Here's the arrangement, up front, because the whole point of this piece is that it runs the opposite way to almost everything written about mornings: my mornings go to other people. The afternoons are the part I try to keep entirely free of meetings, and I hold that about four days in five. The quiet isn't found in the early hours and defended. It's bought, and the morning is what I pay with.

I want to be careful about the four-in-five. It is not a success rate I'm bragging about. It means that roughly once a week the arrangement fails, and I spend an afternoon doing the thing I said I wouldn't do. That's the actual number, and a scheme that works four days out of five is a different claim from a scheme that works.

I'm writing it down because the genre this piece belongs to has settled on the opposite answer so completely that the opposite answer now sounds like a law of nature. Get up before everyone else. Take the first two hours for yourself. Put the inbox after lunch where it can't hurt you. That advice rests on a much thinner empirical base than its confidence suggests, and the part of it that *is* true has nothing to do with what time it is.

The hour on the clock is the least interesting variable

Start with what the research actually says about early rising, because this is where the productivity literature is weakest and loudest.

The study most often pointed at is Christoph Randler's 2009 paper in the *Journal of Applied Social Psychology*, the one that gave us "morning people are more proactive." It surveyed 367 students. It found a correlation between self-reported morningness and self-reported proactivity of about .11. Both sides of that correlation are self-report, the sample is undergraduates, and .11 is a relationship you can see only in aggregate — it tells you almost nothing about any individual. Randler himself has been more careful in interviews than his popularisers: asked whether you could learn to become a morning person, he said you probably could, but that he wasn't sure it would make you more productive. The British Psychological Society's write-up flagged the obvious alternative explanation, which is that it may simply be easier to look proactive in a world already built around rising early.

That's the foundation: one correlational study of students, effect size .11, holding up an entire industry of 5 a.m. clubs.

Now the other direction, which sounds more sympathetic and is equally overstated: that you're a night owl by genetic fate and shouldn't bother trying. That comes mostly from Matthew Walker's *Why We Sleep*, which tells readers night owls "are not owls by choice" and are "bound to a delayed schedule by unavoidable DNA hard wiring." Alexey Guzey has since published a long catalogue of factual errors in that book's first chapter, and later, with the statistician Andrew Gelman, raised a charge about a graph in it with a contradicting bar removed; Walker has conceded that at least one central attribution was misremembered. Draw your own conclusion about the book. Mine is that it isn't what to cite when you want to know how much of your sleep timing is fixed.

For that question, the largest study available is Jones et al., 2019, in *Nature Communications* — a genome-wide association analysis of chronotype in 697,828 people from UK Biobank and 23andMe. It raised the number of loci associated with morningness from 24 to 351, which sounds enormous, and then gives you the number that matters: among the 85,760 participants wearing activity monitors, the 5% carrying the most morningness alleles slept, on average, 25 minutes earlier than the 5% carrying the fewest. Twenty-five minutes, between the genetic extremes. The paper's own heritability estimate for chronotype in UK Biobank is 13.7%.

So: the genetics are real, measurable, and small. What actually determines when people get up is school, children, commutes, shifts, other people's calendars. Which is a more interesting finding than either camp wants it to be, because it means the clock time is not where the leverage is. If moving your whole genome moves your bedtime by 25 minutes, the hour you wake up is not the mechanism anyone is really arguing about.

What is actually substantiated is the cost of switching

The part of this literature I'd defend is not about mornings at all. It's about what happens when you move your attention from one thing to another before the first thing is done.

The key work is Sophie Leroy's 2009 paper in *Organizational Behavior and Human Decision Processes*, titled, with a directness I appreciate, "Why is it so hard to do my work? The challenge of attention residue when switching between work tasks." She coined the term. The mechanism: when you switch tasks, part of your attention stays with the task you left instead of transferring to the one in front of you, and what stays behind is what the new task doesn't get. It's worst when the first task was left unfinished. Her own summary: "Switching attention tends to be difficult for people and subsequent task performance easily suffers."

The follow-up is more useful. In Leroy and Theresa Glomb's 2018 *Organization Science* paper — "Tasks Interrupted: How Anticipating Time Pressure on Resumption of an Interrupted Task Causes Attention Residue and Low Performance on Interrupting Tasks and How a 'Ready-to-Resume' Plan Mitigates the Effects" — participants were interrupted mid-task and told either that they'd have plenty of time to come back to it or that they wouldn't. The ones expecting to be rushed carried more residue and did worse on the interrupting task: less careful with information, missing errors, less likely to find the best answer. The ones who expected to have time were fine. Four studies, including a field study of 202 working professionals.

Read that result carefully, because it isn't the one the productivity blogs extracted. The damage didn't come from the interruption. It came from expecting not to have enough time afterwards. Same interruption, different expectation about the rest of the day, different outcome. That's an argument about the shape of a day, not about its starting hour.

And the intervention they tested is almost embarrassingly cheap: before you switch, write a short note on where you stopped and what comes next. They call it a Ready-to-Resume Plan, and in both lab and field studies the people who did it switched more completely and decided better. If you take one actionable thing from this piece, take that one — it's the only item here with a tested intervention attached.

The figure everyone quotes instead is Gloria Mark's 23 minutes and 15 seconds. Be precise about what it measures: her team observed information workers over three working days and found 81.9% of interrupted work was resumed the same day, on average after that interval. It's elapsed time before someone returns, including handling the interruption and whatever else they did in between — not a measure of how long a brain needs to recover its depth of concentration, though it gets cited constantly as though it were. One observational study. I use it as what it is.

And to be straight about the state of it: I went looking for replication attempts on attention residue specifically and found no public ones. That switching tasks carries a cost is about as well established as anything in cognitive psychology, but Leroy's particular workplace measure of residue is a small set of studies by one research group. I'm citing it because it's the best thing there is, not because it's settled.

The email advice is weaker than you've been told

This is the one I'd push back on hardest, because it's the advice that gets handed out with the most certainty.

The study behind "batch your email" is Kushlev and Dunn, 2015, in *Computers in Human Behavior*: "Checking email less frequently reduces stress." Within-subjects design, 124 adults, two weeks — one week told to check email three times a day with the mailbox closed and notifications off, the other to check as often as they could with it open. Participants reported significantly lower daily stress in the limited week.

Then read the authors' own limitations section, which is more honest than almost all the coverage of it. Because they measured a broad array of well-being outcomes, they write, "it is possible that the significant effects we observed on stress are simply an artifact of the large number of statistical tests we conducted," and the study "should therefore be seen as laying the groundwork for future confirmatory research." Changing how often you check email, they say plainly, "is by no means a panacea." There was no true control condition. And the behaviour numbers are self-reported: the limited group reported checking around five times a day, not three, against a baseline of about fifteen. Everyone in that study was non-compliant, and the paper says so.

Now the finding that almost never gets quoted next to it. Mark, Iqbal, Czerwinski, Johns and Sano, at CHI 2016, tracked 40 information workers for 12 workdays with computer logging, heart-rate monitors and daily surveys. Their abstract: "Batching email is associated with higher rated productivity with longer email duration, but despite widespread claims, we found no evidence that batching email leads to lower stress." Elsewhere they put it flatly — the benefit claims, though widespread, remain unsupported because results conflict. Eleven of their 40 participants were batchers. Small study, measured stress rather than reported stress, opposite result on exactly the thing the popular advice promises.

So: batching email is a defensible preference with a real finding on the perceived-productivity side and no reliable finding on the stress side. That's roughly the status I'd give most of my own working habits, and I'd rather say so than dress a preference up as a protocol — which is why I describe my own arrangement in those terms and not as a stress technique. Email is the only way to reach me — no comment section, no chat, no form that pages me. I answer it myself, and it takes a few days. The few days is not an apology; it's the mechanism. An inbox answered within minutes is an inbox where anyone can insert a task at the front of your queue, and the delay is what stops that. I like quiet, so email is the only way to reach me. That's a policy about who gets to schedule my attention, not a claim about my cortisol.

Paul Graham was right, and the part everyone remembers is the wrong part

The essay this one is arguing with is "Maker's Schedule, Manager's Schedule," from July 2009. If you've read anything on this subject you've read it, or read someone quoting it. Makers, Graham says, "generally prefer to use time in units of half a day at least," and you can't write or program well in one-hour blocks. The manager's schedule, by contrast, is organised in one-hour blocks that any meeting can drop into. And the line that earned the essay its permanent place: "A single meeting can blow a whole afternoon, by breaking it into two pieces each too small to do anything hard in."

I agree with all of it. What I notice is that the essay is almost never quoted on the question of *which* half-day, and that Graham's own answer was not the morning. He programmed at night, because at night nobody could interrupt him, and when he had to meet founders he put the meetings at the end of his working day so they wouldn't break anything.

So the argument has two parts: you need one contiguous block, and you need somewhere to put the fragments. The second one gets ignored. The fragments don't disappear because you've declared your mornings sacred. The calls still have to happen, the people waiting on you are still waiting, and if you push all of it behind your protected block you've only moved the fragmentation to the afternoon and bought a day where you're never fully available and never fully undisturbed either.

Giving the morning away is a decision about where to put the mess. It goes in a part of the day that is already going to be interrupted — if someone needs me, that's when they get me, and they get me properly rather than getting a reluctant twenty minutes carved out of something else. And then the afternoon survives in one piece, which is the only thing Graham's argument actually requires.

There's a cheaper benefit too, which the Leroy and Glomb result predicts. If the morning is the part that belongs to other people, an interruption during it isn't an interruption. Nothing is being left unfinished, because handling other people *is* the task. The residue comes from switching away from something you're not done with, under the expectation of being rushed when you return. Put the switching in a block where there's nothing to switch away from and you've removed the mechanism rather than defended against it.

Why interruption costs me more than the minutes

I care about this more than a calendar would suggest because of the way I keep my machine. It never sleeps and never reboots unless I force it. That's deliberate: there's a desk full of terminals, editors, background services and long-running SSH sessions, and that standing arrangement *is* my working state. Not windows I happen to have open — a context I built and intend to keep.

When I wrote about Claude Desktop killing itself on Windows, the conclusion I kept returning to was that rebooting your way through a recurring bug was never a real fix; it's just deferred cost. What made that bug intolerable rather than merely annoying was that the recommended workaround charged me the one thing I wouldn't pay: tear the standing context down, then rebuild it by hand.

A day chopped into hour-long blocks charges the same fee, except it charges it six times and nobody writes it down. The cost of an interruption isn't the twenty minutes of the call. It's rebuilding afterwards a state I had deliberately left standing — which files were open and why, what I'd just ruled out, what I was about to try next.

Rands puts the subjective side of it better than I can. Writing about his cave, he calls the Zone "a deeply creative space where inspiration is built," and gives the tell that he's been in it: "it's been four hours since I closed the door." Four hours is not a number you reach by defending the nine-to-eleven slot. It's a shape of day, and you either have the shape or you don't.

The counterfeit is what Cal Newport calls pseudo-productivity: visible activity standing in for useful work, because knowledge work has no clean unit of output and watching people look busy is the cheapest substitute. Answering everything fast makes you look productive whether or not anything was produced. His *Slow Productivity* (2024) sets three principles against it — do fewer things, work at a natural pace, obsess over quality — and the first is the one that costs something, because doing fewer things means telling someone no.

For the avoidance of a misreading: we're running three things at once — the export side of lab-grown gemstones plus our own storefront, the iHomePage CMS, and a B2B knowledge base on LLMs and RAG. Not a boast. It rules out the version of this essay where a man with empty days arranges them pleasantly. There's no shortage of things that could fill an afternoon. The binding constraint isn't hours, it's attention, and the arrangement exists because those two aren't interchangeable and get budgeted as though they were. The same goes for output: I write roughly an essay every other week, and only once I'm sure I mean it. Publishing on readiness rather than on schedule is the same decision as protecting the afternoon — it costs regularity, which is apparently what you're supposed to optimise for, and buys the only thing I want from writing.

The strongest objection: this is a thing you can only do if you own the company

I want to deal with this properly rather than put it in a caveat at the end, because it's correct.

Protecting an afternoon is an owner's privilege. If you have a manager, the meeting goes where the manager puts it. If you're on a support rota, the pager doesn't care. If your clients are eight hours away, your quiet hours are their business hours. If you do the school run, the morning is already allocated. Advice that skips past this isn't advice, it's a description of someone's working conditions sold as a method.

The Hacker News threads on this are much clearer-eyed than the blog posts. In the 2024 "Ask HN: What's your morning routine?" thread the parents just state the constraint and move on — one commenter notes that on a school day he's up at seven for the drop-off, which means in bed before midnight, and that on days he isn't, he's up until one and wakes at eight thirty. That's the entire mechanism of his chronotype, and none of it is genetic. In a thread on *Maker's Schedule, Manager's Schedule*, another said he'd happily err on the side of colleagues with kids: he doesn't have them, and if he occasionally takes a 10 a.m. meeting, that's fair.

The forced-schedule reports are the strongest counterweight to the "just become a morning person" line. One commenter pushed onto an early schedule for a season each summer reported it "didn't reveal an untapped reserve of morning energy. It didn't shift my late night energy burst into the morning. In fact, it didn't shift the productivity burst at all — it eliminated it." Another, after fifteen years of attempts — early classes, cooling the bedroom, a puppy that woke him up — said every method "regressed to night-owl schedule within 2 months."

Then the inverse case, which is the honest reason the morning advice exists at all. In the 2012 thread on waking at 5 a.m. to code, a commenter explained that given the meetings and interruptions in his day the early block was the only way to be sure anything got done, so he did the real work then and answered email later. For someone whose afternoons aren't theirs to protect, the morning is the only contiguous block available, and taking it is exactly right. Same structural answer as mine, opposite schedule. That's the point.

So the transferable part isn't the timetable. It's three questions. Which block of your day is actually contiguous, or could be. Who absorbs the fragments, since someone will. And what you're paying for the appearance of availability — because that's the bill nobody itemises.

What it costs me

My side of the ledger, since I think stating a cost is the minimum requirement for writing about your own habits.

One day in five, the arrangement doesn't hold. If you email me, it will be days before you hear back, and some things are worth less when the answer is late. Being reachable only in the morning means I'm genuinely unreachable for a good part of the working day, and the people for whom that's inconvenient are paying for my quiet, not me. I don't have a defence for that beyond saying it out loud.

What I'd say in return is that the alternative isn't free either. It's just that its cost shows up somewhere that's harder to see — in work that's slightly worse than it should have been, in a day that felt full and moved nothing, in a context rebuilt six times instead of inhabited once.

If you want to try any of this

  1. Pick the block before you pick the hour. You need one contiguous stretch long enough to do something hard. Whether it starts at six in the morning or two in the afternoon is the least important decision in this piece.
  2. Decide where the fragments go, explicitly. Calls, questions and other people's emergencies don't vanish because you've blocked out time. Give them a home, and make it a part of the day you'd otherwise lose to interruption anyway.
  3. Write the Ready-to-Resume note. Thirty seconds on where you stopped and what's next, before you switch away. It's the only intervention here with evidence behind it, and it's free.
  4. Don't batch email because it will lower your stress. On the current evidence that part isn't supported. Batch it because you'd rather decide when your queue gets reordered.
  5. Be honest about which part is a privilege. If you can't move the meetings, almost none of this applies, and the right response is to notice that your constraint is organisational rather than personal.
  6. Count the real number and say it. Four days in five, not five. A rule you keep most of the time is still worth having, and claiming more than you manage is how this entire genre went wrong.

After thirty years of this, the thing I'm most confident about is also the least marketable: there's no hour of the day that does the work for you. What there is, is a shape — one block you can live inside, and somewhere to put everything that would otherwise break it. I pay for mine with the mornings, and on the days it holds, I'd make that trade again.