DEEP READING · READ 950
The Art of Stress-Free Productivity · David Allen · 2001 (revised 2015)
The reason you work late and still feel anxious is not that you have too much to do. It is that you have put "commitments that need to be remembered" and "problems that need to be thought about" into the same container — your head — and your head is bad at both jobs: it fails to hold them, and it hands them back to you at random moments, in the shower, in a meeting, at three in the morning. Allen's answer is not to try harder. It is to empty every unfinished commitment out of your head into an external system you actually trust, and translate each one into a specific physical next action. What the emptying frees up is the thing your brain is genuinely good at: thinking.
David Allen is an American executive coach who spent the 1980s and 1990s doing one-on-one workflow consulting for senior managers. Getting Things Done is twenty years of that fieldwork written up as a method, published in 2001 and substantially revised in 2015. It is the most influential book in the personal-productivity genre and the only one to have spawned an entire software ecosystem — OmniFocus, Things, Todoist and a thousand Notion templates are all copying its skeleton. Its division of labour with books like The 7 Habits of Highly Effective People is clean: Covey argues about which direction to walk; Allen only cares about stopping your feet from slipping.
Allen calls anything you have promised yourself or someone else and not finished an open loop — a circuit opened and not yet closed. The category is far larger than you think: the bulb that needs changing, the message you owe a reply to, the vague sense that you should talk to your parents about insurance, the book you bought three years ago and never opened. It does not have to be important. Being unresolved is enough to cost you.
Why does it cost you? Because you have given it no place to live, so it lives in your head — and your head's way of holding something is to keep bringing it up. Worse, the timing is scrambled: it will not remind you of the bulb when you pass the hardware store, it will remind you as you lie down to sleep. Your brain is simultaneously a very small storage room and a broken alarm clock.
There is a stranger property that explains everyday experience even better: the mind barely ranks open loops by importance. Allen's observation is that a dead light bulb and a pending merger occupy the same order of magnitude of mental bandwidth along the one dimension that matters — not yet arranged. It sounds absurd until you remember the week you spent low-level irritated over something genuinely trivial. What consumes bandwidth is not the weight of the thing but the state of being unresolved. Which yields a useful corollary: clearing out a pile of small open loops often produces more relief than advancing one large project, and that is not an illusion — real capacity is released.
This is one of the few places where the book has decent empirical backing. Psychology has an old phenomenon, the Zeigarnik effect: unfinished tasks are remembered better and intrude more readily on consciousness (the effect's own replication record is mixed, so treat it as suggestive rather than law). But a set of experiments by E. J. Masicampo and Roy Baumeister in 2011 produced a sharper result that fits Allen almost exactly: having people make a specific plan for an unfulfilled goal significantly reduced that goal's intrusive thoughts and its interference with attention — even though not a single step of the plan had been carried out. In other words, the mind is not asking you to finish; it is asking for a trustworthy arrangement. That is essentially GTD's spec sheet.
For the target state Allen borrows a phrase from martial arts: mind like water. The metaphor only helps if you unpack it. Throw a pebble into still water and the response is exactly proportionate — the ripple matches the pebble, no more and no less; then the water returns completely to stillness rather than continuing to slosh in memory of the pebble. Mapped back: over-reaction (one email ruins a day) and under-reaction (something important lands and you feel nothing) are both signals of system failure — and "it's over and it's still churning in me" is the commonest failure of all. Note that this is not detachment: water does respond. It just returns to zero afterwards.
If you take one concept away, take this one. Most people's to-do lists do not contain actions; they contain topics: "Mum's birthday," "taxes," "new website," "physical." Each of those items fails every time you look at it, because you cannot execute a topic. You can only stare at it, feel a wash of vague pressure, and skip to answering a message instead.
The definition of a next action is strict: the next visible, physical thing you would do, specific enough that you could start it the moment you were in position. Not "contact the accountant" but "text Sara and ask which documents she needs." Not "look into the new website" but "open three competitor homepages in the browser and write down the three things I like about them."
Why does this work? Because it moves the cost of thinking from the moment of execution to the moment of planning. Inside "Mum's birthday" sits a chain of unmade decisions: are we doing anything, where, who's invited, who books it. Seeing that item on a tired Wednesday evening means making the whole chain right now before you can move, so your brain quite reasonably declines. What you experience as procrastination is very often unmade decisions charging interest. Whereas "text Lee and ask whether that Sichuan place is any good" can be done on a tired Wednesday evening.
How it changes the way you see things: next time you notice yourself avoiding something, your first thought is no longer "I lack discipline" but "the next action on this hasn't been defined." That is a shift in attribution — from a character problem to a system problem, and system problems can be fixed. Allen adds a corollary: anything requiring more than one step is a project, even at two steps (renewing a passport is a project). Projects cannot be executed, so every active project must carry at least one next action, or it is playing dead on your list.
Run it once and the gap becomes obvious. The list said "renew passport" — one thing, apparently, but actually a stack of undecided questions: do I need new photos, where is the old book, is there an appointment system, which day do I go. Broken out, the project is "renew passport," and the only thing that belongs on the list right now is "look up the appointment URL for the passport office." The remaining steps need no thought yet, because once that one is done the next becomes obvious by itself. This is GTD's most practical departure from conventional planning: it never asks you to map the whole path at once, only to keep exactly one step clear at all times.
Step one is capture: pour every open loop in your head — no sorting, no judging, no prioritizing — into an inbox. Allen asks for a thorough mind sweep; most people produce one to two hundred items the first time and lose a few hours to it, and almost everyone feels lighter the moment they finish, despite having completed nothing. The reason is in the previous section: the relief comes from "it has a place," not from "it is done."
Capture has two hard requirements, and dropping either collapses everything downstream. First, it has to be total. If there is one thing you know is not in the system, your mind goes on running a background backup — and the backup is full-volume: it will not confine itself to the one missing item, it resumes carrying all of them. Second, there must be very few inboxes and each must be within reach: one on the phone, one paper tray, not seven. Too many entry points is the same as none.
Which produces the most easily skipped sentence in the book: the operative word in "trusted system" is not system but trusted. The lists have no power in themselves. Your mind stops polling in the background only after repeated confirmation that whatever goes in does resurface when it should. This explains a familiar pattern: people build a beautiful setup and abandon it in a fortnight — because it was never once verified, so the mind never handed over control. And the only mechanism of verification is the weekly review, below.
With a full inbox comes clarify: process items one at a time, asking only two questions of each.
One: is it actionable? Non-actionable items have exactly three destinations — trash; someday/maybe (want to, not now: learn Spanish); reference (nothing to do, but you may need to look it up: the insurance PDF). Two: if it is actionable, what is the desired outcome (is this a project), and what is the next action? Once defined, the next action has three possible fates: if it takes under two minutes, do it now; if it belongs to someone else, delegate it and log it on a waiting for list; otherwise defer it — onto the calendar if it has a fixed date and time, onto a next-actions list if not.
| Verdict | Destination | Example |
| Not actionable | trash / someday-maybe / reference | old flyer / learn Spanish / policy PDF |
| Actionable · under 2 min | do it now | send the one-line confirmation |
| Actionable · someone else's | delegate + log on "waiting for" | ask finance for last month's report |
| Actionable · fixed time | calendar | Thursday 10:00, medical |
| Actionable · everything else | next-actions list, grouped by context | text Sara about the documents |
The operational core of the book is this table: every input has exactly five legal destinations, and "leave it for now" is not one of them.
The two-minute rule deserves its own note, because its justification is an explicit piece of arithmetic: writing something down, re-reading it later, understanding it again, deciding when to do it and finally doing it costs roughly two minutes of overhead by itself; so for anything under two minutes, tracking it is more expensive than doing it. Note that the rule applies during clarifying, not all day long — treating it as licence to pounce on every small thing that appears is the way it is most often misused.
The two clarifying questions have an underrated second use: they can be lifted out of the list and used as general-purpose thinking tools. "What's the desired outcome?" and "What's the next action?" will crack open almost any stalled situation. A meeting that has run forty minutes without a conclusion is usually one where neither question has been asked aloud — everyone is discussing the situation instead of defining the outcome, and nobody can say afterwards who does what next. Forcing those two questions before a meeting breaks up is the cheapest return this book offers a team.
Next-action lists are not sorted by importance. They are grouped by context — the tool, place or person the action requires: @computer, @calls, @errands, @agenda-for-boss. The reasoning is plain: at the moment of acting, your first constraint is not importance but feasibility. You cannot reach the company intranet from the subway, however important that item is; and when you happen to be outside the supermarket, the @errands list hands you five things at once. Grouping by context is how the list adapts to the reality you are physically standing in.
The second rule is stranger: the calendar holds only things with a fixed date or time — Allen calls it the hard landscape. "Finish the proposal Wednesday" does not belong on the calendar unless it genuinely must happen Wednesday; it belongs on a list. Why so strict? Because once you fill the calendar with soft "planning to do this today" blocks, you spend every day dragging unfinished ones forward, and that act trains your mind to stop believing the calendar — at which point the genuinely hard items (flights, interviews, surgery) lose the force they are supposed to have. A calendar is either true in every entry or worthless as a whole.
Take the example nearly everyone has lived. Monday you block "10:00–12:00 write the proposal"; an urgent meeting takes it, so you push it to Tuesday; Tuesday pushes it to Wednesday; by Thursday that coloured block is as transparent to you as the background. The larger loss is not the proposal. It is that you have quietly learned that things on the calendar can be moved — so the next time something truly immovable appears, it gets a discounted level of respect. That is why Allen guards the calendar so jealously: it is the one component of the system that must carry zero false entries.
This is also the sharpest point of disagreement between GTD and another school. The time blocking that Cal Newport argues for in Deep Work does the opposite: nail deep work into the calendar in advance, or it will be squeezed out by everything else forever. Both are right about something, and the disagreement is really about which bottleneck you have. If your problem is "things slip and my head is a mess," Allen is right. If your problem is "nothing slips, but I haven't had one uninterrupted stretch of thinking all year," Newport is right. The mature practice is to run both: GTD for the full inventory of commitments, time blocking to defend the few stretches that must not be interrupted.
Allen is blunt about it: the weekly review is where the method lives or dies. One to two hours, three jobs — clarify (empty every inbox and every stray note), update (walk the project list and confirm each active project carries a next action; check last week and next week on the calendar; scan "waiting for"), create (read through someday/maybe and see what deserves promoting).
His names for the three are easy to keep: get clear, get current, get creative. The order cannot be reversed either — think about the future before clearing the decks and whatever you come up with is eaten immediately by the backlog underneath.
Why is it the vital organ? Return to the trust mechanism: the mind lets go only on the premise that all this gets reconciled regularly, and the weekly review is the reconciliation. Skip it and within two or three weeks your lists grow a crop of stale items, a crop of zombie projects with no next action, and a crop of things already done but still sitting there. The moment your mind detects that the list is inaccurate, it quietly resumes keeping its own copy — which is the death of your system, usually while you tell yourself you have merely been busy lately.
So GTD failures are almost always the same failure: capture and clarify survived, the review was abandoned. That also gives a useful rule of thumb: if you are only going to run part of this, run "capture plus weekly review" — not "buy a beautiful app and build twenty lists."
A common misreading is that GTD only cares about small stuff. The book does have an explicit vertical component; it is just written up lightly.
The natural planning model has five steps: purpose and principles → outcome vision → brainstorming → organizing → identifying next actions. Allen's argument for it is neat — he says he did not invent this sequence, it is what your mind already does when it decides where to eat: first "something nice, we're celebrating" (purpose), then a picture of "quiet, we can talk, no queue" (vision), then a few restaurant names (brainstorm), then a rough ranking (organize), then "call and book" (next action). Where people fail at work is usually by skipping steps: calling a brainstorming meeting with no purpose established, so the discussion diverges past any hope of convergence; or going straight to the Gantt chart with no vision, so the wrong things get scheduled very tidily. He calls the latter reactive planning.
Above that sit the six horizons of focus, counting up from the ground: current actions → current projects → areas of focus and responsibility (the faces a role obliges you to maintain, like "team health" or "my body") → one-to-two-year goals → three-to-five-year vision → purpose and values. Allen's stance is bottom-up: clear the ground first, because a person with two hundred unresolved small things gnawing at them has no bandwidth for three years out.
The layer worth a dedicated hour is the third — areas of focus. These are not goals; they have no completed state. They are the faces you must keep maintaining as long as you hold the role: as a parent, as the person running a team, as the operator of a body, as the manager of a household's finances. Writing them out one by one pays off very concretely: you immediately notice the boxes that have been empty for a while — nothing under "body" for three months, no action under "that old friendship" in two years. A list will never point this out on its own, because you never entered it as an item. The function of this layer is to force out the projects you owe yourself.
Schematic: Allen argues for climbing this ladder from the bottom — someone gnawed at by two hundred loose ends has no bandwidth for the sixth rung. It is also the book's most contested choice.
Finally, how to choose in the moment. Allen offers four criteria: context (what can be done here at all), time available, energy available, and only then priority. Plus a threefold breakdown for seeing your own day clearly: are you doing predefined work, handling work as it shows up, or defining work (emptying inboxes, deciding next actions)? The third is almost always the neglected one, and it is what determines the quality of the other two.
There is really only one thesis; everything else is its operating detail. Separate remembering from thinking; outsource the first to a system and keep the second for your head — and the unit of outsourcing must be the next action, never the topic.
The chain runs: stress comes from unresolved open loops (with support from psychology — making a specific plan measurably reduces intrusive thoughts) → so capture every loop out of your head → but captured items arrive as topics and are still unexecutable, so clarify each into a next action → organize actions by context and keep the calendar for the hard landscape only → whether any of it works depends on whether your mind trusts it, and trust is renewed by the weekly review → and only once the ground is clear does the talk of higher horizons become real.
It establishes something at the level of workflow, and only at that level. It does not tell you what to do, what is worth doing, or how to resist temptation. It guarantees one thing: nothing you have promised will be lost, and at any moment you know what the next step is. Read as a philosophy of life it will disappoint. Read as infrastructure for holding chaos below a baseline, it is the best thing the genre has.
① Your mind is for having ideas, not holding them. It is a very small storage room and a broken alarm clock — it reminds you only when you cannot act.
② Stress does not scale with how much you have to do but with how much is undecided. Until a thing has a "what it is" and a "what's next," it charges interest.
③ The mind barely ranks open loops by size, which is why a trivial errand can occupy a whole week of low-grade irritation.
④ "Mum's birthday" is not an action; "text Lee about that Sichuan place" is. A next action must be concrete enough to do on a tired Wednesday evening.
⑤ Anything needing more than one step is a project, and projects cannot be executed; every active one must carry a next action or it is playing dead.
⑥ Capture has to be total: if you know one thing is outside the system, your mind resumes backing up everything, not just the missing item.
⑦ The operative word is not "system" but "trusted" — the magic is the letting-go that follows the mind's confirmation that nothing slips.
⑧ The two-minute rule is arithmetic: logging it, re-reading it later and deciding when to do it costs about two minutes of overhead by itself.
⑨ A calendar is either true in every entry or worthless as a whole. Put soft "hoping to today" blocks on it and you train yourself out of believing it.
⑩ GTD fails the same way nearly every time: capture and clarify survived, the weekly review was dropped — the lists drifted out of date, the mind took control back, and you assumed you were merely busy.