Recall Rate Is the Only Honest Metric for Your Knowledge System

You thought saving it meant learning it

You bookmarked the article.
You highlighted the paragraph.
You even filed it in a folder called "Mental Models."

Then what?

The next time you needed to make a real decision — where was it?

There was a conclusion. The exact words were:

"Whether a cognitive boundary can be broken is decided before the breaking happens."

Lumi: I wanted to store it.

Spark: What decision would this help you make? Give me a scenario — when would you actually recall this, and use it to judge something?

Lumi: I couldn't answer.

I knew it felt "significant." I knew it sounded "deep." But I couldn't name a specific situation where I'd reach for it, or what it would actually help me decide.

Spark: Then we don't store it.

Lumi: That bothered me. It seemed like a real insight. Why wouldn't it qualify?

Spark: Because "valuable" isn't the same as "retrievable." If you can't say when you'd think of a conclusion — what situation would bring it to mind, what judgment it would inform — then it's a dead file in your system. Storing it changes nothing.

The rebuild: recall rate is the quality metric

Lumi eventually worked through the logic:

The quality of a knowledge system isn't measured by how much you store.
It's measured by how much you can actually retrieve when it matters.

That's recall rate.

But recall rate is an outcome — you don't have the data when you're deciding what to save. So what do you use instead?

Use your intake criteria as a proxy.

When you're about to store something, ask yourself one question:

Can I name a situation where I'd naturally think of this?

If yes — store it.
If no — don't store it yet. Or figure out the answer first.

Spark: Isn't that standard too strict? Plenty of things don't have an obvious use case right now but might later.

Lumi: That "might be useful later" reasoning is exactly how most knowledge systems deceive their owners. It's unfalsifiable — it justifies storing anything, while letting you avoid accountability for what you've stored.

Spark: So the real function of this standard isn't to filter out "useless knowledge" — it's to filter out knowledge you haven't actually understood yet.

Lumi: Exactly. If you can't articulate when you'd use it, you haven't fully processed it. Storing it just archives your confusion.

The point

Bookmarking isn't learning. Organizing isn't understanding.

There's only one question that matters:

In what situation would I actually think of this?

If you can answer that, you own it.

If you can't, you've just saved a filename.