Capture one specific detail within a few minutes of hearing it, and set something up that returns it to you before you next see the person — not a file you would have to remember to open.
That second half is the part almost everyone gets wrong. Thirty years of lifelogging research converges on the same conclusion: capture was never the bottleneck, and an archive nobody revisits does nothing at all.
The short version
- Capture within minutes. The detail degrades fast and you will not reconstruct it tonight.
- One concrete fact beats a summary. "Second-round interview Thursday" is usable; "had a nice catch-up" is not.
- Review is where the value is created, not capture.
- Clever triggering did not beat plain regular intervals in the research, so do not over-engineer when things resurface.
- Referencing a specific detail is what makes someone feel understood — which is the operational definition of intimacy in the relationship literature.
Why remembering matters more than it sounds
This can look like a party trick, and it is not. In the relationship literature, feeling understood, validated and cared for is the operational definition of intimacy — and referencing something specific a person told you is one of the few reliable, hard-to-fake ways to produce that feeling.
The strongest evidence for how load-bearing this is comes from an unexpected direction. Adults with adult-onset amnesia had made almost no new close friends in the decades since onset. Remove the ability to carry episodic detail from one conversation to the next, hold most other capacities steady, and new closeness largely stops forming.
That study is a small clinical sample and amnesia disrupts far more than memory, so it is suggestive rather than conclusive. But it points at which link in the chain is doing the work.
The capture rules
Within minutes, not that evening. The specific texture — that she said it like it was nothing, that he was more worried than he let on — is gone within the hour, and it is usually the most useful part.
One fact, concretely. The test is whether it would form a question later. "Interview Thursday" becomes "how did Thursday go?". "We talked about work" becomes nothing.
Include the date if it has one. Half the value of a remembered detail is knowing when to bring it up.
Do not try to capture everything. Wholesale capture produced enormous archives and no compelling applications, which is the single most repeated finding in this area.
The part everyone skips
A note you have to remember to look up has the same failure mode as the thing you were going to remember. If the retrieval depends on you thinking "I should check my notes about Maya", you have moved the problem rather than solved it.
What works is being handed the detail without asking. In the SenseCam research, reviewing partial captures reliably cued recall months later — and end-of-day review clearly beat no review. The value was in the looking back, not the recording.
One more result from that work is worth knowing because it saves effort: sensor-triggered capture performed no better than capture at fixed intervals. Whatever system you build, the sophisticated part is where the effort gets wasted. A dumb regular rhythm is enough.
Doing it without any app at all
A notes file per person works. Name at the top, dated one-line entries underneath, newest first. Read it for thirty seconds before you see them.
A calendar entry works if you put the detail in the entry rather than the person’s name. "Ask Maya about Thursday interview" on the day you are seeing her is the whole system.
What does not work is a general notes app with everything in one stream, because retrieval requires you to search for something you have forgotten exists.
Where this comes from
Every claim above traces to one of these. Each has its own page with the method and what it doesn’t show.
Recording everything produced huge archives and nothing useful
Sellen & Whittaker (2010) · Communications of the ACM
Reviewing a partial capture cues recall months later
Hodges, Berry, Wood et al. · SenseCam autobiographical memory research
People with amnesia make almost no new close friends
Davidson et al. (2012) · Frontiers in Psychology
Feeling understood is the working definition of intimacy
Reis, Clark & Holmes (2004) · Perceived partner responsiveness
Asking a follow-up question makes people like you more
Huang, Yeomans, Brooks & Minson (2017) · Journal of Personality and Social Psychology
Read next
Out of sight, out of mind: why people vanish when you stop seeing them
You are not cold. The intention to reach out arrives at a moment when you cannot act on it, and intentions that cannot be acted on evaporate.
It’s been too long to message them. Send it anyway.
The reluctance you feel is not information about how the message will be received. It is a bias, and it has been measured.
Friendships don’t end. They drift.
Almost nobody loses a friend to a falling-out. The shared context disappears, nobody initiates, and one day it has been two years.
Kept is what we built from this.
Nine seconds of talking after you see someone. Kept files it against the right person and hands it back the day before it matters. No streaks, no relationship scores, and it never writes your messages.
Or see how it compares to the alternatives, including where they beat it.
