A thought you cannot capture in about a second is, for practical purposes, already gone. Not slower to retrieve. Not filed somewhere inconvenient. Gone, with no copy anywhere, and usually you do not even notice the loss, which is the part that should bother you.

I spent years treating my note-taking like a storage problem. I compared apps by how they organized things: folders versus tags, backlinks versus search, local files versus a synced database. Every few months I migrated to whatever promised a tidier shelf. None of it changed the thing that was actually costing me ideas, because the thing that was costing me ideas happened before storage was ever involved. It happened in the gap between having a thought and getting it out of my head, and that gap is measured in seconds, not in retrieval quality.

There is an old experiment that I think about more than is probably healthy. In 1959, Lloyd and Margaret Peterson gave people three-letter strings to remember, then immediately made them count backward by threes so they could not rehearse. After three seconds of that distraction, people still recalled the letters about eighty percent of the time. After eighteen seconds, recall collapsed to around ten percent. Eighteen seconds, and the letters were gone, because attention had been pointed somewhere else for less time than it takes to find the right app and tap into the right note.

That is the real shape of the problem. Human working memory is not a hard drive that holds your thought patiently until you get around to saving it. It is a leaky bucket, and the leak is fast. So the requirement on the moment of capture is harsher than almost any other requirement in a personal system: it has to be fast enough that catching the thought beats losing it, every single time, including on the days you are tired, walking, holding something in your other hand, or being talked at by a stranger.

Here is where most systems quietly fail, and they fail for a reason that looks like a virtue. They make capture happen inside structure. To write the thought down, you first open the app, then you are looking at a hierarchy, and the hierarchy asks questions. Which notebook. Which folder. Which tag. Does this belong with the other thing or is it new. Each question is a small decision, a few hundred milliseconds of deliberation, and a few hundred milliseconds is enough. While you are deciding where the thought goes, the thought is already draining out of the bucket. You end up with an immaculately organized system that is missing exactly the ideas that mattered most, because good ideas tend to arrive at the worst possible moments for filing.

So I started measuring my own setup with a single number, and it is the only note-taking metric I trust now: how long, in seconds, from the instant I have a thought to the instant it is recorded somewhere safe and I can stop holding it in my head. Not how searchable it is later. Not how nicely it is filed. Just the latency of the catch. When I started timing that honestly, my elaborate system was costing me three to five seconds per capture, sometimes more if I hesitated over where something belonged, and I could feel the hesitation as a kind of low background tax on thinking. I had been optimizing the warehouse while the loading dock leaked.

Cutting that number down turned out to be almost entirely a matter of removing decisions, not adding features. The capture surface has to be dumb on purpose. No folder prompt. No tag picker. No "which notebook." A line goes in, it lands somewhere I trust, and I keep walking. Everything that feels like organizing has to be ripped out of the moment of capture and pushed to later, because organizing is a decision and decisions are exactly what the fast layer cannot afford. The discipline is counterintuitive: to keep more of my thinking, I had to let the capture step get noticeably stupider.

I want to be fair to the other side, because there is a real objection here. Capture that records everything and organizes nothing just relocates the problem. A pile of undifferentiated lines is not a second brain; it is a junk drawer, and a junk drawer you cannot search is arguably worse than not capturing at all. That is true, and I am not arguing against organization. I am arguing against doing it at the wrong time. Curation, structure, linking, deleting the two a.m. thought that did not survive contact with morning, all of that is real work and all of it deserves a place. Its place is later, at a desk, where I have minutes and not a second and a half. The mistake is collapsing two jobs with two completely different latency budgets into one screen and one moment.

The reason this matters for shipping, and not just for note-taking, is that the same friction shows up everywhere I work alone. The idea for the fix to a bug, the realization about why a feature is not landing, the one sentence that finally explains the product, none of these arrive on a schedule and most of them arrive when my hands are full. As a solo developer I do not have a meeting where someone writes my ideas on a whiteboard, and I do not have a colleague who remembers the thing I said in the hallway. If I do not catch it in the first second, there is no backstop. The friction in capture becomes, very directly, friction in what I manage to build, because the things I never recorded are things I never act on.

I have stopped being precious about the storage end as a result. Plain lines, timestamped, in a format I will still be able to read in ten years, is enough for the bottom of the system, and I would rather have a boring durable store and a fast front door than a beautiful database with a slow one. If I have to choose between a system that captures in one second and organizes clumsily, and a system that organizes brilliantly and captures in five, I take the first one without thinking, because the first one keeps the raw material and the second one loses it. You can always build a better reader on top of lines you actually kept. You can never retroactively capture the thought you let drain away while deciding where to put it.

None of this is a productivity hack, and I am suspicious of the genre. It is closer to an accounting correction. For years I was counting the wrong cost. I measured my note system by how it stored and retrieved, when the dominant cost was sitting upstream of both, in the half-second of friction I had never thought to measure. Once I started measuring that number and protecting it, the rest of the system got simpler, not more complex, because most of what I had been adding was structure that the fast layer was paying for and the slow layer should have owned.

So the whole claim, restated as plainly as I can: a thought you cannot capture in about a second is already gone, the cost of that loss is total and silent, and the only way to keep more of your own thinking is to make the moment of capture too dumb to ask you any questions. Optimize the second. The shelves can wait.

What I cannot see is your version of this. When a thought hits you in the middle of something else, what is your real capture latency, one second or ten or never, and do you even notice the ones that get away?

I build Simple Memo on my own: a one-tap iOS app that gets a thought out of my head and into my email in about a second, before I can lose it. I write here every few days about the parts of building solo I had to get wrong first.