To GingerAnne Collins; a friend ever beloved and treasured.

A question that seems simple rarely is. Where should people work? The answer has calcified into camps. On one side: people who treat any office requirement as evidence of managerial pathology, pointing to productivity claims and the absurdity of commuting to a desk you could put anywhere. On the other: executives invoking “culture” and “collaboration” as though those were arguments and not words that sound like arguments. Both sides wield data selectively. Both mistake their preferred arrangement for universal truth. Years inside this question – remote, hybrid, back to the office, back home – have taught me to distrust the confident voices on either side. The truthful picture is messier. It depends on the kind of work. On how long someone has been doing it. On whether the task needs sustained solitary concentration or iterative group thinking. On the phase of a company’s life. On trade-offs that can’t be optimized away, only chosen among. What follows is an attempt to think through those trade-offs without flag-waving – not to crown a winner but to understand what each arrangement costs and provides. The research base is stronger in some places than commonly assumed and weaker in others. Where I know, I’ll say so. Where the evidence runs out, I’ll say that too. Where the numbers being thrown around don’t hold up to scrutiny, I’ll say that too.

I.

Start with productivity, since that’s where most arguments begin. The claim that remote workers are more productive has near-axiomatic status in certain circles – the kind of “fact” people repeat at dinner parties as though it were settled. The evidence is hazier than the confidence suggests. The strongest case comes from a call center in Shanghai. Workers were randomly assigned – actual randomization, not self-selection – to work from home or stay at their desks. The remote group produced 13% more over the nine-month trial. Part of the gain came from working more minutes: fewer breaks, less sick leave, no commute eating the day’s edges. Part came from a quieter environment, with more calls handled per hour when colleagues weren’t chatting nearby. The finding gets cited constantly. It rarely gets contextualized. Call center work is repetitive, individual, and easy to measure. You can count calls. You can time them. The work requires concentration but not collaboration, execution but not invention. Whether the result generalizes to knowledge work – where output resists easy measurement and often emerges from interaction – remains an open question. A more recent experiment offers different evidence. At a large travel company, engineers and other skilled workers were randomized into two groups: five days in the office or three days in the office with two at home. The treatment ran six months; researchers then tracked performance and career outcomes over the following two years. The hybrid group showed no measurable difference in performance reviews, promotion rates, or lines of code written. Quit rates, however, dropped by a third. The hybrid arrangement was in effect free – same output, less turnover, happier employees.

Notice the structure: hybrid, not fully remote. Three days per week of in-person contact. The flexibility benefit without the costs of total distribution. That distinction gets lost when people cite the finding. Now measure productivity differently. One study tracked over 10,000 IT professionals at an Indian firm during the shift to remote work – not through surveys but through monitoring software logging hours and output directly. Productivity fell somewhere between 8 and 19 percent. People worked longer hours but produced slightly less. Time in meetings and coordination rose. Uninterrupted work time shrank considerably. One caveat the number needs: this measured the abrupt forced shift to remote during COVID, when neither workers nor management chose the arrangement and neither had infrastructure built for it. It’s a good study of sudden disruption, not of considered remote work at companies that design for it. An uncomfortable pattern emerges even so. Studies using self-reported productivity consistently favor remote work. Studies using objective measurement show flat or negative effects. The implication isn’t that remote workers are lying. It’s that people are poor judges of their own output, or that what feels productive and what is productive may drift apart. You can end a day feeling effective – messages answered, meetings attended, a few tasks closed – without having produced much. The pleasant sensation of busyness is a reliable illusion. Maybe the divergence doesn’t matter much if workers are happier and more likely to stay. Maybe it matters enormously if it compounds across years. The truthful answer is that we don’t yet know.

II.

Where the evidence converges is on retention and preference. Remote and hybrid arrangements keep people. The travel company experiment found quit rates dropping by a third. The earlier Shanghai study found attrition cut in half. The findings replicate across organizations and industries: companies offering work-from-home options see turnover drop by somewhere between a quarter and half. Employee preferences are unambiguous to the point of comedy. Surveys find that 98% of remote workers want to continue working remotely at least some of the time. When surveys ask how much pay people would give up to work from home, the responses are striking – in some recent surveys, close to 70% of respondents claim they’d accept lower pay for the privilege. That precise number deserves care. It comes mostly from surveys run by remote-job listing companies, which have a stake in the answer, and stated-preference data has a known problem: what people say they’d trade in a survey tends to be larger than what they’d actually trade when facing real consequences. When economists run experiments that force real choices – careful willingness-to-pay estimates from studies including the travel company experiment mentioned earlier – the revealed valuation of remote work lands closer to 8% of salary. Meaningful. Worth taking seriously. Nothing like 70%. The gap between stated and revealed preference is itself the more interesting finding. It tells you something about how people feel about remote work versus what they’d actually sacrifice for it. Strong feelings, modest willingness to pay. Both numbers are true, in their way – one measures emotional temperature, the other measures economic value. Any discussion citing only the dramatic 70% figure captures the feeling and misses the tradeoff.

The underlying preference isn’t mysterious, whatever its precise magnitude. Commuting is miserable. Across diary studies that ask people to rate how they feel during different activities, the commute lands reliably among the least pleasant parts of the day – worse than housework, worse than most work itself. Reclaiming that time, whether an hour or ninety minutes or, in some cities, two hours daily, improves quality of life. Add the flexibility to handle errands, the chance to be present for family, the option to work in comfort rather than open-plan fluorescent noise, and the appeal becomes obvious. Cost savings exist on both sides, though the specific dollar figures that circulate are softer than they appear. Employers save on real estate, utilities, and related overhead; employees save on commuting, parking, lunches, and work clothes. The widely repeated per-employee savings estimates – the ones cited in just about every remote-work article – come from consulting firms’ proprietary models rather than from measured outcomes at actual companies. Directionally correct, numerically speculative. The qualitative truth is that savings exist; the quantitative truth would require better data than currently exists in public. The psychological costs get tallied less often. About three-quarters of remote workers report feelings of isolation. The boundary between work and home blurs until there is no boundary. The commute, for all its misery, once served as a transition – a liminal space between professional and personal life, a chance to decompress with bad radio or a podcast before walking through the door. Strip it away and work seeps into evenings and weekends. The workday has expanded since widespread remote adoption; after-hours and weekend work have risen noticeably. Whether that represents efficient flexibility or the slow colonization of private life by paid labor is a question of interpretation. Probably it’s both, depending on the person and the organization.

III.

Remote work’s appeal rests partly on the promise of deep work – uninterrupted time for the cognitively demanding tasks that produce real value. The theory is intuitive. Without colleagues stopping by, without ambient office chatter, without the impromptu meeting scheduled on top of your morning, you can finally think. Some tracking data supports parts of the theory. In studies where remote workers’ focus time has been measured through software, they log meaningfully more uninterrupted work hours than their office counterparts. The classic finding that workers need more than twenty minutes to recover from an interruption gives the extra focus time value. Interrupted four times daily, you might lose an hour to recovery alone. The picture is more complicated than advocates acknowledge. Those same remote workers, freed from office interruptions, tend to fill the space with meetings. Video calls metastasize across calendars. Coordination that once happened in passing now requires scheduled synchronous time. The monitoring data that showed productivity dropping at the Indian IT firm also showed uninterrupted work time shrinking, not expanding.

Domestic distractions replace office interruptions rather than erasing them. A third of remote workers report struggling with staying home too much. Children interrupt. Spouses interrupt. Pets interrupt. The refrigerator interrupts in its own tempting way. Delivery trucks ring the doorbell during calls. The barking dog three houses down interrupts. The environment changes; the interruptions shift in character without disappearing. Video fatigue was a constant topic during the pandemic’s first year – the strange exhaustion of watching your own face in a little box for hours, the cognitive load of reading body language flattened through a screen, the unnatural experience of making eye contact with a camera rather than with a person. The fatigue has faded somewhat as novelty wore off. Brain-scan research suggests videoconferencing still requires more cognitive effort than face-to-face conversation. The fatigue may have become normalized rather than solved. The deep work argument isn’t wrong, exactly. Some people in some roles do concentrate better at home. Writers working alone, programmers building something solitary, analysts running complex models – for work that benefits from hours of unbroken focus, remote arrangements can offer genuine advantages. The error is assuming the finding generalizes to all knowledge work when much knowledge work needs exactly the kind of spontaneous interaction that remote arrangements make harder.

IV.

The strongest empirical case for in-person work concerns how knowledge moves through organizations – the kind of knowledge that can’t be written down or transmitted through formal channels. The junior engineer – call her Ayanna – is three months into her first real job. She’s been given a bug to track down, something about a database query that times out intermittently. She’s been staring at it for two hours and she’s stuck. Two desks over, a senior engineer is on a call troubleshooting a different problem. Ayanna isn’t listening on purpose. She’s just in earshot.

“Yeah, what does the log say around the timeout? Pull the one from before the deploy – no, before, not after.” A pause. “OK so if that index got rebuilt, check whether the query plan changed. Sometimes the planner picks a different path and you wouldn’t know unless you look.“

Ayanna’s own bug has nothing to do with indexes. The logic, though – start with the logs, correlate with deploy times, check if the query plan changed – is a move she hasn’t seen before. When she returns to her own problem that afternoon, she tries it. The plan has changed. She finds the bug in twenty minutes. She’s not quite sure when she learned the technique. She just has it now. That kind of learning doesn’t show up on any syllabus. It can’t be distilled into a document because the person demonstrating it would not think to document it – the moves feel obvious to him, part of how he works, not a thing he could name. It’s tacit knowledge. It transmits through proximity or it doesn’t transmit at all. When the world went remote in 2020, a great deal of that ambient learning simply stopped. A study of one large technology company – over 60,000 employees, using the company’s own communication data before and after the shift – found cross-group collaboration time dropped by about a quarter. Networks became more static. People talked more to the people they already knew and less to new connections. Communication shifted toward asynchronous channels, which are better for information transfer but worse for relationship building. The researchers warned the changes might impede knowledge transfer and degrade output quality over time. They acknowledged effects on innovation might take years to detect. We’re still inside that measurement window. Other findings are more immediate. At one Fortune 500 tech firm, software engineers who sat in the same building as their teammates received about 20% more code feedback than engineers spread across multiple buildings. When offices closed entirely, the feedback gap between co-located and distributed teams vanished – but overall feedback dropped. Everyone got less mentorship. The effects concentrated on junior employees and women, the people who most benefited from informal access to senior expertise.

The link between physical proximity and innovation turns out to be surprisingly quantifiable, at least in one rigorous attempt to measure it. A working paper tracked geolocation data from hundreds of thousands of phones, using face-to-face encounters between employees at different companies as a predictor of subsequent patent citations. If half the employees at two companies work from home, the authors estimated, meetings fall by about a third and patent citations decline in the low double digits. The research design allows for causal inference rather than mere correlation. The setting is Silicon Valley, which limits how far the finding generalizes, but the direction – physical proximity produces innovation outputs – is consistent with what other studies have found through different methods. The mechanism involves what network theorists call “weak ties” – the acquaintances and casual contacts who connect us to information outside our immediate circle. Your close colleagues tell you what you already know. It’s the person you chat with occasionally in the elevator, the former teammate you bump into at a company event, the acquaintance from a different department who mentions something over coffee – those connections bridge different pools of knowledge. Remote work systematically undermines weak-tie formation. In one analysis of a research campus that went remote, weak tie formation declined dramatically – a finding from a narrow population of academic researchers, admittedly, but consistent with the broader pattern showing up in corporate settings.

Strong ties – your actual collaborators, the people you work with daily – may even strengthen in remote settings. You schedule calls with them. You message them regularly. The relationship is explicit and maintained. The weak ties, though, the ones that emerge from unscheduled encounters rather than deliberate contact, depend on proximity. Without shared physical space, they don’t form at the same rate. The implications concentrate on people early in their careers. Someone who has already built an extensive network, who knows whom to call for what, who has absorbed the tacit knowledge of her profession over years of proximity to skilled colleagues – that person can work remotely and draw on accumulated capital. Someone who hasn’t built that network yet, who is still learning how skilled practitioners think and work, depends on proximity in ways that may not be obvious until the deficit becomes apparent years later. Ayanna, on a fully remote team, might never overhear that diagnostic approach. She might learn it eventually through other means, or she might not. What we can say is that the default path to absorbing it has closed. We don’t have great longitudinal data on whether early-career remote workers develop skills at the same rate as their office-based counterparts. The pandemic was too recent and the shift too sudden for the multi-year tracking that would settle the question. The theoretical reasons for concern are serious, though, and the early evidence on mentorship, feedback, and network formation supports them.

V.

Career progression presents a pattern that should worry advocates of fully remote work. The Shanghai call-center study – the same experiment that found a 13% productivity gain – also found something less often quoted: remote workers received roughly 50% fewer promotions than their office-based counterparts, conditional on performance. More productive, less promoted. Government data from before the pandemic found employees who worked mostly from home were notably less likely to receive bonuses. More recent analysis suggests remote workers get promoted about 30% less often than hybrid or in-office workers. The reasons aren’t subtle. Managers view remote employees as more replaceable – about two-thirds express this view in surveys. A majority believe full-time remote work damages career advancement. Nearly half admit they sometimes forget about remote workers when assigning tasks. The vast majority of executives say they notice in-office contributions more readily than remote ones. Proximity bias is real. Face time matters. Visibility counts. Being present when decisions are made and assignments are handed out shapes who gets considered for what. It shouldn’t, in some ideal world – work should be judged by outputs, not attendance. Pretending the bias doesn’t exist doesn’t make it go away. If anything, the pretense compounds the damage for the people who can least afford it.

A potential escape valve exists. Hybrid work may neutralize the promotion penalty. The travel company experiment that found no performance differences over two years of follow-up also found no promotion differences between five-day office workers and three-day hybrid workers. The career penalty may concentrate in fully remote arrangements rather than hybrid ones. Three days per week may be enough presence to maintain visibility; zero days may not be. The demographic implications sit uncomfortably. Women are considerably more likely than men to prefer full-time remote work. Women and people of color show less interest in returning to offices. If proximity bias persists and these preferences don’t shift, flexible work could compound existing inequities rather than reducing them. The people most likely to choose remote arrangements are also those who already face disadvantages in advancement. That doesn’t mean remote work is bad for these groups – it may still be preferable to their alternatives – but the career trade-offs aren’t evenly distributed.

VI.

Remote work imposes what amounts to an intentionality tax. Everything that happens organically in offices requires explicit scheduling when people are distributed. The encounter in the hallway becomes a calendar invitation. The quick question becomes a message that may or may not get answered promptly. The impromptu gathering to solve an urgent problem becomes a video call that needs to be set up, joined, waited for, managed by someone. Meetings expanded enormously during the initial remote transition. Exact figures are hard to pin down generally – the headline statistics tend to describe specific platforms during specific windows rather than general patterns – but the direction is clear across many measures. Remote workers spend more time in scheduled synchronous communication than their office counterparts ever did. That’s the overhead of coordination across distance. It doesn’t disappear as people get used to remote work. It’s structural. Decision-making slows. Virtual teams take more time to reach conclusions. Over 80% of executives, asked in surveys, prefer in-person meetings for difficult decisions and complex strategic discussions. The reasons track with how high-bandwidth communication works: you read someone’s face, you catch the hesitation before the words, you feel the room’s temperature on a hard question. A request you can see someone’s face while making is harder to dismiss than one that arrives in an inbox alongside forty others.

Silos form predictably. The same large-company analysis that found cross-group collaboration declining also found groups becoming more interconnected internally while reducing bridges between different parts of the organization. People talked more to their immediate teammates and less to the broader network. The pattern – intensified local communication, reduced global communication – is exactly what impedes organizational learning and creates fragility. The fragility question deserves more attention than it tends to get. When knowledge is spread across a network of informal relationships, when several people understand how something works, when context transmits through daily proximity – the organization is resilient. Any one person can leave without catastrophic loss. When knowledge concentrates in a few individuals who work in isolation from each other, when no one has absorbed their context by sitting nearby, when their understanding is invisible to others until it’s needed – the organization becomes brittle. The “bus factor,” the number of people who’d have to be hit by a bus before a project collapses, shrinks. This is hard to measure directly, and I don’t know of rigorous studies isolating the effect. The direction of concern follows from the research on silos and network structure. Concentrated knowledge is a risk. Remote work, by reducing informal knowledge sharing, likely raises that concentration.

VII.

The claim that return-to-office mandates are really about control is partially accurate and partially unfair. The accurate part: surveys of managers find a meaningful fraction – around a third – identifying “tracking employees” as a main goal of their company’s return-to-office policy. A quarter of executives admit they hoped mandates would cause voluntary turnover, using office requirements as a kind of soft layoff. When analysts compare mandate-driven companies to their peers, the policies correlate with managerial desire for control rather than with evidence-based strategies for improving performance. The mandates often don’t improve financial metrics. The gap between how employees view their productivity and how leaders view it is vast. Nearly 90% of employees report being productive. Only about 12% of leaders have full confidence their teams are productive. That gap, roughly 75 points, represents a basic disconnect. Leaders don’t trust that work is happening without direct observation. Employees resent the distrust.

Surveillance technology has proliferated accordingly. By most industry estimates – the surveys come from monitoring vendors, VPN companies, and HR software firms, which should temper confidence in the specific numbers – somewhere around three-quarters to four-fifths of companies now use some form of digital employee monitoring. Mouse movement trackers. Keystroke loggers. Software that takes screenshots at random intervals. The effects run counter to what leaders seem to expect. Over half of monitored employees feel stressed or anxious. Over half would consider quitting over excessive surveillance. Monitoring lowers productivity by inspiring resistance rather than compliance. A sizeable share of companies report losing employees who refused to be surveilled. Nearly half of remote workers admit to faking online activity, using mouse-jiggling software or scheduling Outlook emails to send at odd hours to signal they were working late. The surveillance creates theater instead of visibility. Workers and bosses play their parts. Nobody is fooled.

The unfair part of the control narrative: real concerns exist that critics too quickly dismiss. Security requirements in finance, healthcare, and government impose genuine constraints – you can’t let a contractor take a restricted dataset home. Mentorship and training programs sometimes need in-person components for valid pedagogical reasons. Some teams have documented coordination failures that a location policy might address. Labeling every return-to-office decision “control” oversimplifies what are often difficult organizational trade-offs. The distinguishing feature is whether leadership can articulate specific, evidence-based rationales, and whether they follow the same rules they impose. When executives mandate five days in office for employees while continuing to work from their own homes, when “culture” and “collaboration” get invoked as slogans rather than explained as mechanisms, when the policy appears suddenly after a layoff as an alternative to severance – skepticism is warranted. When leadership can explain specifically what problems the policy addresses, binds itself to the same requirements, rolls the change out gradually and remains open to feedback – the calculation is different. Both patterns exist. An honest assessment requires acknowledging that some mandates are control-driven or cost-cutting dressed up as culture, while others respond to genuine operational challenges.

VIII.

Hybrid arrangements – two or three days weekly in person, the rest remote – have the strongest research support of any configuration. The travel company study is the clearest evidence: three days in-office produced the same productivity, the same promotion rates, the same performance reviews, with a third less turnover. The revealed-preference estimate from that study put employee valuation of the flexibility at roughly 8% of salary – smaller than the survey numbers suggest but economically meaningful. An efficiency gain either way: same output, happier workers, lower attrition, without the costs of full distribution. The logic is intuitive. Hybrid schedules preserve enough in-person time for the ambient benefits – spontaneous encounters, relationship maintenance, tacit learning through proximity. They preserve enough remote time for the focus benefits and the flexibility workers value. Neither pure arrangement captures both. Hybrid tries to. Implementation determines whether hybrid delivers on the promise or becomes what one executive called “the hell of half measures.” The worst outcome is uncoordinated schedules where people never overlap with their actual collaborators. They commute to the office, pay for parking and childcare, and then sit on video calls all day – incurring the costs of both arrangements while gaining the benefits of neither. Walk through such an office and the absurdity announces itself: the floor half-empty at any given moment, everyone present nominally, all on headsets, talking to colleagues in other buildings or other states.

The solution is organized hybrid: coordinated days where teams are in person together, predictable rhythms rather than chaos. Some companies designate anchor days – everyone in on Tuesday and Wednesday, say – while leaving the rest flexible. Others let teams pick their own coordination patterns. The specifics matter less than the principle. In-person time only adds value when the right people are present together. Equity issues persist regardless of implementation. Proximity bias means whoever shows up more gets more visibility. If some people choose five days and others three, the five-day people acquire an advantage in advancement – the same proximity penalty in miniature. If leaders exempt themselves from office requirements, the policy collapses. People follow behavior rather than stated rules. Everyone watching executives work from home while demanding their teams commute draws the obvious conclusions. The recommendations that emerge from the research are consistent. Everyone should follow the same schedule, or at least schedules that don’t create systematic differences in visibility. Evaluation should shift from presence-based to output-based metrics, though that’s harder to implement than it sounds. Monitoring technology is almost uniformly counterproductive – it generates stress, inspires gaming, and measures activity rather than results.

IX.

Different industries and roles have different needs, and pretending otherwise confuses the debate. Technology has the highest remote work rates – around two-thirds of workers at least partially remote. Telecommunications and insurance follow, with professional services close behind. These are information-handling industries where the core work product is knowledge, communication, and decisions rather than physical objects or in-person service. Remote work is structurally possible in ways it isn’t elsewhere. Real estate, construction, manufacturing, healthcare – industries requiring physical presence for the core work product – naturally have lower remote rates. A surgeon can’t operate remotely. A factory worker can’t assemble products from home. The question of remote work is moot for a large portion of the economy. The interesting cases are ones where remote work is possible but optimal arrangements vary by role and task. Finance is instructive. Major banks push aggressive return-to-office policies while the available data suggests hybrid works fine for most functions. The insistence on five days seems to reflect cultural conservatism more than operational necessity – a sector norm maintained by inertia and executive preference, not evidence about results.

Role characteristics predict remote viability better than industry labels. Independent task work suits remote arrangements. Collaborative and creative work may benefit from in-person time, at least for certain phases. Client-facing roles often need presence. Large meetings with ten or more participants are significantly worse on video than in person. Small meetings of two to four show comparable effectiveness across modalities. The pattern suggests not a “remote versus office” question but “which activities benefit from which setting.” Seniority effects cut against themselves. Entry-level employees need more in-person time for mentoring and relationship building, and younger workers are also most likely to prefer flexibility. The preferences and the developmental needs point in opposite directions. Companies that offer fully remote options may attract young talent while failing to develop that talent effectively. Companies that mandate full-time office work may develop their junior employees better while losing them to competitors offering flexibility. Company age correlates with flexibility: most companies founded in the last decade or so offer flexible arrangements, while considerably fewer pre-2000 companies do. Whether that reflects different organizational DNA – companies born in the digital age designed differently from the start – or just different leadership ages and attitudes isn’t clear. Some of both.

X.

The wave of return-to-office mandates in 2024 and 2025 offers a kind of natural experiment, though one whose results won’t be fully visible for years. Major technology companies have pushed aggressive timelines. One large e-commerce company mandated five days weekly for over 350,000 corporate employees, with surveys finding nearly three-quarters considering quitting. A major computer manufacturer moved to five days despite its own CEO having stated remote work was “absolutely here to stay” just three years earlier. Telecommunications companies implemented similar mandates, with competitors actively recruiting dissatisfied employees. Financial services pushed harder still. One of the largest banks mandated five days for over 300,000 employees, with leadership shutting down internal forums where employees criticized the policy. Employees reportedly contacted unions about organizing. Another major bank is on its seventh return-to-office mandate, with its CEO calling remote work an “aberration.” The rhetoric often invokes collaboration and culture without explaining how those abstract goods are produced by physical presence or damaged by its absence. The speed and inflexibility of many mandates – announced quickly, implemented with minimal accommodation for individual circumstances – suggests something other than careful organizational design. It suggests conviction in search of justification rather than the reverse.

Early attrition data should give pause to executives issuing these mandates, though the data comes mostly from HR consultancies and recruiting firms with their own perspectives on the question. With that caveat: roughly 40% of companies with aggressive return-to-office policies report higher-than-anticipated turnover. Around 80% report losing talent due to mandates. Companies with return-to-office policies take noticeably longer to fill open roles, suggesting the talent market hasn’t fully bought the premise that in-person work is worth insisting on. Most telling in its consistency across surveys: high performers are more likely to have low intent to stay when facing return-to-office requirements. The mandates may drive out exactly the people companies most want to keep. None of this means return-to-office is never justified. Some organizations may need more in-person time than they currently have. The question is whether blunt mandates, implemented abruptly and without clear explanation, are the right mechanism – or whether they’re destroying trust and losing talent in pursuit of benefits that could be achieved more carefully.

XI.

Against the backdrop of return-to-office mandates, some companies have built large, successful organizations on fully remote foundations. One major DevOps company operates with roughly 2,000 employees across more than 60 countries, with no offices anywhere, reporting nearly $600 million in revenue with 37% year-over-year growth in its most recent fiscal year, non-GAAP operating profitability, and recognition as an industry leader. One automation company, remote since founding, reached a multi-billion-dollar valuation and profitability within a few years of launch. One web-publishing company has operated with over 1,500 employees across more than 80 countries for two decades, powering a large fraction of all websites globally. These aren’t fringe experiments. They’re sizable, profitable companies operating at scale without physical offices. They suggest remote-first models can work, at least for certain industries, cultures, and leadership approaches.

Aggregate analysis comparing flexible companies with mandate-driven ones finds flexible companies growing revenues faster, even after adjusting for industry and size – though that analysis comes from organizations with flexibility-friendly orientations, and the correlation doesn’t prove causation. Flexible companies may have better leadership overall, may be in faster-growing sectors, may attract more talented employees through their policies. The correlation does undermine the claim that remote work inevitably damages performance, though. The counter-examples are too numerous and too profitable. The fair interpretation is that remote-first work is viable for companies that build for it from the start, that hire people who thrive in the arrangement, that develop the coordination practices and documentation habits and communication norms the model requires. It’s a different organizational form with different requirements, not the same organization minus buildings.

XII.

Where does this leave us? Hybrid work – two or three days weekly in person – has the strongest evidence base. It reduces turnover without measurable productivity cost. It preserves enough in-person time for the benefits of proximity while offering enough flexibility for the benefits employees value. If there’s a “best” answer in the abstract, this is probably it. Hybrid also appears to neutralize the career penalty that fully remote workers face, which is not a small consideration. Fully remote work carries costs that are real even when the numbers attached to them turn out to be soft. Mentorship thins. Weak ties fail to form. Junior employees miss the ambient learning that happens through proximity. Career advancement suffers in organizations that haven’t rebuilt their evaluation systems. These costs fall most heavily on people early in their careers and on groups who already face disadvantages, and they compound over time in ways that may not be apparent for years. Companies that go fully remote should understand what they’re trading away, even if they decide the trade-off is worthwhile. Fully in-person work is increasingly a competitive disadvantage for recruiting, given the strength of employee preferences for flexibility. Organizations that insist on it will pay a premium in salaries or accept lower-quality talent or both. The premise that physical presence is required for good work has been sufficiently disproven by the pandemic and its aftermath that asserting it without evidence won’t persuade people who have experienced otherwise.

Productivity claims are overstated by both sides, and the measurement problem is deeper than most of the debate acknowledges. Self-reported productivity diverges systematically from objective measurement, and both are hard to interpret in knowledge work where “output” resists easy definition. Anyone claiming certainty about whether remote workers are more or less productive than office workers is overconfident. It depends on the type of work, the individual, the organizational context, the quality of management. Surveillance almost uniformly backfires. Monitoring technology generates stress, inspires gaming, and measures activity rather than results. It signals distrust and produces resentment. The companies relying on it are fighting their own employees rather than working with them. Significant research gaps remain. We don’t have longitudinal studies tracking skill acquisition rates for early-career remote versus in-person employees over multi-year periods. The innovation effects researchers warned might take years to measure remain in that measurement window. The fragility concerns – knowledge concentration, reduced bus factor – lack rigorous study. Much of the startup-specific evidence is anecdotal. And many of the numbers that circulate most confidently in the remote work discourse turn out, on inspection, to come from consulting firms and vendors with positions in the outcome rather than from studies designed to measure what they claim to measure.

What’s clear is that “should people work remotely?” is poorly posed. The better questions are these. Which people? Doing which work? At which career stage? In which organizational context? With which management practices? Under which conditions does in-person work add enough value to justify its costs in commuting, real estate, and reduced flexibility? Under which conditions do remote work’s advantages outweigh its costs in coordination, knowledge transfer, and relationship formation? The questions have different answers for different circumstances. The senior engineer working on solo projects, established in her network and skilled in her craft, may be perfectly served by full remote work. The junior engineer still learning how senior engineers think, still building the relationships that will define her career, may need more proximity than she realizes. The team in rapid creative iteration may benefit from being in the same room. The team in execution mode may need fewer meetings, not more presence. Neither side in this debate has a monopoly on evidence or wisdom. The challenge is matching arrangement to circumstance rather than imposing ideology on complexity.

XIII.

A thought experiment clarifies the trade-offs. Imagine two versions of the same company, identical in every way except one works fully in person and one works fully remote. The in-person company has higher overhead: real estate, facilities, the friction of commuting. Its employees are less satisfied with work-life balance and more likely to leave for competitors offering flexibility. Junior employees, though, absorb tacit knowledge through proximity. Cross-functional relationships form spontaneously. Problems get solved in hallway conversations. The network of weak ties is dense and continually replenished. The remote company has lower overhead and happier employees. It can hire from anywhere and access talent pools the in-person company can’t reach. Its senior employees are productive and satisfied. Junior employees develop more slowly, though. Knowledge concentrates in individuals rather than spreading through the organization. When a key person leaves suddenly, context goes with them. Innovation that emerges from spontaneous encounter happens less often.

Which company is better? There’s no abstract answer. It depends on industry, competitive landscape, composition of workforce, nature of the work, quality of leadership. Both models carry costs. Both offer advantages. The choice is a trade-off, not a triumph. The wise approach is to stop treating this as a binary question at all. Hybrid arrangements exist precisely because pure models impose pure costs. Some time together, some time apart, coordinated thoughtfully rather than mandated rigidly – the combination captures most of the advantages while mitigating most of the costs. Hybrid can fail too. It can become the worst of both worlds rather than the best – the cost of commuting combined with the isolation of distributed work, the overhead of offices combined with the fragmentation of distribution. Getting hybrid right requires intention, coordination, and willingness to experiment rather than decree.

XIV.

The question of where people should work is ultimately about what we value and how we trade off competing goods. Employees value flexibility and autonomy. They value reclaiming hours lost to commuting. They value being present for their families and comfortable in their homes. These are not trivial preferences to be overridden by executive fiat. Organizations value coordination, knowledge flow, innovation. They value developing junior talent, maintaining culture, enabling the spontaneous encounters that produce unexpected solutions. These are not trivial needs to be dismissed as pretexts for control. Neither set of values trumps the other. The work is finding arrangements that respect both – offering employees the flexibility they want while preserving the organizational benefits of proximity. That’s harder than mandating a simple policy and enforcing compliance. It requires ongoing negotiation, experimentation, and willingness to adjust when something isn’t working. The companies that tend to thrive are those that think clearly about which activities benefit from which settings, that match work mode to work type, that trust their employees enough to grant flexibility while building the rituals and structures that maintain connection. The companies that struggle are those that issue edicts – whether “everyone remote” or “everyone in office” – without attending to the needs of their work and their people.

What we learned from the pandemic is that remote work is possible at a scale no one had tested before. What we’re learning now is that possible isn’t the same as optimal, and that optimal varies by context. The ideological positions – remote as liberation, office as oppression, or the reverse – miss the point. The interesting questions are practical. What works? For whom? Under what conditions? How do we find out? Those questions won’t be settled by research papers or mandate announcements. They’ll be settled by thousands of organizations experimenting, observing, adjusting, and – in some cases – failing. The answers that emerge will be plural rather than singular, contingent rather than universal. That’s unsatisfying if you want a simple rule. It’s appropriate if you believe complex systems resist simple answers. The geography of work is being redrawn. The new boundaries are still being negotiated. The only thing that seems certain is they won’t look quite like what came before, and they won’t match any ideology’s prescription. They’ll look like trade-offs, carefully made – or carelessly made, in the organizations that refuse to think them through.

Featured song:

First time I heard this, the moan-moan-moan was an instant turn off, and a skip… until I heard the bridge being sung in a dream during a reunion with an old friend, Tola Akunji. Then, the song grew on me in waking life. Have loved it – and Cissie – ever since.

Similar Essay: Severance
Image Source