The Infinite Workday: When Removing the Schedule Removed Only the Schedule

There is a small, recognizable moment near the end of any given Tuesday — somewhere between the last meeting and the dishes — when a message arrives. It is from a colleague three time zones away who has just started their afternoon, and the message is phrased in the polite, low-pressure register of asynchronous communication. No rush, whenever you have a moment. The phrase is almost always honest. The person sending it does not, in fact, expect an answer in the next ten minutes. But the device has chimed, and the message has been read, and the small executive function in the brain that was about to release for the evening has now been re-engaged. The reply will be drafted in the head while the pasta cooks. It will be sent before bed. The total elapsed time at the keyboard, measured by any timesheet, will be perhaps four minutes. The total elapsed time of the workday, measured honestly, will be something much larger than that. This is the experience now being called, with somewhat clinical precision, the infinite workday, and the phrase is not the marketing of a fatigue trend. It is the shorthand a Microsoft research team chose for a pattern its own product telemetry could no longer ignore.

The promise that produced the pattern was a good promise. The shift toward asynchronous work, accelerated by a pandemic but designed in advance by a generation of distributed-software companies, was supposed to liberate knowledge workers from the worst inheritances of the industrial schedule: the open-plan office's chronic interruption, the meeting-as-status-symbol, the requirement to be visibly present rather than usefully productive. The mechanism was simple and, on paper, elegant. Replace the meeting with the document; replace the synchronous interruption with the message that can be answered when the recipient surfaces; let the team's work happen across the hours and time zones in which it is most effectively done, rather than during the narrow synchronous window in which everyone happens to be at their desks. The argument, articulated cleanly across the writings of the GitLab and Automattic founders and the leadership of any number of distributed startups, was structurally compelling and not particularly controversial. The trouble was not the argument. The trouble was the implementation. The organizations that adopted the async-first vocabulary largely kept the always-on availability expectation of the old culture in place, and removed only the schedule that used to give that expectation an edge. The result is the working pattern that the Microsoft 2025 Work Trend Index report measured with a clarity that has made the data difficult to look away from.

What the Telemetry Saw

The 2025 Microsoft Work Trend Index, the special report published under the title Breaking Down the Infinite Workday, drew on what the company described as trillions of aggregated and anonymized productivity signals from Microsoft 365, supplemented by an Edelman Data and Intelligence survey of thirty-one thousand knowledge workers across thirty-one markets conducted in early 2025. The numbers reported are easy to summarize and difficult to dismiss. Chat messages sent outside of standard working hours rose by fifteen percent year over year. The average knowledge worker now receives fifty-eight chat messages that arrive before or after their nominal working day. The same worker, during the working day, is interrupted on average every two minutes, with a daily ceiling of up to two hundred and seventy-five distinct interruptions. The same worker receives, on a typical day, one hundred and fifty-three Teams messages and one hundred and seventeen emails. Meetings starting after eight in the evening are up sixteen percent year over year, driven principally by cross-time-zone collaboration. The data is, in the strict sense, internal to the Microsoft platform — which means it is biased toward the kinds of organizations that adopt Microsoft 365 — but it covers a wide enough cross-section of the modern white-collar economy that no honest reader of the report can treat it as the experience of an unrepresentative minority.

The shape of the day these numbers describe is no longer a day in any historically meaningful sense. It is a long, low-amplitude wave of small contacts, interspersed with synchronous meetings, with a tapering tail that runs from the dinner hour to bedtime and resumes at the breakfast hour the following morning. It is not, strictly speaking, that workers are working more hours — though several of the country-level studies suggest they are — but rather that the hours during which the day is actively present in the worker's attention have expanded to fill nearly the whole of the waking period. The distinction matters. A worker who is at the keyboard from nine to six and walks away has a complete set of non-working hours. A worker who handles four messages between eight in the evening and ten at night, four more between six and eight in the morning, and another fifteen across the workday's nominal core does not. The total active typing time may be lower than the older worker's, but the worker's psychological boundary between at work and not at work has been erased. The infinite workday is not a quantity of labor. It is the loss of an off state.

The Research that Named the Mechanism

The framework that has done the most useful work in explaining the experience predates the infinite workday by more than a decade. In 2014, Larissa Barber, then at Northern Illinois University and now at San Diego State University, and her colleague Alecia Santuzzi published a paper in the Journal of Occupational Health Psychology introducing a construct they called workplace telepressure. The construct was defined precisely: a preoccupation with electronic messages received from work, combined with an urge to respond to them quickly, regardless of the importance of the message or the time at which it arrived. The behavior was familiar enough to anyone with a smartphone, but the formal naming of it allowed the construct to be measured, and the measurement allowed it to be correlated against the outcomes that mattered. The correlations, reported across a sequence of studies that Barber and her collaborators have published in the intervening twelve years, are remarkably consistent. High telepressure scores predict worse sleep quality. They predict higher rates of self-reported burnout. They predict more frequent health-related absences from work. They predict lower satisfaction with work-life balance. The pattern holds across industries and across cultures, and it holds independently of the worker's total reported hours.

The finding from this research that has the most implications for the infinite-workday problem, however, is one that the productivity-advice industry has been almost entirely silent about. In a recent study published in Group and Organization Management in early 2026 — co-authored by Barber, Santuzzi, and Xinyu Hu — the team examined whether the presence of formal disconnection policies in an organization was associated with reduced telepressure among that organization's employees. The answer was no. The presence of a policy, considered as an independent variable, had essentially no measurable effect on whether the workers in that organization felt the urge to respond to messages quickly. What did predict reduced telepressure were implicit norms about after-hours availability — what the leadership of the organization actually modeled, what colleagues actually did, and whether the organization's culture treated rapid response as a marker of competence or merely as a behavior. The policy, in other words, was almost irrelevant. The culture was nearly everything. Barber's own summary of the work is the kind of sentence that should be printed and laminated and stuck inside every HR office in the developed world: telepressure is a workplace problem, not a worker problem. The advice industry's standard response — set your status to Do Not Disturb, batch your email, turn off notifications after six — is responsive to the second framing and not the first, which is the principal reason it almost never produces a durable change in behavior.

What Async Was Actually Supposed to Do

It is worth being precise about the original promise, because the failure mode of the infinite workday has obscured a real intellectual achievement underneath it. The case for asynchronous-first work, made most clearly by the GitLab founder Sid Sijbrandij and the Automattic founder Matt Mullenweg across a decade of essays and internal documentation, rested on a small number of structural observations. The first was that the synchronous meeting, considered as a unit of work, is one of the lowest-bandwidth modes of communication available to a knowledge-work organization, and that most of what gets said in a meeting could be more accurately said in a document and more thoroughly absorbed at the reader's own pace. The second was that interruption, in the technical sense studied by Gloria Mark at the University of California, Irvine, exacts a context-switching tax estimated at twenty-three minutes per meaningful interruption, which means that a working day with eight interruptions has already lost three hours to the recovery cycle alone, regardless of what else is happening in it. The third was that the requirement to be present during synchronous hours discriminates against parents, against neurodivergent workers, against workers in non-headquarters time zones, and against anyone whose best thinking happens at a clock face the head office did not designate as the official one. The combination of these observations produced a coherent design: replace meetings with documents wherever possible, replace interruptions with messages that can be answered in batches, and let workers choose the hours in which they are most effective. The design has been in production at GitLab, Automattic, and a number of similarly structured companies for more than a decade, and where it has been implemented in its full form it has produced the outcomes its proponents predicted. The mistake was not the design. The mistake was the partial adoption.

The companies that took on the vocabulary of asynchronous work without the culture of asynchronous work — which is to say nearly every large traditional organization that retrofitted distributed tooling onto a pre-existing synchronous culture during the pandemic — produced what the design was specifically engineered to avoid: a workday that retained the synchronous expectation of immediate response while abandoning the synchronous container of fixed office hours. In the original design, the document existed precisely so that the writer did not need to be present when the reader read it, and the reader did not need to interrupt anyone to absorb it. In the partial-adoption version, the document was added to the system without removing any of the meetings, and the chat message was added to the system without removing any of the emails, and the expectation of rapid response was extended across the new channels rather than being relaxed on the old ones. The total number of inputs grew. The expected response time shrank. The hours during which inputs could arrive expanded to cover the entire waking day. Each component of the change was individually defensible. The aggregate produced an architecture in which no hour of the day was structurally protected from contact, and the workers inside it began to act, on the evidence, exactly as the architecture had been designed to make them act.

Why the Schedule Was Doing Quiet Work No One Noticed

The schedule of the industrial-era workday — clocking in at nine, clocking out at five, the explicit weekend, the protected dinner hour — was a target of much justified criticism for most of the twentieth century. It was rigid; it served the employer's measurement needs more than the worker's productive ones; it discriminated against any worker whose life shape did not fit a uniform eight-hour synchronous block. The dismantling of the schedule, in the white-collar context, was correctly understood as a liberation. What the dismantling failed to notice was that the schedule had been performing a second function in parallel with its tracking function. The schedule was the boundary — the agreed-upon point past which no one expected to reach you. It did not matter very much whether a particular worker stayed late occasionally. What mattered was that the structural default of the day had an end. The end existed not because the worker had chosen to enforce it, but because no one had ever proposed an alternative. The end was held in place by collective expectation. The worker did not have to do any individual work to keep it standing.

The removal of the schedule removed both functions simultaneously, and the discussion at the time focused almost exclusively on the tracking function. The dialogue inside most organizations went something like we are no longer going to require you to be at your desk between nine and five — we trust you to do the work — and we are going to give you the flexibility to set your own hours. The phrase that did not appear in the announcement, in nearly any case, was and accordingly we will not contact you outside the hours you have set. The second half of the deal was simply omitted. The first half was framed as a worker benefit, and accepted as one. The omission of the second half was framed, when it was framed at all, as a worker responsibility — you are responsible for setting your own boundaries — which is a structurally different proposition from the organization will hold a boundary on your behalf. The latter is cheap because no individual is paying the cost of enforcement. The former is expensive because every individual is paying it constantly, against constant low-grade incoming pressure, in conditions of asymmetric social information. The first model produced an off state by default. The second model produced an on state by default, with an off state available only as a costly opt-in. The behavioral telemetry that Microsoft published is what an entire economy looks like when an off state is reframed from a structural condition into a discretionary effort.

The Statute as a Second Schedule

The jurisdictions that have noticed the structural problem have, in roughly the past decade, begun to attempt a legislative correction. The earliest and most extensively studied of these is the French droit à la déconnexion, which entered into force on the first of January, 2017, as Article 55 of the El Khomri labor law — formally the Loi Travail — amending Article L2242-8 of the Code du travail. The text of the statute is more modest than the foreign press coverage of it tended to suggest. It does not prohibit work email after six in the evening. It does not impose criminal penalties on a manager who calls an employee on a Saturday. What it requires is that any French firm with fifty or more employees negotiate, with its employee representatives, the conditions under which digital communication tools may be used outside of working hours, and that, in the absence of an agreement, the employer publish a charter unilaterally setting those conditions. The mechanism is procedural rather than substantive. It does not specify the rule; it specifies that there must be one, and that the rule must have been the product of an explicit deliberation. The theory of the law is that the existence of an explicit institutional boundary, openly negotiated, will do most of the work of restoring the collective off state that the dissolution of the schedule had removed.

Portugal followed in 2021 with a stricter version, imposing direct fines on employers who contact workers outside agreed hours except in cases of force majeure. Ontario added a right-to-disconnect requirement to its Working for Workers Act in late 2021, in force the following year, requiring employers of twenty-five or more to publish a written policy. Australia's amendment to the Fair Work Act, in force from August of 2024, gave workers an explicit right to refuse to monitor or respond to contact from their employer outside working hours, with that right enforceable by the Fair Work Commission. The early enforcement data is, predictably, still sparse, but the case law that has emerged in France over the eight years the statute has been in force is instructive. The cases have not, by and large, been the dramatic ones the original commentary anticipated — punitive damages against firms with brutal cultures. They have been more often the quieter ones: a worker successfully claiming overtime for a pattern of after-hours email responses that the employer had treated as voluntary; a manager successfully challenging a termination that had cited the manager's failure to respond to weekend messages as evidence of disengagement. The statute's effect, in these cases, is not that it changes what is permitted. It is that it changes the default presumption when the question is litigated. The boundary that the schedule used to hold up structurally is being reconstructed, slowly and incompletely, through procedural law.

What Individuals Can Actually Do

The honest version of the personal-practice section of any piece on this subject is that the available individual interventions are narrower and less powerful than the productivity industry implies, and that the most powerful of them are not the ones the industry has most loudly recommended. The standard advice — turn off notifications, batch email to two windows a day, set a Do Not Disturb status from a fixed time — is not wrong, exactly. The Barber and Santuzzi work shows that these practices, where adopted, do correlate with marginally lower telepressure scores at the individual level. The correlation is small. The dominant variance, in their studies, is explained by the workplace's implicit norms rather than by the worker's personal practices, which means that the worker who turns off notifications inside an organization that rewards rapid response will, on the evidence, either reverse the practice within a few weeks or pay a measurable career cost for maintaining it. The intervention is not strong enough, on its own, to overcome the gradient it is being asked to push against.

The interventions that the studies suggest do work at the individual level fall into two narrow categories, and the productivity press has been largely uninterested in both because neither produces a tidy listicle. The first is making the boundary visible to other people in the form they can read. A worker whose calendar publicly shows hours they have committed not to work — not "busy," not "blocked," but explicitly visible as off-hours — is markedly less likely to receive messages during those hours than a worker whose calendar shows the absence of a meeting and is consequently legible as available. The mechanism is informational, not normative; the colleague who would otherwise send the message is given a costless way to defer it, and most of them, on the evidence, take it. The second is responding to off-hours messages at an off-hours pace, even when one is awake and able to respond faster. This is the practice that the telepressure literature most cleanly recommends, and it is also the practice that organizational cultures most actively resist. The worker who responds to a Tuesday-night message on Wednesday morning is, in many organizational cultures, training the rest of the team — slowly and quietly — to expect that the message will be answered on a Wednesday-morning timeline. The training works. The cost is the social discomfort of holding the line during the first several weeks while it is taking effect, which is real, but bounded.

Both of these interventions, it should be noted, are structurally identical to what the right-to-disconnect statutes do at the policy layer. They reinstate, at the individual level, a default that the dissolution of the schedule removed at the cultural level. They are an attempt to recreate, by hand and in single-worker units, the off state that the old industrial schedule produced for free. The reason they work, where they work, is the same reason the old schedule worked: they make the question of when one is available not a question that has to be relitigated every day.

What the Pattern Is Confessing

The deeper finding underneath the infinite-workday data is not that workers are over-contacted. The deeper finding is that the structural off state which the industrial schedule provided as an unintended by-product was doing significant load-bearing work, and that the dismantling of the schedule, performed in the name of liberating the worker, removed the off state as collateral damage. The liberation was real. The off state was real. The two were entangled in ways that were almost impossible to see while the system was intact, and that became visible only when the system was disassembled and the absence of the off state began to be felt directly. This is a familiar pattern in the history of institutional change. A function carried quietly by an old institution disappears when the institution is reformed, and the consequence of its absence is felt long before the function is correctly identified.

The reasonable response to this is not to reconstruct the old schedule. The schedule had real costs, and those costs are why it was dismantled, and the costs have not gotten any cheaper. The reasonable response is to be precise about which functions the schedule was performing, separate them, and reconstruct only the ones the new system needs. The tracking function — be at your desk between nine and five — is the one most of the discussion focused on, and the one whose dismantling was correctly identified as a liberation. The boundary function — do not expect contact past six — is the one almost no one named at the time, and the one whose dismantling has produced the cost the Microsoft telemetry is now measuring. The two were never the same function, but they had been packaged together in the same institutional artifact for so long that the dismantling of the package felt, while it was happening, like a single act. The post-schedule organization that is going to work in the long run will be one that has done the careful work of reconstructing the boundary while continuing to leave the tracking dismantled — by policy, by published default, by the small set of individual practices that survive contact with the surrounding culture, and by the patient effort of restoring an off state in a system that has lost its memory of having had one.

A calendar, read carefully, is a kind of confession. It says who has been allowed to write on the page, when they have been allowed to write, and what hours have been left clean. The arithmetic of the infinite workday is the visible arithmetic of a calendar in which no hours are clean — in which every hour of every day is, by structural default, available to the next request. The relief of the rare day on which the pattern doesn't hold is the relief of a calendar that has remembered how to end. The pursuit of that ending, in any of the forms it can take, is the older project of trying to keep some share of the week, however small, in a register that the workplace is not actively writing in.