Time Blindness: The Hour That Turned Out to Be Four
What it actually looks like
You said an hour. You are four hours in and it is not finished.
The overrun is not the interesting part; anything can run long. The interesting part is that you were not hedging. You believed it — and tomorrow, asked about a task the same size, you will say an hour again.
It is 6:10 in the evening. The last moment you can place with confidence is opening the file after lunch. In between you did real work, but the afternoon has no shape in your memory. It did not feel like five hours. It felt like a stretch of indeterminate length that ended when you looked up.
This week's calendar was filled in a fortnight ago by a version of you who felt generous. Three calls, a deliverable and a quick review, all landing Thursday. That person is not available for comment. You are the one paying what they promised.
When you sized the work, you sized the part you can picture. The code, the draft, the deck. Not the review round, the migration, or the half day lost to the dependency that behaves differently in production. None of it is a surprise; it happens every time. It is not in the picture your mind produces when somebody asks how long.
And the number left your mouth before you had thought about it — a question on a call, a pause of about a second and a half, an answer. You double the figure now, quietly, before you say it out loud. You still run over.
Who this happens to
This is common among people who can disappear into work.
The ability to drop into a problem and stay there is a real advantage, and it is the same capacity that removes your access to the clock. The sense of elapsed time is built out of noticing — changes of task, changes of light, small breaks in attention. Absorption removes every one of those markers.
It is common in founders who have never billed by the hour. Consultants get corrected constantly: someone writes down what was spent, a client argues about it, and within two years they know what a discovery call costs. You invoice outcomes. No one has handed you a timesheet, so nothing has contradicted your sense of duration. Twelve years in, your estimates are still a beginner's.
It happens to optimistic planners, whose default relationship to a new project is appetite — a real asset in a business that asks you to believe before there is evidence. Optimism does not switch itself off in the presence of a calendar.
And it is common in minds that run fast: quick, associative, several threads open at once. For many minds the felt sense of duration runs differently — an hour of something absorbing and an hour of something dull do not report the same length, and neither reports the true one. That is not a flaw in your character. It is a gauge that was never calibrated, read by someone who assumed it had been.
What sets it off
Being asked out loud, live, with a person waiting. The figure you give in conversation is not a calculation but a reflex with a number attached. The silence you would need in order to think is longer than the conversation tolerates, so you fill it.
Planning more than a week out sets it off. Your future self is a stranger with an empty calendar, no outage and no bad day. You are not scheduling into a real week but an idealised one.
"This is basically the same as the last one" sets it off, because it usually is not, and because your memory of the last one has been edited down to its successful pass.
Long absorbing stretches with nothing external in them — no meeting, no meal, no change of room — set it off hardest. Those are the days that end at 6:10 with a hole in them.
And one quiet trigger: being asked by someone you want to impress. The number that comes out then is not a prediction. It is an offer.
Why it keeps happening
Your estimate is not built from data. It is built from a memory, and memory compresses.
What you kept from the last job of this kind is the shape of it working: the approach that succeeded, the hour it came together. What was discarded is the two false starts, the afternoon spent waiting on someone, the version you threw away. You estimate from a highlight reel and measure the result against a full recording. The gap is arithmetic, not failure.
Kahneman and Tversky named this the planning fallacy: people systematically underestimate how long their own tasks will take, while often judging other people's accurately. The bias is not ignorance, which is why knowing about it does not remove it. It comes from estimating while standing inside the plan, so the correction has to arrive from outside it.
Then there is the reason it never improves. You have never written down what anything actually took, so nothing contradicts you and every estimate is a first estimate — twenty years of unmarked guesses is one guess, repeated. Experience teaches only where there is feedback, and duration is the variable nobody scores.
Underneath it sits a belief worth saying plainly: I should be able to tell. That sentence turns an uncalibrated instrument into a character problem, and the remedy for a character problem is effort — care more, focus harder, be realistic this time. Effort does nothing to a gauge. You cannot concentrate your way into a reading.
So ask the honest version. Name one project in the past year that took as long as you said it would. Not roughly — as long. Most founders, pressed, cannot name one. That is not carelessness but a bias with a consistent direction, and a bias with a consistent direction is correctable in exactly one way.
What actually helps
Log actual durations for one week before you trust another estimate. Start time, stop time, one line of what it was. Not a tool with a dashboard — a note file or a paper page. The point is not discipline; it is a record. A week gives you four or five real numbers, and those beat every instinct you hold about your pace.
Then quote from the record, not from the feeling. "The last one like this took nine hours" is a sentence you can say out loud in a meeting. "It feels like about a day" is the sentence that has already cost you several weekends. When the two disagree, the record wins: it was written at the time, and the feeling was assembled afterwards.
Buy yourself the gap between being asked and answering. Let me check and come back to you before the end of the day. Two seconds of mild discomfort for a number that is looked up rather than produced. Almost nobody who asks needs it inside the conversation; they ask there because that is when it occurred to them.
Make time visible instead of felt. A clock in your line of sight. A timer in the corner of the screen — not to alarm you, to show you. Absorption deletes the internal signal; sight restores an external one, on precisely the days that currently vanish.
At day's end, set the estimate and the actual side by side. One line, thirty seconds. The gap is the only thing that recalibrates anything, and after two weeks you will find it is roughly the same multiple every time. That multiple is the discovery: it turns a private embarrassment into a correction factor you can apply to the next thing you are asked.