The paper uses US data to argue that after the pandemic, workers in occupations more capable of being done remotely spent more time alone, withdrew from social activity outside work, and showed worse mental-health indicators than workers in less-remotable occupations. The headline claim that landed hardest was not “remote work is bad” in general, but that the effect was strongest for people living alone, who were more likely to go entire days without human contact and to show more distress and more use of mental-health care.
The strongest reaction was methodological skepticism. A lot of people zeroed in on the design choice the paper itself makes explicit: it compares occupation-level shifts in “remotable” versus “nonremotable” jobs, not a clean remote-versus-office split inside the same jobs. That immediately opens obvious confounders. Tech, finance, and other remote-capable sectors got hit by layoffs,
AI anxiety, and different post-pandemic pressures than fields like nursing or retail. Several readers thought that weakness was big enough to make the causal language feel overstated, even if the underlying observation about isolation sounds plausible. Others pointed out the paper says it tested AI exposure separately and found the effect loaded on remotability rather than an
AI exposure index, but that did not persuade most skeptics that the identification problem was solved.
Once the causality fight was set aside, the conversation got more useful. People largely accepted the narrower point that remote work can become socially dangerous when it removes default daily contact and nothing replaces it. The distinction that kept coming up was not remote versus office, but supported versus unsupported remote work. Living with a partner, kids, roommates, or a real local community changes the experience completely. So does having coworking spaces, clubs, volunteering, meetups, or regular rituals outside work. People with those buffers described remote work as a clear mental-health win because it cuts commuting, lowers office stress, and frees time for chosen relationships. People without them described slow drift into days or weeks with barely any real conversation.
A sharper theme underneath all of this was that many societies have offloaded too much social life onto work. In-office work can hide that problem by supplying hallway chat and
weak ties by default. Remote work exposes it. For some, those weak ties are enough to keep the day from feeling sterile. For others, office interaction is performative, political, or actively draining, and losing it is no loss at all. That is why the practical conclusion was much narrower than the paper’s framing. Remote work is neither inherently healthy nor unhealthy. It works well for people who already have social infrastructure or can build it. It fails badly for people whose only frictionless human contact came from the workplace.