The tech campus perfected something the medieval castle understood: if life inside the walls is complete, the walls become invisible. Meals, gyms, clinics, barbers, shuttles, laundry: the canonical Silicon Valley office was an all-inclusive settlement, and its generosity had a geometry. Every amenity moved one more reason to go home inside the perimeter. The campus was not primarily a collaboration machine. It was a presence machine, and presence was the currency in which commitment was counted.
The pandemic broke the machine's monopoly. For a few years, work left the walls and, by the account of the people doing it, mostly went fine. The struggle since, usually filed under return to office, is best understood as a dispute about what the castle was for. If offices existed for the work, the work's portability is decisive evidence. If they existed for visibility, control and the display of loyalty, then empty buildings are an institutional crisis regardless of output, which is roughly how some leaderships have behaved.
What the numbers say about the standoff
The data suggests the drama is louder than the movement. A Stanford and Atlanta Fed survey of over 1,000 executives found only 12 percent of U.S. executives with hybrid or remote workers planned any return-to-office mandate in the following twelve months, and planned returns would cut the share of paid days worked from home by less than half a percentage point, from 21.2 to 20.8 percent (Stanford Report, 2025). Remote days had settled at roughly a quarter of all paid working days in the U.S. Most employers, in other words, quietly kept the arrangement; they are satisfied with the savings on space, the productivity, and the retention.
The mandates that did happen produced a natural experiment in who obeys a summons back to the keep. Ding and Ma (2024), using synthetic-control methods over 260 million résumés, found that RTO mandates at Microsoft, SpaceX and Apple were followed by disproportionate departures of senior employees, with the effect scaling with the mandate's strictness: the strictest regime, SpaceX's full-time presence requirement, was associated with shifts of up to 15 percent in the seniority distribution. The seniors did not leave for startups; they moved to large direct competitors. And in the Survey of Working Arrangements and Attitudes, only 44 percent of U.S. workers said they would comply with a full-time RTO mandate; the rest said they would quit or start looking (SWAA, December 2024).
Read as sociology, this is a story about castles and free people. A command to be present, detached from any demonstrated need of the work, is a loyalty test, and loyalty tests are most expensive to the people with the most alternatives. The mandate selects for those who must stay and sheds those who could choose. A retention architecture that filters out exactly the people with options is running in reverse.
The older evidence about presence
The assumption underneath the all-inclusive campus (that adults must be watched, or at least gathered, to work) has been tested before, in a harder setting than software. As Ricardo Semler has told it, Semco let factory workers set their own hours in 1985. Management predicted the assembly line would collapse for lack of synchronization. Instead the workers coordinated among themselves; parents took mornings, single workers took later shifts, and the operating rule was simply to talk to the person next to you. His summary: adults do this. The plant did not need a bell because the people were not children, and it turned out the bell had been for the managers.
That is the uncomfortable translation for the modern office. If a factory line could self-coordinate physical presence four decades ago, a design team's need for commanded attendance deserves scrutiny. There are real reasons for gathering (apprenticeship, trust-building, the unplanned collision), but they argue for purposeful assembly, occasions with a reason, not for daily residence within the walls.
After the castle
None of this means the office is finished; it means the office has to earn what it once commanded. The all-inclusive campus was an elegant solution to a problem employers no longer admit to having: how to maximize the hours of life spent on company ground. Its successor has to answer a different question: what is worth traveling for? Companies that answer it with mandates are reasserting the castle. Companies that answer it with occasions (work that is genuinely better together, chosen and scheduled by the people doing it) are doing something the castle never had to do: making the inside worth entering when the gate stands open.
The gate is open now. That, not the square footage, is the historic change.
Sources
- Stanford Report / SWAA — Bloom, Davis & Barrero (2025). Return to office? Not everybody is doing it. Only 12% of U.S. executives with hybrid/remote staff planned RTO mandates; planned returns would cut WFH share from 21.2% to 20.8%; only 44% of workers said they would comply with a full-time mandate (SWAA, Dec 2024). https://news.stanford.edu/stories/2025/03/return-to-office-not-everybody-is-doing-it
- Ding & Ma (2024). Return-to-Office Mandates and Brain Drain (arXiv 2405.04352). Synthetic-control analysis of 260M résumés: RTO mandates at Microsoft, SpaceX and Apple were followed by disproportionate senior departures, scaling with mandate strictness (up to 15% distribution shift at SpaceX), with seniors moving to large competitors. https://arxiv.org/pdf/2405.04352
- Semler (1989). Managing Without Managers (Harvard Business Review). First-person account of Semco's participative model, including workers setting their own hours. https://hbr.org/1989/09/managing-without-managers