The Learning Loop — How to Study Every Topic

In plain words: like learning to ride a bicycle. Watching someone else ride (reading a tutorial) doesn't mean you can ride. You only really know it once you get on, wobble, fall, and get back up — that's what 'explain without notes' and 'break it, fix it' do for a system design topic.

The loop, verbatim: Learn → Close the tutorial → Explain without notes → Implement → Break it → Fix it → Explain trade-offs.

Explanation without notes is the checkpoint that actually transfers knowledge — if you can't, you re-learn that day.

Implement with intent: write the code yourself from the design, not from memory of the sample.

Break it on purpose: change a requirement (new vehicle type, provider failure) and watch which seams the design has.

Fix it: every break reveals a missing abstraction or a missing test — the fix is the learning.

Java/Spring translation drill: 'generic concept → Spring implementation → production problem' for every topic.

The Kafka example from the syllabus: don't stop at 'distributed broker' — be able to explain partitions, consumer-group rebalances, broker death, ordering, retries, duplicates, and idempotency (all of it is in Phase 10).

The rate-limiter example: LB → Spring Boot instances → Redis — implement it, then ask: what if Redis is down, what at 100k req/s, how do instances coordinate (Phase 12.2 answers all of it).