 kilometres to nearly 1,000 over the period. Yet the chance that a paper was disruptive fell from 28% for teams in one city to 22% for teams more than 600 kilometres apart. For patents, it fell from 67% to 55%. The gap shrank but held after controlling for field, period, team size, career age and the individual researchers involved, and it appeared both within and across time zones.
On-site teams conceive, remote teams carry out
To see why, the researchers collected author contribution statements from papers in Nature, Science, PNAS and PLOS ONE published between 2003 and 2020. These list who conceived the research, who ran experiments, who analysed data and who wrote the paper.
When the same scientist moved from an on-site to a remote team, the chance they helped conceive the research dropped sharply, from 63% to 51% in the preprint data. Remote collaborators instead clustered in late-stage technical work involving codified knowledge, such as experiments and analysis. Conceptual work, which relies on tacit knowledge that is hard to share at a distance, stayed with people working together in person.
That pattern also shapes careers. In on-site teams, junior and senior researchers were about equally likely to conceive ideas together. In remote teams, the less-cited a researcher was relative to a collaborator, the less likely the pair were to co-conceive the work, so emerging scientists were more often handed technical tasks.
Why better video calls have not closed the gap
The authors conclude that despite striking improvements in digital technology, remote teams are less likely to integrate their members’ knowledge into new, disruptive ideas. Remote papers did gather more citations within five years, which shows that the disruption measure captures something different from popularity.
For research institutions, the finding points to a practical division of labour: distance works for executing a project, while the early stage, when a new idea is still forming, benefits from people sharing a room.
Study Details:
- Title: Remote collaboration fuses fewer breakthrough ideas
- Authors: Yiling Lin, Carl Benedikt Frey, Lingfei Wu
- Journal: Nature
- Publication Date: November 29, 2023
- DOI: 10.1038/s41586-023-06767-1