When something happened is part of what it meant.

Two people tell you the same sentence: I left my job. In one case it happened the month after a parent died. In the other it happened in an ordinary spring, after eight months of quietly deciding. The sentence is identical. Almost nothing else is.

This is obvious when stated and almost universally discarded in practice. Most systems that hold personal information hold facts — where you work, who your people are, what you like. Facts are tidy, queryable, and stripped of the one property that made them mean anything, which is where they sat in a sequence.

The two clocks again

Start with a distinction that sounds pedantic and is not: there are two different times attached to anything you write down.

There is when the thing happened. And there is when you wrote about it.

These come apart constantly, and the gap is informative in its own right. Writing about an argument the same night is a different act from writing about it three weeks later. The first is closer to the event and further from any perspective. The second has been through something. A record that stamps both with the same timestamp — the moment you typed — has lost the distinction entirely, and with it the ability to notice that you tend to write about your father only after a delay, and never on the day.

The delay is not noise around the signal. Often it is the signal.

Order is not the same as sequence

A chronological list gets you order. It does not get you what surrounded a thing, and surroundings are what do most of the work.

The clearest evidence for this comes from an odd direction. Daniel Kahneman and Donald Redelmeier studied how patients remembered painful medical procedures and found that remembered pain was predicted almost entirely by the worst moment and the final moment — not by duration.¹ A longer procedure with a gentler ending was remembered as better than a shorter one that ended at peak intensity. The experience in total was worse; the memory was better.

What this establishes, generalised, is that a memory of a period is not a summary of that period. It is shaped by where the sharp moments fell within it. Which means that knowing what happened in a season of your life, without knowing where it sat relative to everything else, tells you very little about how that season is stored in you.

Your November is not a list of November's contents. It is what November's worst evening was, and how the month ended.

Time is doing the noticing

There is a category of thing about a life that is invisible at every scale except the long one.

A friendship does not end. It thins. Someone who was in every week of your writing is in one week in four, and then one month in three, and no single step down is large enough to register as an event. There is no moment you could have written about, because nothing happened. The whole change is made of nothing happening, repeatedly, for a year.

The same is true in the other direction. Something you were certain about in March you no longer believe in September, and you did not change your mind on any particular day. You cannot locate the switch because there was no switch.

These are the changes most worth seeing and they are structurally unavailable from inside the present. You cannot introspect your way to them, because introspection samples the current moment and the change is only visible as a difference between moments. The only way to see it is to have kept the earlier moment, dated, and to compare.

Time is not the backdrop to this kind of noticing. It is the instrument.

Why memory drifts toward the wrong dates

There is a further reason to keep dates externally rather than trust recall, which is that dating is the part of memory that fails first and most confidently.

People are generally poor at placing autobiographical events in time, and the errors are not random — they cluster and shift in ways the rememberer cannot feel. Work on the distribution of autobiographical memory across the lifespan has repeatedly found systematic distortions in temporal placement, including the tendency to pull events toward salient landmarks.² Dorthe Berntsen and David Rubin's work on cultural life scripts shows something related: we hold a shared template of when things are supposed to happen in a life, and recall bends toward it.³

So when you say that was about two years ago, you are producing an estimate shaped by how significant the event feels now, not by a stored date. Things that mattered feel recent. Things that resolved feel long ago. The estimate is confident and it is a reconstruction.

A written record with real dates is not a better version of this. It is a different kind of object — one that does not get renegotiated every time you consult it.

What we do with it

Concretely, this means a few things about how the record is kept.

Both clocks are stored. When it happened, and when you wrote it, as separate facts.

Dates carry their own vagueness honestly. Around March stays around March. It does not get quietly resolved into the fourteenth because a precise date is easier to store. A record that invents precision is worse than one that admits it does not know.

Things that stop being true keep their end date. The job you left is not deleted; it becomes the job you had until August. What changed and roughly when is the part that turns a set of facts into a life.

And once a week, the letter can say this is the fourth week running, or not since February — sentences that are only possible because something was counting, and that we would not otherwise be able to give you.

If any of this resonated, you can read more about what a year of writing shows you or why the thoughts you keep coming back to matter most.