The photo survives. The story doesn’t.

Open your camera roll and scroll back three years. You’ll find a photo you cannot explain. A shoe on a windowsill. A drawing of what might be a horse. Your kid holding something up to the lens, delighted, about something. You took that photo on purpose. The purpose is gone.

This is the quiet failure of the modern childhood record. Most advice about how to remember childhood details starts with taking more photos, and the photos were never the problem. We are the most-photographed generation of parents in history and we are producing archives our own kids won’t be able to read. The image survives perfectly — better than any generation’s before it — and the reason evaporates within months.

Context decays faster than images

Memory researchers call it cue-dependent forgetting: the memory is often still in there, but the retrieval cue is gone. The photo was supposed to be the cue. But a photo of a half-eaten sandwich cues nothing, because the thing that mattered — he would only eat it if the crusts were cut into triangles, specifically triangles, and he checked — was never attached to it.

The picture was never the memory. It was a pointer to the memory, and pointers rot.

The failure hides behind a very reasonable feeling: I’ll remember what this was. And in the moment, you would. The triangle inspection is vivid on Tuesday, still there in October, gone by the time it matters. Vividness now is not a promise of retrieval later — it’s almost the opposite, because the details that feel unforgettable are exactly the ones nobody bothers to write down. Nobody captions the sandwich. Everybody captions the graduation, which is the one photo that never needed a caption.

Paper keepsake books solved this accidentally. The prompt was printed on the page: “First word.” “What made them laugh.” The answer was written under the question, and forty years later both are still there, together. Digital photography threw away the question and kept only the answer, a million times over.

It’s worth sitting with how odd that is. A keepsake book printed in 1986 on acidic paper, filled in with a ballpoint pen, is more legible today than a cloud album assembled in 2021. Not because the paper is a better medium — it’s a much worse one — but because the book stored two things per page and the album stores one. The question turns out to be half the record.

What a prompt actually does

A prompt does two jobs. Before the moment, it directs attention — you notice their hands today because something asked you to. After the moment, it becomes the label the memory is filed under. It’s both the assignment and the index card.

Photo

Photograph their hands. Just their hands.

Twenty years from now, that image with that sentence attached is not a random close-up. It’s an answer to a question, dated, with an age on it. Searchable. “What he was afraid of.” “The word she wouldn’t let us correct.” You can find it, because the finding-word was saved with the thing.

The before-the-moment job is the one nobody prices in. Attention is the scarce input here, not storage. On an ordinary Tuesday you are not going to spontaneously notice the way your kid holds a crayon in a fist like a dagger, because nothing is asking you to. A question asks you to. The entry it produces isn’t extra work on top of your day — it’s ten seconds of your day you would otherwise have spent not noticing.

The after-the-moment job is what makes the archive readable. Consider what you can actually type into a search box in twenty years. Not a filename. Not a date you don’t know. What you have is the shape of a question: what was she afraid of, what did he call the dog, what were they obsessed with the year we moved. If the record was built out of questions, those searches land, because the language you remember in is the language the archive was filed in. If it was built out of timestamps, the same searches return nothing, and the archive is technically complete and practically unreadable.

How to remember childhood details, actually

None of this needs software. If you keep photos anywhere, the practice that changes what your archive is worth in twenty years is small enough to start tonight:

• Write the caption the same day. Two sentences. The reason decays in weeks, not years, so the same-day sentence beats the someday paragraph.
• Keep the question with the answer. If something made you take the photo, write that down, not just what’s in the frame. “He checked the triangles” outlives “sandwich.”
• Date it, and put an age on it. “Age 3” is the sort you’ll actually search by. Doing the arithmetic in 2044 from a filename is how archives go unread.
• Favor the ordinary Tuesday. Birthdays photograph themselves. The details that vanish are the weekday ones — the bedtime negotiation, the word said wrong, the current obsession.
• Don’t trust “I’ll remember what this was.” You are reading an essay that exists because nobody does.

Question

What did they say this week that you want to remember exactly?

Notice what that question does that a camera can’t. Some of the best context never had a visual in the first place — the sentence said from the back seat, the theory about where the moon goes, the ten minutes on the floor being a patient in a stuffed-animal hospital. A record built only out of photographs quietly drops everything that didn’t face a lens. Questions don’t have that bias. They catch whatever actually happened, including the days when what happened was a sentence.

And lower the bar until it can’t intimidate you. This isn’t a journaling practice with a page-a-day quota; it’s a caption habit. A hundred photos with two sentences attached will be worth more in 2044 than ten thousand without — not a little more, categorically more, because the hundred can be read by someone who wasn’t there. That is the eventual audience: someone who wasn’t there, or was there and was three.

A notebook and a list of good questions will do all of this. The 133 questions Vellum asks are published, free, for exactly that use — and if you’d rather the question arrived on its own and filed the answer where you can find it, that’s the whole app.

That’s the whole design argument: keep the question. Everything else — the app, the timeline, the search — is plumbing for that one decision. Ask one good question tonight and write down the answer, and you’ve already built the part of the archive that was actually at risk.