How I got into sleep

July 2025

I watched Matthew Walker's conversation with Andrew Huberman on the Huberman Lab podcast, and by the end of it I had ordered Why We Sleep.

That is not a distinguished origin story. An enormous number of people currently interested in sleep arrived by exactly that route, and I get the impression some of them are faintly embarrassed about it — as though a subject only counts if you found it somewhere obscure.

I am not embarrassed. Recruiting people into a field is precisely what popular science is for, and it is hard to do well. It worked on me, and a year later I am reading sleep papers for fun. By any reasonable measure the book did its job.

Mechanisms, not tips

Looking back at the parts of that episode I actually marked, the pattern is embarrassingly obvious: REM and non-REM sleep and the paralysis; sleep cycles and the structure of the night and hormones; waking in the night and fragmented sleep; sunlight; caffeine timing; alcohol.

Not one of those is a tip. Every one is a mechanism.

The paralysis was the one that got me. During REM sleep the body is immobilised — and the point of the immobilisation is that the mind can run a vivid simulation without the body attempting to perform it. Then the exemptions: the muscles that move the eyes, and the muscles of the inner ear. A system with a reason, and exceptions that make the reason legible. That is what a good explanation feels like, and I had not expected to find one in a discussion about going to bed.

The rest followed the same shape. Deep sleep clusters in the first half of the night because it is driven by a pressure that is highest at sleep onset and discharges as the night goes on — so REM expands into the second half almost by subtraction. Caffeine does not add energy; it blocks a receptor for a molecule that has been building since you woke, which is why the tiredness arrives all at once when it wears off.

Before that summer, sleep was overhead. It was the eight hours standing between me and everything I wanted to do with a day, and the only interesting question about it was how little I could get away with. Afterwards it was a system with stages and pressures and a chemical clock. Systems can be understood, and things you understand stop being an enemy.

Then I looked at my own week

Understanding a system makes you want to look at your own instance of it, and mine was a mess.

I was getting a reasonable amount of sleep by total hours — something around eight most nights — but at times that varied wildly. Late one night, early the next, a long recovery lie-in at the weekend. On paper it was enough sleep. It felt awful. I was tired in a flat, background way that having-had-eight-hours did not explain.

Later, on a stretch where I was sleeping closer to six hours but going to bed and waking within about the same half-hour every single day, I felt substantially better. Two hours less. Noticeably better.

That did not fit anything I had absorbed, where the number of hours was the whole story. It took me a while to find the literature that made sense of it: work on sleep regularity, the day-to-day consistency of when you sleep rather than how long, which turns out to predict some health outcomes more strongly than duration does.

I want to be careful, because this is the part of my own story most likely to be misread. It is not evidence that six hours is enough, and I am not recommending it. Six hours is below what the American Academy of Sleep Medicine recommends for adults, my six-hour stretch was a symptom of a timetable rather than a plan, and the well-documented trap with short sleep is that people feel fine while measurably impaired. What the experience did was make a finding legible to me that I would otherwise have read straight past. The guideline numbers, and what they do and do not license, are here.

Meeting it again, formally

Then the syllabus caught up with the podcast.

Meeting material a second time in a physiology course, after you already care about it, is a completely different experience from meeting it cold. You are not learning it, you are checking it against something. And the formal version is less quotable and more precise: the same stages, but with the actual proportions, the actual variability, and — this was the part that reorganised my thinking — error bars.

Medical school does not mainly teach you facts. It teaches you, largely by catching you out, that a number has a provenance. That “studies show” is not a citation. That a percentage means nothing until you know what it is a percentage of, and that a striking figure in a book is a pointer to a study rather than a finding in itself.

That is a slightly deflating education. It is also the single most useful thing I have to offer on a website about sleep, because it is in desperately short supply in this subject.

What I have learned since

Two things, and they sit together more comfortably than you might expect.

The first is that popular science is a doorway, not a destination. I have since read carefully argued criticism of Why We Sleep — Alexey Guzey's 2019 essay is the thorough one, and Walker has publicly responded to parts of it. I read all of that well after the book, as someone already committed to the subject rather than as a sceptic. Working out where I land on it is a separate project and I am not going to pretend I have finished it here.

The second is that this is normal, and finding it uncomfortable is the mistake. The tempting resolutions are both cheap: decide the author is a fraud, or decide the details do not matter. Neither is thinking. A person can be substantially right, genuinely important to a field, and wrong in specific places, all at once. Noticing the third thing is not a betrayal of the first two — it is what taking someone seriously eventually looks like.

Why this site exists

Because I wanted a bedtime, and every sleep calculator I found gave me one without telling me what it had assumed to get there.

So I built one that says. It uses the same arithmetic as all the others — there is no clever model here and nobody else has one either — but the four assumptions are printed on the homepage, the cycle length is a slider rather than a constant, and short nights are flagged as short. Then I wrote an article arguing that the central claim of the entire category, my own tool included, is unproven, which is an odd thing to publish next to your own product and is the whole point.

The rest of the site is the same instinct applied more widely: go and read the thing, report what it actually says including the parts that are inconvenient, and link to someone better when they exist.

None of which I would be doing if a podcast episode in July 2025 had not made sleep interesting enough to bother with.

Where the facts in this come from

This is an essay rather than a reference page, so it does not carry a reference list of its own. The factual claims it leans on are each sourced in full on the page that makes them properly:

• Sleep stages, the paralysis, and why deep sleep is front-loaded — What a sleep cycle actually is.
• The AASM's seven-or-more-hours recommendation, and the sleep regularity finding — How much sleep do I actually need.
• Criticism of Why We Sleep — Alexey Guzey's 2019 essay, linked in the text above.

The personal parts are not sourced, because they cannot be. They are one person's recollection, and the disclaimer below says what that is worth.

Related reading

• What a sleep cycle actually is — the mechanism that hooked me, with the numbers and their ranges.
• Are sleep calculators accurate? — the same scepticism, turned on my own tool.
• How much sleep do I actually need — why the guidelines are ranges.
• About me — including what I am not qualified to tell you.