In a 2008 lab at the University of California, Irvine, people wrote emails while a supervisor broke in with questions. Gloria Mark, Daniela Gudith, and Ulrich Klocke timed the work, counted errors, and asked how the hour felt. The interrupted writers finished faster than the uninterrupted ones. The emails were shorter. Error rates did not differ in a statistically significant way. What did differ was the body of the afternoon: more stress, more frustration, more time pressure, more effort. The authors noted that those workload ratings were already higher after about twenty minutes of interrupted performance.
That is the all-nighter in office clothes. You "catch up" by going faster, the page count drops, and the next morning's estimate is written by someone who slept badly and still believes the faster pace was a plan.
What the sleep guidance actually says
The Centers for Disease Control and Prevention's sleep page, current as of the May 2024 update on that page, lists seven or more hours a day for adults ages 18 to 60. Ages 61 to 64 are listed at 7 to 9 hours, and ages 65 and older at 7 to 8. The CDC ties enough sleep to attention and memory for ordinary daily tasks, and it says to talk with a healthcare provider if sleep problems are regular or if signs of a sleep disorder show up. Disorders named on the page include insomnia, restless legs syndrome, narcolepsy, and sleep apnea. A diary of bedtimes, wake-ups, naps, exercise, alcohol, caffeine, and medications is one tool a clinician may ask for. This article is not that diary and not a diagnosis.
The CDC's adult figure rests on a 2015 consensus statement from the American Academy of Sleep Medicine and the Sleep Research Society, led by Nathaniel Watson and colleagues and published in Sleep. That statement says adults should sleep seven or more hours per night on a regular basis to promote optimal health, and that sleeping less than seven hours on a regular basis is associated with impaired performance, more errors, and a greater risk of accidents, along with a list of health outcomes that belong in a clinic, not a productivity slogan. The same statement says individual need varies with genetic, behavioral, medical, and environmental factors. Seven hours is a population floor for ages 18 to 60 in the CDC table, not a personal prescription and not a trophy for sleeping less.
An all-nighter is a deliberate trip under that floor. You can still ship a draft at 4 a.m. The consensus statement is about regular short sleep and performance, not a moral verdict on one bad night. The practical point is narrower: if the way you hit deadlines is repeated short sleep, you are borrowing attention from the next estimate. The next estimate is the thing that creates the next all-nighter.
Speed is not the same as a finished block
Mark, Gudith, and Klocke ran a controlled email task, not a field study of a newsroom. Same-context interruptions and different-context interruptions did not produce a significant difference in how long the task took once interruption time was removed. The disruption was the break in the task, not whether the question matched the topic. Personality measures mattered a little: higher openness to experience and a higher need for personal structure predicted faster completion of the interrupted task in their regression, and the authors treat that as a hint, not a hiring test.
Read that beside the heroic story people tell about crunch. Crunch often looks productive because the clock is visible and the interruptions are either banned or constant. Banned, you might get a long email, the baseline condition in the lab, where messages were longest. Constant, you get the shorter email and the stress spike. Neither condition is a calendar design. A calendar design decides, before the week starts, which hours will resemble the uninterrupted condition.
A block that beats the all-nighter has four properties. It is on the calendar as an appointment with a name, not a wish labeled "focus." It is long enough to produce a piece of the deliverable, not a mood. Mark's lab measured on the order of twenty minutes before the stress ratings moved; a useful block for real work is usually longer than that, because starting up is part of the cost. It ends while you can still sleep. And it is defended against the kind of drop-in the lab simulated, including the "quick question" that shares the project's topic. Same topic did not make the interruption free.
Place the block before the panic has a vote
People schedule recovery. They rarely schedule the work that would have prevented the recovery. Try the reverse on one deadline this month. Count backward from the external due date to the last day you are willing to be editing rather than inventing. Put a single block on a day in the first half of the remaining time. Name the block with a physical output: "methods section drafted," not "work on the report."
During the block, treat interruptions the way the experimenters did: as a measured cost, not as proof you are needed. If a question can wait ninety minutes, it waits. If it cannot, write down what you were in the middle of before you switch, because the lab's whole point is that reorienting is the expensive part. The authors suspected reorientation time was being paid for by working faster and writing less. You can see that bargain in a short email. You see it in a short, shaky section of a report the next morning.
Sleep is the other edge of the block. If the only hour you "found" starts at 11 p.m., you did not find an hour. You moved the hour out of the CDC's recommended window. Shift the block earlier in the day even if it is shorter. A shorter uninterrupted block plus a night of seven hours beats a longer block that deletes the night, because tomorrow's attention is part of the project. The Watson statement links regular short sleep with more errors. You do not need a personal error rate to take the association seriously. You also should not treat one late night as a medical event. If short sleep is the pattern, the CDC's instruction is to talk with a healthcare provider, not to download another timer.
A week sketch, labeled as a sketch
Here is an illustrative example, not a case study and not a person. Picture a Thursday report due next Thursday. The invented plan puts a 90-minute block on Monday at 9:30 named "draft the two tables," a 45-minute block on Wednesday named "replace the two weakest paragraphs," and nothing after 7 p.m. any night. The external deadline stays Thursday at 4. The Monday block is allowed to produce ugly tables. Wednesday is allowed to stop at forty-five minutes even if a sentence is still clumsy. The hero version of the same sketch has empty Monday, a "quick look" Wednesday that becomes email, and a Wednesday night that runs past midnight. The hero version feels more serious. The lab result says the serious feeling is partly stress, and the shorter pages are not a sign of efficiency.
If the Monday block gets raided by meetings, the problem is no longer sleep. It is that the calendar sold the hour before you sat down. That problem has its own page: meeting debt and the hours you never had. The estimate that assumed the block would survive is the subject of estimating work without lying to yourself.
Limits, stated plainly
Nothing here says a scheduled block will finish the project, raise your wage, or treat insomnia. The Irvine study used a short email task with a small lab sample. The CDC table is public health guidance, not a bedtime assigned to you. Some jobs have real nights: hospitals, news, freight, on-call engineering. Those nights need staffing and recovery, not a blog post scolding the shift. If you are awake because of pain, anxiety, or a sleep disorder, the next step is a clinician, which is what the CDC page already says.
The all-nighter survives because it produces a visible artifact and a story of effort. The block produces a smaller artifact earlier, when the story is boring. Boring is the point. You still get to decide whether this week's deadline gets a named hour before the night shift volunteers.
Where the claims come from
- Centers for Disease Control and Prevention. (2024, May 15). About sleep. cdc.gov/sleep/about
- Watson, N. F., Badr, M. S., Belenky, G., et al. (2015). Recommended amount of sleep for a healthy adult: A joint consensus statement of the American Academy of Sleep Medicine and Sleep Research Society. Sleep, 38(6), 843–844. PMC4434546
- Mark, G., Gudith, D., & Klocke, U. (2008). The cost of interrupted work: More speed and stress. Proceedings of CHI 2008, 107–110. doi.org/10.1145/1357054.1357072