Working near each other

The management researcher Ethan Bernstein and Stephen Turban put badges on employees at two corporate headquarters before and after a move to open plan, recording whom each person faced and whether anyone was speaking, with email and instant messaging logged on the servers. At the first firm, 52 participants went from 5,266 minutes of face-to-face interaction over 15 days to 1,492, which is 72% less, or from 5.8 hours a day each to 1.7. At the second, 100 employees in 1,830 pairs spent between 67 and 71% less time face to face. Instant messages rose 67% and the words in them 75% at the first firm; email between pairs rose 22 to 50% at the second. Bernstein and Turban read it as withdrawal: people who cannot get away from each other physically retreat into a channel they control.

The urban researcher Arianna Salazar Miranda used a decade of MIT publication records against a decade of building renovations. Between 2005 and 2015 those renovations moved researchers between buildings for reasons unrelated to whom they worked with, which turns the campus into an experiment. 2 researchers put in the same building co-authored 0.8 more papers per hundred pairs in the year of the move, 2.7 more by the third year, settling at 1.85 after five. For context, the average MIT researcher publishes about one paper a year, so sharing a building with 100 new colleagues is worth almost one extra paper a year against them. Buildings holding related fields and several departments did more than buildings holding one.

The urban researcher Ilaria Mariotti surveyed 326 people using coworking spaces in Italy, drawn from 549 active spaces in 2018. 73% said they had developed new products or services since joining, and 52% of those had done it with other members. Revenues had risen for 39% of the members and 29% of the firms, and 97% said they were satisfied with the space. The study asks people to attribute their own results, which is the weakest evidence here.

The design researcher Chiara Tagliaro asked 188 managers at an Italian utility company to estimate how their groups split the week between working alone, working with their own team, working with other teams and working on the move. Where a group sat in the company’s collaboration network predicted its teamwork time better than its job function did, and predicted nothing at all about its time alone. Tagliaro’s argument is that space planning uses job function because it is on the org chart, while the thing that actually varies is who talks to whom.

The medical researcher Kristiane Roed Jensen rated 96 video-recorded obstetric emergencies against the sound level in the delivery room, with experts scoring the teams on a validated scale. The 23 teams that worked without any sound above 90 decibels reached high performance 91.3% of the time, the 73 teams exposed to it 58.9%, at p below 0.001. The result survived adjustment for team size, non-technical performance, how fast the patient was bleeding, the type of hospital, the cause, the duration and the time of day. The noise came from instruments hitting the floor, cupboard doors and people crying.

Bernstein and Turban put badges on 152 people. Salazar Miranda used a decade of publication records. Jensen watched 96 recordings of work as it happened. Mariotti and Tagliaro both asked people to estimate their own patterns, 326 in one case and 188 managers in the other.