The setup

Picked database indexing, the thing I know best, deliberately, to give the test every chance to confirm the fear: that I don’t actually understand things, I just reproduce them.

What happened

The explanation came out complete. Problem, mechanism, tradeoff, in order, no gaps. Then, immediately after finishing it, I predicted I’d already lost whoever was listening. That prediction wasn’t supported by anything in the explanation itself. It came after a correct answer, not during a struggling one.

What the result actually says

So the original hypothesis, a retrieval gap, doesn’t hold, not for this. What’s better supported is narrower and more interesting: knowing the material was never the problem. Trusting a finished answer to actually be finished, and reading a room well enough to know how much of it to say, that’s the real, untested skill. Different problem. Different fix.

What I took from it

  1. The test disproved my own hypothesis. That’s the test working, not failing.
  2. The doubt arrived after a complete answer, so it wasn’t about the material.
  3. The skill to practice is trusting a finished answer and judging how much of it to say.