The Open Office Gambit
Every few years, a company tears down its cubicle walls in the name of collaboration. The logic seems airtight: remove physical barriers and people will talk more. By 2020, roughly 70% of American offices featured some form of open plan. The trend rests on a bedrock assumption that proximity breeds interaction, rooted in decades of sociological research showing that people who are physically closer form stronger social ties.
Ethan Bernstein and Stephen Turban, researchers at Harvard Business School, decided to test that assumption with something no prior study had used: objective measurement. Not surveys, not interviews, not activity logs completed each evening from memory. Wearable sociometric badges that tracked every face-to-face exchange in 10-millisecond intervals, combined with email and instant messaging data pulled directly from company servers. Their results, published in Philosophical Transactions of the Royal Society B, were unambiguous.
What Two Companies Actually Measured
The first study tracked 52 employees at a Fortune 500 headquarters for 15 workdays before their floor was redesigned, then another 15 workdays after a three-month adjustment period. The sociometric badges captured infrared proximity, speaking patterns, and body orientation simultaneously. If two people faced each other, alternated speaking, and stood within 10 meters, the system logged a face-to-face interaction. When any condition broke for more than 5 seconds, it ended.
Before the walls came down, each employee averaged 5.8 hours of face-to-face interaction per working day. After the redesign, that number fell to 1.7 hours, a 72% reduction that stunned the executives who had commissioned the project. Email volume told the mirror-image story: messages sent to colleagues increased 56%, messages received jumped 20%, and CC'd messages climbed 41%. Instant messaging surged 67% in message count and 75% in total word count. Workers did not stop communicating; they retreated to screens, replacing corridor conversations with inbox threads.
The second study, at a different Fortune 500 company, tracked 100 employees for eight weeks before and eight weeks after their open-plan conversion. Face-to-face time dropped between 67% and 71%, depending on the statistical model. Email volume rose 22% to 50%. Of 1,830 tracked employee pairs, 643 reduced their face-to-face interaction while only 141 increased it. That is a 4.6-to-1 ratio favoring withdrawal.
Why People Withdraw in Open Space
The mechanism behind this withdrawal is straightforward, rooted in a tension between architectural openness and the human need for privacy. In a cubicle, a conversation is semi-private. Two people can talk without performing for an audience of peers. In an open plan, every exchange is public. The person three desks away can hear your voice, read your body language, and draw conclusions about your priorities and work habits. Under these conditions, employees erect what Bernstein calls a "fourth wall"โa social barrier built to replace the physical one that was removed.
The strategies are visible to anyone who has worked in such a space. Noise-canceling headphones worn as do-not-disturb signs. Eyes locked on screens to discourage interruption. And a quiet migration to Slack channels and email threads where conversations leave no ambient trace. The office becomes a library with keyboards.
Physical distance between workstations still mattered, but far less than expected. In Study 2, walking distance between desks had a statistically significant but small effect on face-to-face interaction, and no significant effect on email. The open-plan transition effect dwarfed it completely, suggesting that visibility matters more than proximity. Being ten meters closer made almost no difference when the entire floor was exposed.
Scaling the Damage
Here is a calculation the original paper did not run. Study 1 employees each lost roughly 4.1 hours of daily face-to-face time (5.8 hours before minus 1.7 after). For a company of 5,000 employees undergoing the same transition, that amounts to 20,500 lost person-hours of face-to-face interaction per working day. Over 250 working days, the annual total reaches 5.125 million person-hours. That is equivalent to erasing every in-person conversation across 2,562 full-time employees for an entire year. The open office does not eliminate collaboration infrastructure. It hollows it out and replaces speech with typing.
The Strongest Counterargument
The most substantive challenge comes from workplace design researchers who argue Bernstein and Turban tested the wrong kind of open office. A bare-bones open floor with rows of desks and no acoustic zoning represents the cheapest and most extreme implementation of the concept. Ben Waber, whose company Humanyze commercialized the sociometric badge technology used in these studies, has argued that well-designed activity-based workspaces with varied zones can preserve or even increase face-to-face interaction. A 2021 study by Appel-Meulenbroek and colleagues in Building and Environment found that offices combining open and enclosed spaces produced higher satisfaction and interaction quality than either pure cubicle or pure open layouts.
This counterargument carries real weight, and the research supporting it is growing. Both of Bernstein and Turban's field sites converted from cubicles to fully open floors, the architectural equivalent of removing a car's doors and declaring it more aerodynamic. More nuanced designs may avoid the worst effects. But the counterargument also concedes the central finding: the default open office, the one most companies actually build because it is cheapest, actively suppresses the collaboration it was meant to create.
What We Didn't Prove
Both field sites were single floors at Fortune 500 headquarters in the United States, staffed primarily by knowledge workers in corporate functions: technology, sales, HR, finance, and product development. Industries with different collaboration norms (hospitals, trading floors, newsrooms) may respond differently. The studies measured volume of interaction but not its quality; it is possible that the remaining 1.7 hours of daily face-to-face time were more focused and productive than the 5.8 hours that preceded them, though nothing in the data suggests this. Both studies captured a transition period, meaning some portion of the decline could reflect temporary adjustment rather than permanent behavioral change. OpenCo1 executives, however, reported that internal productivity metrics remained depressed well after the transition.
The Bottom Line
Open offices do the opposite of what their architects promise. When you strip away physical boundaries, people erect social ones. The result is less conversation, not more, mediated through screens rather than speech. Two independent field studies, both using objective sensor data rather than self-reports, converged on the same 70% decline. The finding is large, consistent across both sites, and deeply inconvenient for anyone who just signed a ten-year lease on a boundaryless floor.
What You Can Do
If you manage a workspace, resist the all-or-nothing redesign. Create zones: open areas for spontaneous interaction, enclosed rooms for focused work, and small pods for two-to-four-person conversations that need privacy. If your company already has a fully open plan, advocate for retrofitting acoustic partitions and designated quiet hours. As an individual, track how much of your communication has migrated to email or messaging since your last workspace change. If the answer is "most of it," the architecture is working against you. Seek out face-to-face conversation deliberately by scheduling walking meetings and claiming a conference room for the 15-minute discussion that would otherwise become a 12-email thread. The sensor data confirms your instinct to withdraw in open space is natural. Fighting that instinct is worth the effort.