Metacognition and Learning — Mistake¶

When it's worth the full process, and when it's not¶

Do it when: you'll depend on this concept later — production code, an on-call shift, a system you'll own — not just a one-time answer to move past right now.
Do it when: you notice you've skimmed the same doc two or three times and still can't explain it. That's a clear sign to stop rereading and switch method.
Skip the full process when: you just need one fact to unblock yourself right now, like a function's exact argument order. Not every lookup needs to become a learning project.
The real cost: building something yourself and coming back to it later both take more time up front than just skimming a doc once. It's worth that time only for ideas you'll actually reuse.

Mistaking "this feels familiar" for "I understand this." Rereading a doc or rewatching a tutorial makes it feel smooth and easy, and that smooth feeling gets mistaken for real understanding — researchers call this the illusion of fluency.
Fix: close the material and try to explain or rebuild the idea before deciding you "get it."
Using rereading and highlighting as your main study method. A large review of ten study techniques found rereading and highlighting are some of the weakest, even though they're the most common way people study.
Fix: replace them with testing yourself (retrieval) or explaining why something works (elaboration) — the same review rated both far higher.
Reaching for a complicated word instead of a plain one. Using jargon to describe something can hide a fuzzy spot instead of showing real mastery.
Fix: if you can't say it in plain words, you don't understand it yet — that's the whole point of the Feynman technique.

Skipping the guided example when a concept is truly brand new. Jumping straight to "figure it out yourself" on something you have zero background in wastes time rediscovering things a five-minute example would have shown you right away.
Fix: for a genuinely new idea, follow one worked example first — it's the right amount of help for a beginner.
Staying on guided examples too long once you're not a beginner anymore. The same research that shows worked examples help beginners also found they stop helping — and can even get in the way — once you have real background in that area.
Fix: once you're not starting from zero, move to building on your own sooner rather than following another tutorial.

Cramming everything into one sitting, right before you need it. It feels efficient, but a single session mostly fades within days.
Fix: spread your practice across a few days instead of one long session, even though it feels slower.
Treating "I finished the tutorial" as "I learned it." Finishing a guided example only proves you can follow someone else's steps — it hasn't tested whether you can produce the idea yourself yet.
Fix: the tutorial isn't done until you've built a small version alone, with it closed.