"Explain it simply" is a trap

Ask for a simple explanation and you get an analogy. Analogies feel wonderful. They also stop you from asking the next question, because you now feel like you understand.

"A closure is like a backpack the function carries around."

Lovely. Now: does the backpack hold copies of the items, or the items themselves? If you don't know, you don't understand closures — you understand a backpack.

The ladder

Ask for the same concept at three heights, in one prompt:

Explain [CONCEPT] three times, labelled:

RUNG 1 — to a curious 12-year-old, using one everyday analogy.
RUNG 2 — to a junior developer who knows [YOUR ACTUAL BACKGROUND].
RUNG 3 — precisely, using the real terminology, no analogy at all.

Then tell me the single most important thing Rung 1 gets WRONG.

Why three rungs and not one:

• Rung 1 gives you something to hold onto — a hook for memory.
• Rung 2 is the one you'll actually use when you code.
• Rung 3 is the vocabulary you need to search documentation and talk to colleagues.
• The final question is where the learning happens.

The precise version feels the least useful because it's the hardest to read. It's also the only rung that lets you search documentation, read a Stack Overflow answer, or describe your problem to a colleague. Skipping it is how people stay dependent on asking an AI to translate for them.

The killer follow-up

That last line is the whole technique. Variants worth memorising:

Where does that analogy break down?

What does that analogy make me likely to get wrong?

Give me a case where the intuition from Rung 1 predicts the wrong answer.

Recalibrating mid-conversation

Don't restart. Steer.

• Too advanced → Back up. Assume I don't know what [term] means.
• Too basic → I already know that part. Skip to [the bit I asked about].
• Too long → Same answer, one third the length, keep the example.
• Lost in jargon → Rewrite that using no term you haven't defined for me.
• Suspiciously smooth → What's the caveat you left out to keep that clean?

That last one is unreasonably effective. Explanations get simplified by omission, and asking directly for the omission is faster than discovering it in a bug.

Checking your own understanding

Before you leave a concept, run this:

I'll now explain [CONCEPT] back to you in my own words.
Point out anything I've stated too confidently or got subtly wrong.
[your explanation]

If you can't write the explanation, you haven't finished the chapter.