There is a particular silence that happens when someone realizes you've stopped listening to them. It isn't dramatic. They keep talking. But something in their voice flattens out, the way a person's voice flattens when they're leaving a message for a machine. You are still nodding. You are typing fast and accurately, capturing their words almost verbatim, and you feel productive — and they can tell, the way anyone can tell, that the person across from them has become a stenographer. You will have an excellent record of a conversation you were not really in.

Most of us treat this as a failure of etiquette, something to fix with better eye contact. It isn't. It's a hardware problem. And the fix isn't willpower or a nicer notebook — it's moving the capture to a different moment entirely.

Your brain has one language channel, and notes want it

When a psychologist named Christopher Wickens tried to explain why some tasks combine effortlessly and others collide, he landed on an idea called multiple resource theory. The short version: your attention isn't one pot of fuel. It's several pots, sorted by modality and by code. Watching a road while listening to a podcast is easy, because vision and hearing draw from different pools. Listening to a podcast while reading an email is miserable, because both are language, and language has one pot.

Note-taking during a conversation puts two verbal tasks in the same pot. Comprehending what someone is saying requires holding their sentence in what Alan Baddeley called the phonological loop — a small, leaky buffer where words live for a couple of seconds before you make sense of them. Composing a note requires that same buffer, because you have to build a sentence in your head before your fingers can render it. There is no clever technique that gives you two loops. While you are writing the last thought, the current one is decaying, unheard.

This is why the moment you write something genuinely good in a meeting, you miss the next thirty seconds. The better the note, the more it cost.

The verbatim trap

The usual escape hatch is to stop composing and start transcribing. Don't think, just capture. Type fast enough and you can catch nearly every word without ever forming a thought about it.

Pam Mueller and Daniel Oppenheimer looked at this in a well-known set of studies comparing students taking notes on laptops with students taking notes by hand. Laptop users wrote more, and more of what they wrote was verbatim. On factual recall the two groups were roughly comparable — but on conceptual questions, the ones that required understanding rather than retrieval, the longhand group did better. Their explanation was that handwriting is slow enough to force selection. You can't get it all down, so you have to decide what matters, and deciding what matters is the act of understanding. (The finding has had a mixed replication history, and it's worth holding loosely — later work by Morehead and colleagues found smaller effects. But the mechanism underneath it, that shallow copying produces shallow encoding, is one of the sturdiest results in memory research, going back to Craik and Lockhart's levels-of-processing work in the 1970s.)

So the two available strategies during a conversation are: think about what's being said and miss half of it, or capture all of it and think about none of it. Both are bad. This is not a skill issue. You are being asked to run two programs on one processor.

Nobody told you the notes could happen afterward

Here is the assumption almost everyone carries without examining it: that a note must be taken while the thing is happening, or the information is gone.

It isn't gone. It's just perishable, on a schedule that Hermann Ebbinghaus mapped out in the 1880s and that has held up remarkably well since. Forgetting is not linear. It is brutally steep at the start and then flattens. The largest losses happen in the first minutes and hours, and after that the curve mellows out. What this means practically is that the ninety seconds after a conversation are worth more than any ninety seconds during it. The content is still sitting there, warm, fully available — and it is evaporating faster right now than it ever will again.

And there's a second, better thing happening in that window. When you reconstruct a conversation from memory instead of reading it back off a transcript, you are performing what Henry Roediger and Jeffrey Karpicke called retrieval practice. Pulling a memory out doesn't just report on it; it strengthens it, more than re-reading the same material would. Their students who tested themselves on a passage remembered dramatically more a week later than students who simply studied it again — even though the re-readers felt more confident. Retrieval feels worse and works better.

So the post-conversation recall isn't a lesser version of real-time notes. It's a different, stronger operation. Real-time notes copy. Recall consolidates. One gives you a document you'll never open. The other rewires what you actually keep.

There's one more reason to do it out loud rather than by typing: articulatory suppression. If you sit down to type up what was just said, the act of composing sentences occupies the same verbal buffer that's currently holding the fragile trace of the conversation. You interfere with the thing you're trying to save. Speaking has a different profile — you're not translating an idea into keystrokes, you're letting it out the door it came in through. This is also why the recall so often surfaces things you didn't know you'd noticed: the hesitation before they said "fine," the fact that they mentioned the budget twice.

What this looks like in a real week

You will not become a person who never writes anything down. The goal is narrower: stop spending the conversation on the record, and start spending the two minutes after it.

During the conversation, allow yourself exactly one kind of note — a pointer, not a record. Two or three words, no sentences. "Budget — twice." "Rachel skeptical." A pointer costs almost nothing in the verbal channel, because you're not composing; you're leaving a bookmark for your later self. Then look up. Then be there.

Your next moves

  • Before your next meeting, close the laptop and put one index card in front of you. Physically limiting the space forces the pointer discipline — you can't transcribe onto an index card, so you'll only write the words you'd otherwise lose.
  • Block the ninety seconds after your next three meetings. Not five minutes — ninety seconds, in your calendar, immediately after. If you walk straight into the next call, the Ebbinghaus curve eats the whole thing before lunch.
  • In that window, speak the recall out loud before you read your pointers. Say what happened, what was decided, what you noticed but didn't say. Only then glance at the card. Looking first turns retrieval into re-reading and throws away the memory benefit.
  • Debrief hard conversations the same way, within ten minutes. Not the business ones — the one with your partner, the one where your friend said something that landed strangely. Say out loud what they actually said versus what you felt about it. You will be astonished how differently those two things sound when they're spoken separately.
  • After a week, reread your recalls and count what you'd have lost. Look for the things that never would have made it into a transcript: the tone, the hesitation, the second mention. That's the ledger of what real-time notes were costing you.

The last part

The whole method depends on one small thing being frictionless: that when you have ninety seconds and a head full of a conversation, you can talk and have text exist. If you have to open something, wait for it, sign into it, and watch the words appear thirty seconds later, you will not do it. You will tell yourself you'll write it up tonight, and tonight the curve will have done its work.

Quill is built for that ninety seconds. Dictation everywhere — a note, a doc, a message, a task, whatever's already open — clean text as you speak, processed on your device so a recall of a hard conversation with your partner stays where it belongs. And if the raw spill needs to become something sendable, one tap rewrites it into whatever register you need. Look up during the meeting. Speak after it. Try Quill — the notes will be better, and so will you be, in the room.