You sit down at eight with ninety minutes and every intention of using them.
The first fifteen are bad. You reread what you wrote last time, open two things you don't need, and produce a paragraph you'll delete. Somewhere around twenty past, without any transition you noticed, you're working properly.
That ramp happens to almost everyone and almost nobody plans for it. Which is a shame, because it's the most manipulable part of a session.
The ramp is not wasted time
The first instinct is to treat those fifteen minutes as a failure of discipline and try to eliminate them. That doesn't work, and it isn't the right goal.
Getting into a piece of work involves loading it: remembering where you were, rebuilding the shape of the problem, recovering the decisions you'd made. That's real cognitive work with a real cost, and it happens whether or not you've arranged for it.
What you can change is what you spend it on. Right now most people spend it on the difficult part, badly — producing a paragraph they'll delete, or a decision they'll reverse — and arriving at the good stretch with some of their energy already gone.
The alternative is to spend it deliberately on something that helps.
What good warm-up work looks like
Four properties: related to the task, low-stakes, requires no decisions, and produces something small.
Rereading the last session's output, changing nothing. This is the best one for writing, and it does double duty — it loads the state and it tells you where you are.
Tidying something adjacent. Formatting, filing, sorting out the references. Genuinely useful, requires nothing of you, and puts your hands on the material.
Redoing something easy you've done before. Musicians have known this forever — scales aren't there to develop technique at year twenty, they're there to get you into the instrument. Programmers have the equivalent in a small refactor or an easy bug.
Or writing down what you're about to do. Two minutes, three lines, and it converts an unfocused start into a task.
What none of these are is a different task. Answering email is not warm-up, because it loads the wrong state and leaves residue you'll then carry into the real work.
The obvious risk: warm-up work is comfortable, and comfortable activity adjacent to difficult work is exactly what avoidance looks like.
The distinction is time-boxed versus open-ended. Ten or fifteen minutes of rereading is a ramp. Ninety minutes of rereading is a session where you avoided writing, and it'll look identical from outside.
So set the boundary before you start. Fifteen minutes, then the real thing, whether or not you feel ready. The feeling of readiness is not a reliable signal and waiting for it is how a ramp becomes an evening.
The tell that you've crossed over: three sessions in a row that were all warm-up. One is a slow evening. Three is the pattern, and it means the difficult part has become something you're not approaching.
The other end of the session
There's a matching discipline at the finish, and it's worth more than the warm-up.
Stop before you're empty. Write down where you got to and what's next. That note is what makes tomorrow's ramp shorter — you arrive at a session with the state partly loaded, rather than starting cold every time.
And stop mid-thought where you can. Finishing something tidy makes the next session begin with a decision. Stopping halfway through a paragraph you know how to complete means tomorrow starts with typing, which is a far easier thing to start.
Done together, the closing note and the mid-thought stop can cut the ramp from fifteen minutes to five, which across a hundred sessions is a substantial amount of work recovered from nowhere.
It's worth noticing that both of those are done at the end of a session, when you're tired and want to stop, and that both take under three minutes. Almost everything in this practice that pays well is like that — small, badly timed, and easy to skip.
Where this doesn't apply
Two honest exceptions.
Some work has no ramp at all. Physical training, routine admin, practising something well-established — these start at full speed, and building a warm-up ritual around them adds ceremony without benefit.
And short sessions can't afford one. If you have thirty minutes, spending fifteen on warm-up leaves you very little, and the better use of a short session is contact rather than production — do the small thing, write the line, keep the goal alive. The ramp matters for sessions long enough to have a middle.
There's also individual variation that no general advice covers. Some people genuinely do start cold at full concentration, and some need forty minutes. Worth measuring your own rather than adopting a number from an article.
Ninety minutes, two arrangements
Tuesday evening, half seven to nine, drafting a chapter.
The first arrangement: he opens the file, doesn't know quite where he is, and starts trying to write. Twenty-five minutes produce about two hundred words, most of which go. Around ten past eight he finds the thread and works properly until nine. Useful output: roughly fifty minutes.
The second: he opens the file and reads the last two pages, changing nothing. Ten minutes. Then he writes three lines about what this section needs to do. Two minutes. Then he writes, from about a quarter to eight, and it's decent from the start because he knows where he is.
At nine he stops mid-paragraph and writes one line: stopped halfway through the bit about damp surveys, next is the case study.
Useful output: about seventy minutes, from the same ninety. And Thursday's ramp is five minutes rather than fifteen, because the note tells him where he was.
What the timer is actually for
The session timer here isn't there to measure your discipline or produce a total. It's there to bound the session, and the bounding is what makes both ends work.
A bounded session has a defined start, which makes a deliberate ramp possible rather than accidental. And it has a defined end, which is when it asks you for a line — at the one moment when what happened is still in your head and costs nothing to write.
The ten minutes at the beginning and the ten seconds at the end are doing more for the next session than the middle is.
Related: Ninety minutes is the unit · Attention residue · How your log becomes the thing that restarts you
Share: