Have Desk, Will Commute

I’m in the job market right now, and I keep telling recruiters something that seems to catch them off guard: I prefer an in-office role where I actually go in, have a real desk, interact with colleagues in person, and go to a building most days of the week. In a market where remote work has become the default perk, saying you want to go to the office can sound like a glitch in the system.

It isn’t. The case that pushed me to write this came from Scott Galloway, the NYU professor, on an episode of Real Time. He called remote work the second worst thing to happen to young people,1 and named what they give up by staying home. One in three relationships begin at work, he argued. It’s where you find not just a paycheck but friends, mentors, and mates, and where the young absorb the social guardrails that hold early adulthood together.

He was talking mostly about people half my age, and I’m well past needing an office to meet a mate, but the things on his list don’t stop mattering with age. The conversations that never happen, the mentor who never peeks over the wall, the friend never made: those are losses at any age. From where I sit now, as the person who’d be doing the mentoring, I can watch them disappear for exactly the people I’d most want to bring along. It crystallized something I’d already come to believe from years on engineering teams.

The trouble with the work-from-home debate is that it has calcified into a single, boring question. Are people more productive at home or at the office? Measured that way, the people who want to stay home have a real case. So let me grant them their strongest version of it, and then explain why I’m still looking for a desk.

The bullpen

I’m not arguing this from theory. I lived the version of work I’m describing, and I’d take it back tomorrow.

Some of my fondest professional memories come from a bullpen with thirty engineers in one room. Someone always had Turntable.fm going, and we took turns playing DJ, queuing tracks for the room while we worked. You could lean over and bounce a half-formed idea off whoever sat next to you and have it sharpened, or demolished, in about thirty seconds. Stand-ups happened in a corner of the room, on our feet, done in five minutes. Some of it was on the calendar and some of it wasn’t, but the part that made the place hum, that easy back-and-forth, you couldn’t have scheduled if you tried. All of it added up to a culture.

This wasn’t unique to that one bullpen, either. Even at my last company, before COVID, I knew the people in the cubes around me. I knew their families, the new baby, the move someone was planning for the summer. You pick all of that up without trying, just from sitting a few feet from a person month after month.

And you felt the place. When someone left the company, you knew. They gave handshakes, they gave hugs, and you actually saw them walk down the hall and out the door. To lighten the mood, we’d cue up “The Lonely Man,” the wistful piano theme that closed every episode of the old Incredible Hulk series, and play them out as they went. It was a joke, and it wasn’t. You don’t get that in a remote setting. Typically a teammate leaves and you get an email, one you may have missed because the goodbye got buried in your inbox. Their status dot goes gray, and a week later you notice you haven’t seen their name in a while. That bullpen scattered more than ten years ago, and I’m still close with some of the people who sat in it. A status dot doesn’t do that.

The relationships we stopped having

In 2023, the U.S. Surgeon General did something unusual: he issued a formal public-health advisory about loneliness,2 placing social disconnection in the same category of national concern as smoking and obesity. The comparison wasn’t rhetorical. A landmark meta-analysis by Julianne Holt-Lunstad,3 pooling 148 studies and more than 300,000 people, found that weak social connection carries a mortality risk comparable to smoking and greater than obesity. It is the research behind the now-common warning, repeated by the Surgeon General himself, that chronic loneliness can be roughly as deadly as smoking fifteen cigarettes a day.

Now ask where adults make their close friends. Plenty of us keep friends from school and the old neighborhood, and some of those bonds last a lifetime. That isn’t luck. Sociologists since the 1950s have pointed to three conditions that close friendship seems to require: proximity, repeated and unplanned interaction, and a setting that lets people drop their guard. As the sociologist Rebecca Adams has noted,4 that mix is exactly why so many people meet their lifelong friends in college. A school hallway or a dense childhood block is saturated with all three.

Adulthood strips them away. Once school is behind you, the daily unplanned collisions mostly stop, proximity turns into something you have to schedule weeks out, and the settings that invite you to be vulnerable get rare. One place still reliably supplies all three for most adults: work. So it isn’t surprising that more than half of us form a close friendship at work,5 and that we’re likelier to make a new friend there than almost anywhere else in adult life. School and the neighborhood gave us the friends of our youth. Work is where most of us go on making them as adults.

But the office does something subtler than that, too. Every coworker you grow close to comes attached to a whole world you’d never have reached on your own. You meet their college roommate at a birthday party. You get invited to the wedding and leave with two new numbers in your phone. Their sister needs a contractor, and it happens you know a good one. A single work friendship is rarely just one friendship. It’s a doorway into someone else’s entire network, and adult life hands you very few of those doorways. The office was never only a place to produce output. For a lot of people, it was the last reliable place to meet someone, and the last reliable place to meet everyone that someone knew.

A hit sitcom already made this argument for us. The Office ran nine seasons on one premise: put people in a Scranton branch office and let the human friction do the rest. Jim and Pam, Dwight and Angela, Michael and Holly: nearly every romance in it began across a desk, Galloway’s one-in-three statistic acted out for laughs. Jim doesn’t suspend Dwight’s stapler in Jell-O over a video call, and the found family never forms in a Slack channel. A remote version of The Office isn’t a different show, it’s no show at all. There’s nothing left to film.

Pull everyone out of that building and the laughs aren’t the only thing you lose. BetterUp found that the share of workers who say they actually know their colleagues as people fell from 79 percent in 2019 to 68 percent in 2024.5 Gallup’s global data6 adds a stranger wrinkle. Fully remote workers report the highest engagement with their jobs, and also the highest rates of loneliness, stress, and anger. They like the work. They are coming apart around it.

To be fair, remote work didn’t invent the loneliness epidemic. The amount of time Americans spend with friends has been falling for two decades, long before Zoom, but remote work removes the incidental contact that used to catch people before they fell through: the hallway hello, the unplanned lunch, the break room encounter where someone asks how your weekend was, the colleague who notices you’ve gone quiet. It doesn’t cause the disease. It removes one of the few places that was treating it.

What’s filling the silence

This is the part that unsettles me, because we lost the workplace relationship and something has already moved into the space it left.

By 2026, more than a billion people are using generative AI,7 and a growing share of them aren’t asking it to write emails. They’re asking it for company. Replika, one of the largest AI-companion apps, has tens of millions of users, and by the company’s own reporting a majority of its paying subscribers describe themselves as being in a romantic relationship8 with their AI. Character.AI’s users reportedly spend around two hours a day talking with their virtual friends, more than the average person spends on TikTok.8 One report put the share of teenagers who’ve formed some kind of relationship with an AI at nearly three-quarters.8

The appeal isn’t a mystery. As a counseling psychologist with the American Psychological Association9 recently put it, AI companions are endlessly validating and never argue back. They offer the sensation of being heard with none of the friction. Real relationships are messy and unpredictable. An AI is patient at 2AM, agrees with you, and never needs anything in return.

And that’s the trap. An OpenAI and MIT Media Lab study9 found that light use of a chatbot could ease loneliness, but heavy daily reliance correlated with more loneliness, not less, because it displaces the real thing. A year-long study tracking thousands of users7 found the same pattern. Leaning on chatbots in lonely moments tended to deepen the loneliness over time. The AI hands you the feeling of connection while removing the reason to go find the genuine version.

Now put that next to remote work. The coworker bond, reciprocal, occasionally annoying, fully human, is precisely the kind of relationship most at risk of being swapped for the frictionless substitute. When the colleague who’d have helped you debug is a Slack handle you’ve never met, and the AI in your editor answers instantly and never judges you for asking, which one are you going to lean on? Working from home doesn’t just remove the human. It installs a more convenient stand-in right where the human used to sit. I build with these tools every day and I think they’re remarkable, but a remarkable tool is a terrible reason to stop talking to people.

The contributions nobody logged

The productivity charts miss something, and it’s the part I feel most personally.

I’ve spent a good chunk of my career as the senior person in the room, the one whose job was to listen to what someone was working on and flag the pitfalls to watch for before they became a major outage at 2AM. I’ve tried to do that same job over a video call and a code-review thread, and it is not the same thing. Something gets lost in translation: the half-question the junior dev would’ve asked in person but swallows in Slack, the context I’d have offered if I’d seen the look on their face.

A video call is a closed loop. One person speaks at a time, the call ends, and whatever you meant to raise but didn’t is gone with it. In a shared room, the person stays in your line of sight all day, and that presence does the remembering for you. You glance up, catch them deep in the thing you were worried about, and it nudges you to go flag it. Remotely, that nudge never fires, because the person you’d have reminded yourself to talk to isn’t anywhere you can see.

I wasn’t imagining it either. A team of economists, Natalia Emanuel, Emma Harrington, and Amanda Pallais, studied software engineers10 at a Fortune 500 firm whose campus happened to be split across two buildings a few blocks apart. When the offices were open, engineers who sat in the same building as their teammates received about 20 percent more feedback on their code than those whose colleagues were elsewhere. When the offices closed, that advantage largely evaporated.

And the people who lost the most were the ones who could least afford to: new hires and junior engineers, who tended to stop asking for help once they couldn’t do it face-to-face. The feedback they were missing wasn’t noise. It was the slow, invisible transfer of judgment from people who’d already made the mistakes. Mentorship doesn’t show up on a sprint board. It shows up three years later, in who turned out to be good at the job.

And the professional loss is only half of it. The junior engineer who never gets that feedback is also the young person Galloway was worried about, the one who isn’t in the room where friendships form, where a third of people meet a partner, where you learn by watching older colleagues what the work and the life around it are supposed to look like. I’m not the one missing a first real friendship or a partner. I’m the one who used to be the senior guy those things happened around. Remote work doesn’t only cost that engineer a faster path to competence. It costs them the place where a good deal of a life gets built, and it costs the rest of us the chance to help build it.

The economics of the empty building

Walk through a downtown core on a Tuesday and you can see the other bill coming due.

U.S. office vacancy has been stuck at historic highs, hovering in the high teens nationally11 in 2026, with the hardest-hit downtowns like Seattle running past 25 percent, after the national rate approached a record 20 percent12 at its peak. Some markets are recovering, but the scale of empty space is unlike anything in living memory. Economists have a name for what it threatens: the urban doom loop. Columbia’s Stijn Van Nieuwerburgh, the economist who popularized the term, describes the mechanism plainly.12 Falling office values mean lower property-tax revenue, which forces cities to cut services, which makes them less livable, which pushes more people out, which lowers values further. A spiral that feeds itself.

And the people caught in it mostly never got a vote. The remote-work decision belonged to office workers and their employers, but a full building keeps far more people employed than the ones with laptops. There are the cafeteria workers who served lunch, the network specialists who kept the in-office systems running, the landscapers who maintained the grounds, the maintenance crews who serviced the plumbing, the custodians who cleaned up after everyone went home. Empty the building and every one of those jobs empties with it. From there the loss radiates outward: the deli owner whose lunch rush evaporated, the bartender who used to know the regulars by their drink orders. A Boston bar owner told ABC News13 that office workers had been half his customers, and then they simply stopped coming. He didn’t get to work from home. He just got the empty room.

This is the part of the ledger that the convenience of working from home never accounts for, because it’s external by design. The benefit is private and immediate: your morning back, your dog at your feet. The cost is public, diffuse, and paid by someone you’ll never meet.

The productivity question, taken seriously

The work-from-home side has a real argument here, and it deserves a fair hearing.

A senior engineer alone at home, undisturbed, with no one tapping their shoulder, will often ship more and cleaner code than the same engineer in an open-plan office full of interruptions. Stanford’s Nicholas Bloom has tracked this for over a decade. In a randomized controlled trial at Trip.com, hybrid workers matched in-office output while resignations fell by a third.14 His earlier call-center experiment found home workers were 13 percent more productive15 on their own tasks. I’ve felt it myself. The deepest work I’ve ever done, I did sitting in an empty house.

But it raises the question the productivity charts never ask. More productive for whom? A senior developer siloed at home is optimizing one number, their own output, while the most valuable thing in the building stays locked in their head. Remember that proximity study.10 Sitting near coworkers lowered senior engineers’ short-term output, because they spent that time teaching. A chunk of the celebrated work-from-home productivity gain is just the senior person no longer mentoring anyone. That isn’t a gain. It’s a cost we deferred and stopped counting.

Where I come down is this, and it’s why blanket return-to-office mandates fail. The answer isn’t to herd everyone into a room to do the exact same solo work they’d do at home, only with worse coffee and more interruptions. That really is the worst of both worlds. The answer is to design the office around the two modes that matter. Protect real deep-focus blocks, hours where engineers work heads-down and undisturbed, the way they would at home, because that solo focus is real and worth defending. Then deliberately build in the handoff: the review, the walkthrough, the session where the senior’s finished work becomes the junior’s education. Solo focus and knowledge transfer aren’t enemies. They’re two different modes, and a well-run office can schedule both on purpose. A home office can only ever give you one of them.

That’s the part of the job I want to do. Not just write good code, but make the people around me better at writing it. You can’t do that from a silo, no matter how productive the silo feels.

What I’m looking for

None of this is an argument for marching everyone back to a cubicle five days a week and pretending 2019 never ended. Flexibility is real, and the time people got back is not nothing. Anyone who’s traded a 90-minute commute for breakfast with their kids knows that.

It’s an argument for counting the whole bill. Remote work is a trade, not a free upgrade, and we’ve spent four years celebrating one side of the receipt. We got our mornings back. We lost a little of the connective tissue that holds careers, friendships, and downtowns together, and we started handing what was left to machines that are happy to fake it. Unlike productivity, you don’t feel that loss in a quarter. You feel it in a decade, when the junior engineer never got mentored, the new hire never made a friend, and the city outside the empty tower got a little harder to love.

So that’s what I’m looking for. Not a mandate to grudgingly comply with, but a room I want to be in, a team that still believes the office is for more than headcount and is willing to run it like it means it. The office was never really about the work. That was always the easy part. The things it’s really for are the things I want to show up for.

Sources

  1. Galloway, S. Remarks on remote work and young people (video, 2025). https://www.youtube.com/watch?v=I0COHGgySro ↩

  2. Office of the U.S. Surgeon General. Our Epidemic of Loneliness and Isolation: The U.S. Surgeon General’s Advisory on the Healing Effects of Social Connection and Community (2023). https://www.ncbi.nlm.nih.gov/books/NBK595227/ ↩

  3. Holt-Lunstad, J., Smith, T. B., & Layton, J. B. “Social Relationships and Mortality Risk: A Meta-analytic Review.” PLOS Medicine 7(7): e1000316 (2010). https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1000316 ↩

  4. Williams, A., on the sociology of adult friendship (citing Rebecca G. Adams’s three conditions for close friendship). NPR / The New York Times (2012). https://www.npr.org/2012/07/17/156920115/recipe-for-a-good-friendship-form-by-age-30 ↩

  5. “How modern work is fuelling an epidemic of loneliness.” HR Future (2024), citing BetterUp workplace-connection data. https://www.hrfuture.net/talent-management/culture/how-modern-work-is-fuelling-an-epidemic-of-loneliness/ ↩ ↩2

  6. “Remote Work Productivity Study.” Great Place To Work (2025), summarizing Gallup’s State of the Global Workplace engagement and well-being findings. https://www.greatplacetowork.com/resources/blog/remote-work-productivity-study-finds-surprising-reality-2-year-study ↩

  7. “Turning to chatbots when lonely may exacerbate feelings of loneliness, study finds.” PsyPost (2026). https://www.psypost.org/turning-to-chatbots-when-lonely-may-exacerbate-feelings-of-loneliness-study-finds/ ↩ ↩2

  8. “AI Relationships: Why People Are Falling for Chatbots.” Built In (2026). https://builtin.com/articles/ai-companionship-changing-human-relationships ↩ ↩2 ↩3

  9. Andoh, E. “AI chatbots and digital companions are reshaping emotional connection.” APA Monitor on Psychology (2026), including the OpenAI / MIT Media Lab usage findings. https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection ↩ ↩2

  10. Emanuel, N., Harrington, E., & Pallais, A. “The Power of Proximity to Coworkers: Training for Tomorrow or Productivity Today?” NBER Working Paper 31880 (2023). https://www.nber.org/papers/w31880 ↩ ↩2

  11. “U.S. Office Market Report, May 2026.” CommercialCafe. https://www.commercialcafe.com/blog/national-office-report/ ↩

  12. “US office vacancy levels hit record high, are we on the precipice of an ‘urban doom loop’?” AOL / Moneywise, citing Moody’s Analytics and Columbia’s Stijn Van Nieuwerburgh. https://www.aol.com/finance/vicious-cycle-us-office-vacancy-102800947.html ↩ ↩2

  13. “How US cities are reimagining the future as office vacancy rates soar.” ABC News (2024). https://abcnews.go.com/Business/us-cities-reimagining-future-office-vacancy-rates-soar/story?id=115968925 ↩

  14. Bloom, N., Han, R., & Liang, J. “Hybrid working from home improves retention without damaging performance.” Nature 630, 920-925 (2024). https://www.nature.com/articles/s41586-024-07500-2 ↩

  15. Bloom, N., Liang, J., Roberts, J., & Ying, Z. J. “Does Working from Home Work? Evidence from a Chinese Experiment.” Quarterly Journal of Economics 130(1): 165-218 (2015). https://doi.org/10.1093/qje/qju032 ↩

Back to top

Built with Jekyll and Just the Docs | © 2025 Rich Zapata