Hari's office door has a piece of paper taped to the inside. The paper has five questions written on it.
For Stolen Focus, the questions were on his door for three years. Is our attention really getting worse? If it is, why? Who is doing this to us? How do we fix it? Where in the world is fixing it — I want to go see them.1 He can still recite them years later. He looked at them every time he walked out of the room. He looked at them every time he walked back in. The book that exists is the book those five questions produced.
He's specific about how strict the discipline is. When he found something genuinely interesting during research that didn't answer one of the five questions, the interesting thing didn't make the book. He calls this kill anything that isn't serving the questions you're asking and the story you're telling. He says it's stricter than Fitzgerald's kill your darlings. Fitzgerald told you to kill the beautiful sentences that didn't earn their place. Hari tells you to kill the best insights if they don't fit the questions. What you put into your book is not the most interesting stuff you found out. It's not the most lyrically written stuff that you managed to produce. It's not even the best insights.1
The five-question discipline is what makes the book a journey rather than a collection. Without it, three years of research produces an interesting set of findings. With it, the findings cohere into something a reader can walk through from start to finish.
The list lives on a door, not in a notebook. Hari is operationally specific about this. A list in a notebook gets opened occasionally and forgotten between openings. A list on the back of the door gets seen every time you walk out of the room and every time you walk back in. The forcing function is environmental.
This sounds small. It isn't. Three years is a long time to hold five questions in working attention. The mind drifts. New interesting things appear. The temptation to follow them is constant. The door is what keeps the writer attached to the original questions even when the writing-life is presenting alternatives every week.
If you take Hari's discipline seriously, the door is part of the methodology, not an incidental detail. The same five questions in a Google Doc don't function the same way. The questions have to be where the writer's body keeps encountering them.
Hari doesn't make up the questions to be marketable. He starts every book with something happening in his own life. Addiction in his family. His own depression. His own attention-crisis. The questions emerge from what he actually doesn't know but needs to. The personal stake is what sustains the three-year discipline. Without it, the questions become abstract and the door becomes irritating to look at.
The five questions also form a chain. For Stolen Focus: is the problem real → why → who's responsible → how do we fix → where it's being fixed. Each question depends on the previous one. The chain produces the book's spine. Random questions sitting side-by-side don't make a book; chained questions do.
This page is the input apparatus that feeds Bee-and-Jar Research Method (Hari). The list anchors the chase. Without the door, the bee-chasing wanders. With the door, every bee-chase pulls back to one of the five anchor-points. The two pages work as a pair.
It pairs with Animating Question as Essay Engine (Karlsson) from a different angle. Karlsson says an essay is born from one question that the writer is genuinely held by. Hari says a book needs five chained questions on the door. Same underlying claim — questions are the work's spine — at different scales of the work.
The discipline runs against Karlsson 5-Step Writing Method on one specific axis. Karlsson lets the question emerge over a year of incubation. Hari fixes the questions before the research starts. The choice depends on what kind of work you're doing. Karlsson's emergent-questions serve discovery-essays where the writer doesn't yet know what they'll find. Hari's fixed-questions serve investigation-books where the writer knows the territory and is generating evidence to answer questions they've already articulated.
Imagine you're a year into a book project. You've been doing interviews. You've been reading. You've drifted somewhat — the latest cluster of interviews has been about something tangentially related to your original interest. You haven't articulated the drift to yourself.
You walk back into your office one afternoon. The five questions are on the door. You stop. You read them. You realize the last six interviews don't directly answer any of the five questions. They were interesting. They were producing material. But they weren't producing material for this book.
You sit down. You ask: are the five questions still right, or has my actual investigation revealed a different set? If the five questions are still right, then the last six interviews need to be set aside. They might become a different book or no book at all. If the five questions have actually shifted, you need to revise the door — write the new five questions, replace the old ones. But you need to do that deliberately, not by drifting.
This is what the door produces. Not the prevention of drift — drift is going to happen across three years. The prevention of unconscious drift. You can drift if you choose to, but the door makes the drift visible enough that you have to choose.
You're starting a project that's going to take more than a year. The setup matters.
A blank afternoon. You sit at the kitchen table with paper. Phone in another room. You ask yourself what you actually want to know. Not what you want to argue. Not what you want to teach. What you want to know. You write the questions out, one per line. You sit with them. You revise. You aim for five to seven. Fewer than five and the project is probably too narrow; more than seven and you've smuggled in scope-creep.
The questions should chain. Each one should depend on or extend the previous one. Your chain will be different from Hari's, but it should have the same kind of dependency-structure. Random questions side-by-side don't constitute a book; chained questions do.
You print the list. You tape it to the inside of your office door, or above your desk, or somewhere you'll see it every time you sit down to work. You do not keep it in a Google Doc. You do not keep it on your phone. You keep it in the room with your body.
Three years from now, you'll have walked past it ten thousand times. The questions will be memorized. They'll keep your work attached to itself even when the world keeps offering alternatives.
You'll know the project has drifted when you can't articulate why your most recent chapter belongs in this book. When you find yourself defending the chapter's inclusion with phrases like it adds depth or it shows the texture of the world. Those defenses are the project drifting. The chapter either answers a question on the door or it doesn't. If it doesn't, the chapter belongs in a different book or in nothing at all.
You'll know the discipline is holding when the book converges in the late stages. The chapters start visibly connecting because they were all anchored to the same five questions. You can describe the book's spine in a sentence to a stranger because the spine is what the questions named.
The principle keeps showing up across long-form journalism. Caro's notebook system worked similarly. McPhee has described his version. Patrick Radden Keefe operates with explicit advance-question structures. The journalists are not copying each other; they keep arriving at the same discipline because the discipline is what the form requires.
The open tension: highly-emergent topics may resist pre-specified question-lists. Some real subjects don't fit five chained questions until the research has already started revealing what they actually are. Hari himself says he had some sense of the questions going in but they evolved during research. The discipline isn't to refuse all revision — it's to revise rarely and deliberately. A list revised every month is no list; a list revised once at the year-one mark when something foundational has shifted is still a list.
You spend six months reading a book by an essayist like Karlsson. The pleasure is in not knowing where it's going. Karlsson follows his own attention. You follow his attention with him. The piece arrives at conclusions neither of you could have predicted at page 1. The animating question emerged during writing, not before.
You spend a different six months reading a book by Hari. The pleasure is the journey through specific questions you also wanted answered. You knew before you started that Hari was going to take you through is attention getting worse, why, who's responsible, how to fix it, where it's being fixed. You signed up for that journey. The book delivers it.
Two different reader-experiences, two different writer-processes. Karlsson is right that some work requires the questions to emerge. Hari is right that some work requires the questions to be fixed before research begins. The vault holds both because the same writer working on different forms moves between them. The question discipline depends on what the work is for.
What Hari is doing with the door belongs in two other places worth noticing.
Behavioral Mechanics: Commitment Devices as Attention Architecture — Thaler and the behavioral-economics literature named the mechanism three decades ago. A commitment device is a self-imposed constraint that protects your future-self from the drift of your present-self. The questions on the door are a physical, environmental commitment device. The same architecture that helps people stick to diets and exercise routines helps writers stick to coherent projects. The writer who has internalized question-list discipline is doing the same operation as the saver who automates retirement contributions and the patient who hands their phone to a friend before therapy. The mechanism is environment-over-willpower. The reason all these examples work is the same.
Eastern Spirituality: Sankalpa as Intention Anchor — Yogic practice traditionally opens with the formation of a sankalpa, an explicit intention that anchors the rest of the session. The teacher tells the student to state the sankalpa before beginning practice and to recall it before closing. The recall is structurally the same as Hari's reading-of-the-door — explicit intention as gravitational anchor for the rest of the work, repeated often enough that the body learns it. The yogi and the journalist are running the same operation at different scales of stake. The vault should recognize that question-list-as-attention-architecture is not a craft idiosyncrasy. It's an old discipline appearing in new clothes.
The Sharpest Implication Most ambitious nonfiction projects fail not because of bad research or bad writing but because the writer never did the upfront work of articulating what the book is actually asking. The proposals get accepted, the books get written, and most of them underperform because the missing question-discipline produces incoherent output. The fix is upstream of the writing. Five questions on the door before research begins would dramatically improve the average quality of long-form nonfiction. Most writers don't do it. The reasons are partly cultural (the discipline looks pedestrian, not literary) and partly cognitive (the writer doesn't yet know what they don't know). Hari's career is partial evidence that the discipline pays off. It's also evidence that the discipline can be sustained over decades by writers willing to be slightly less interesting in the short term in exchange for substantially more coherent in the long term.
Generative Questions