← Blog

The Interstitial Journal blog · Built on evidence, not hype

Where Did the Day Go? Why Memory Is a Poor Timesheet

Six o'clock, and the day is gone. You were busy the whole time, and if someone asked where the hours went you could name about three things. That is not a personal failing. Remembered time is built differently from lived time, and the researchers whose job is to measure how people spend their days stopped relying on the remembered version a long time ago.

By comun.ai · 5 October 2026 · ~6 min read

Two clocks: the one you watch and the one you rebuild

Psychologists separate two ways of judging how long something took. A prospective judgment is made when you know in advance you will be asked — in effect, you are keeping an eye on the clock. A retrospective judgment is made afterwards, without warning, from whatever the interval left behind in memory. Almost all end-of-day accounting is the second kind.

Richard Block and Dan Zakay's meta-analysis of 20 experiments found that the two are not the same measurement: prospective judgments came out longer and less variable than retrospective ones, and the pattern was best explained by attention in the first case and by memory in the second.¹ Moderate Remembered duration is a reconstruction, not a reading.

What it is reconstructed from is change. In a later experiment, Zakay and Block had 96 students work through a 12-second run of simple naming tasks that switched task either once or three times. People who knew they would be timing it reproduced the high-switching interval as shorter; people asked only afterwards reproduced it as longer.² Early One lab study of seconds is a clue, not a law about workdays. But it points somewhere familiar: a fragmented afternoon flies while you are in it, then leaves a pile of pieces behind, and at six o'clock the pieces are all you have to rebuild from.

This is the mirror image of the planning fallacy, which is about forecasting how long work will take. That one is a failure to consult the past. This one is a past that was never recorded in a form you could consult.

Why time-use researchers switched to diaries

The people who measure national time use met this problem long ago. A question like "how many hours do you usually work?" asks you to define work consistently, search memory for every episode of it, and add them up — in a few seconds. A time diary asks for something easier: what you did, in order, over one specific day or week.

When John Robinson and colleagues set the two side by side, using the American Time Use Survey and Belgian surveys in which people kept a diary for a full week, employed respondents' estimates of their work hours ran about 5–10 percent higher than the hours in their own diaries, and most of the overestimate came from people who reported long workweeks.³ Moderate The authors' conclusion is the useful part: estimates and diaries are related, but they are not the same measurement.

Diary-methods researchers make the general case — a report made close to the moment captures the particulars of experience in a way a later summary cannot.⁴ Daniel Kahneman and colleagues' Day Reconstruction Method exists for the same reason, walking people back through yesterday episode by episode to reduce recall bias.⁵ We covered it, and why ordinary days compress to nothing, in What to Write When Nothing Happened.

The same gap, when a machine keeps score

Screen time gives the cleanest test, because phones and computers log usage automatically. Douglas Parry and colleagues ran a pre-registered meta-analysis of studies that measured both, pooling 106 effect sizes. Self-reported use correlated only moderately with the logs, self-reports were rarely an accurate reflection of logged use, and measures of "problematic" use tracked the logs even more weakly.⁶ Strong

That evidence is about media use, not your working day. Carrying it over to how you remember a day of meetings, messages and actual work is an inference. Given the duration research above it is a reasonable one, but it remains an inference.

What a timestamped line fixes

If the end-of-day account is a reconstruction, the fix is to stop asking memory to do a ledger's job. A line written at the seam — "10:40 — budget sheet half done, on to the vendor call" — is made while the episode is still present. Nothing has to be rebuilt later; the time is already on it.

In this app that is the default shape of an entry. The timestamp is automatic, the day reads back as a chronological narrative rather than dated pages, and finishing a timed task logs the focused time as an entry you can tag. Read across a week, the lines give you something memory cannot: a sequence, with its gaps visible.

What it can't fix

• It only records what you wrote. A gap between 10:12 and 13:40 is not three empty hours; it is three unrecorded ones. A work journal is a self-selected record, not an automatic time tracker, and the missing stretches are exactly where memory will fill in.
• Recording can change what is recorded. Diary researchers list reactivity among the method's known drawbacks: reporting on your experience can change how you experience it, or what you do.⁴ Your log describes a week in which you were keeping a log.
• It is not a productivity score. Entries per day and focus time are counts, not measures of output or of whether the work mattered. We covered how far a few weeks of entries can be read in What Your Journal Patterns Can and Can't Tell You.

Where the evidence stands

What this means for this app

The calibrated one-sentence version: your end-of-day sense of where the time went is rebuilt from whatever changed, so a line with a timestamp on it, written at the seam, is a better record than the evening's memory — and a partial one.

That makes a daily log a way to see the day you had, not a way to rate it. This is reflection, not treatment. More on our science page, how your notes stay private on our privacy page, and the method itself in the interstitial journaling guide.