The 45-Hour Week That Burns You Out Faster Than the 90-Hour One

The problem

Every burnout retro ends the same way: cut the meetings, cap the hours, mandate PTO. Six weeks later, the same people are just as fried.

Meanwhile you've got a founder pulling 90-hour weeks who seems to be thriving, and an engineer working a perfectly reasonable 45 who is visibly done. If hours were the mechanism, this shouldn't be possible. It is, constantly, and it means we've been treating the wrong variable.

Why it happens

The model that actually explains this is older than most of the productivity advice built on top of it. Robert Karasek published the Job Demand-Control model in 1979, and it's held up remarkably well because it doesn't measure workload on one axis — it measures two: how much is being demanded of you, and how much control (decision latitude) you have over how, when, and what you do about it.

Cross those two axes and you get four quadrants, not a line:

• Low demand, high control — comfortable, low strain.
• Low demand, low control — understimulated, disengaged, a different failure mode entirely.
• High demand, high control — Karasek calls this an "active" job. Hard, fast-paced, and — this is the counterintuitive part — protective rather than harmful, because you're the one steering it. This is your 90-hour founder, or a senior engineer who chose the scope they're carrying.
• High demand, low control — "high strain." This is where burnout actually lives. Not high demand alone. High demand with no say in how you respond to it.

Control isn't a nice-to-have that makes a hard job more pleasant. In this model it's the thing that determines whether demand turns into strain at all.

On-call is close to a textbook high-strain design if you don't deliberately engineer around it. The demand is high and unpredictable — you don't choose when the page fires. And the control is often close to zero: you can't usually decide that the root cause gets fixed before the next feature ships, you can't set your own escalation policy, and half the time you can't even declare "this alert is noise, mute it" without going through someone else. Forty-five reasonable-looking hours with zero latitude over the worst parts of them will wreck someone faster than ninety hours they actually chose.

What to do about it

This is why "just work less" so often fails as burnout advice: it only pulls the demand lever. If the actual mechanism is low control, reducing hours without touching authority leaves the high-strain structure fully intact — people get a shorter version of the same trap.

The interventions I've watched actually move the needle target control instead, often without touching hours at all:

• Give the on-call engineer real authority to block new feature work until a recurring root cause is fixed — not a suggestion they can raise in a retro, actual authority to stop the line.
• Let the person closest to an alert decide it's noise and mute it, instead of routing that judgment call up a chain.
• Let people negotiate their own swap and escalation terms instead of inheriting a rotation designed by someone who's never been paged from it.

None of that reduces the number of hours worked. All of it reduces strain, because it changes who's steering.

A useful diagnostic for any recurring stressor, on-call or otherwise: does the person experiencing this have the authority to change what's causing it? If yes, you've got a demand problem, and reducing load will probably help. If no, you've got a control problem, and cutting hours will just give someone a smaller version of the same powerlessness.

One caveat worth keeping: Karasek and Theorell later added a third axis, social support, because a high-strain job with strong peer and manager backing is more survivable than the same job in isolation. Support doesn't replace control, but it buys people time while you fix the actual structure.

Key takeaways

• Burnout correlates with low control under high demand — not with hours worked in isolation.
• High demand plus high control ("active" jobs) is a different, more sustainable category than high demand plus low control ("high strain").
• On-call and other interrupt-driven work is structurally high-strain unless it's deliberately paired with real authority over the interrupt.
• Fixing "how many hours" without fixing "who decides" rarely works. The reverse often does, even without touching hours at all.
• Diagnostic question: does the person under stress have authority to change what's causing it?