How to practise: plans and mock interviews

A design round rewards practice. A candidate who only read lessons knows the ideas, and often runs out of time or forgets the numbers.

A weak plan only reads. A strong plan trains the clock, the numbers and the spoken answer.

Think of a cricket player. Batting against a real bowler helps more than reading about batting. After each session, a coach names one thing to fix. You need a plan, a mock and a review.

A mock interview is a full practice round under real conditions. You get a new problem and a timer. Someone plays the interviewer.

A weekly plan mixes five kinds of work. Lessons teach the ideas. A full problem, spoken aloud, trains the six steps of lesson 2. A short drill trains one trade-off or incident answer (Tradeoff's Drill takes 7 minutes). Review brings old mistakes back. A mock tests everything under the clock. As the interview comes near, move time from lessons to mocks.

How to run a mock.

1. Take a new problem and set a timer for 45 minutes. Use the clock of lesson 2: requirements 5, entities 2, API 4, data flow 3, design 10, deep dives 18, closing 3.
2. The interviewer gives a one-line problem, answers your questions, and adds one twist at minute 24 (lesson 114). Give hints only if asked. Count them.
3. Speak aloud all the time, or record your voice. Stop at minute 45.

How to review. Do it within one hour. Answer six questions with yes or no:

1. Did I ask questions and get numbers before I drew?
2. Did every step stay inside its minutes?
3. Does each box have a number?
4. Did I say a trade-off sentence for each big choice (lesson 6)?
5. Did I handle the twist with the five moves (lesson 114)?
6. Did I finish without a hint?

Take the two worst answers. Write each as a rule: "When ___, I will ___." Redo that step on a new problem within 3 days. In spaced review, a card comes back after 1 day. Each right answer makes the wait longer: 3, 7, 16, then 35 days. A miss sends it back to 1 day. Tradeoff's Review page works like this.

Assumption: you have 5 hours (300 minutes) a week, and a lesson takes 45 minutes.

The total is 90 + 60 + 21 + 50 + 75 = 296 of 300 minutes.

Now a mock log: requirements 9, entities 2, API 4, data flow 3, design 10, closing 3. That is 9 + 2 + 4 + 3 + 10 + 3 = 31 minutes without deep dives. So the deep dives get 45 - 31 = 14 minutes. This answers the predict box. The guide gives 18, so you lost 4 minutes.

In the review, questions 2 (steps in time) and 6 (no hint) are "no". They give two rules:

1. "When the requirements pass minute 5, I will say 'assumption' and start the entities."
2. "When I am stuck, I will say what I know and make an assumption, before I ask for a hint."

The redo for rule 1 is the requirements step alone, on a new problem, with a 5-minute timer, within 3 days.

A mock with a person solves live speaking and honest feedback. It costs their time. I would use an AI interviewer, or my own recording, when nobody is free. I would add lessons if a mock shows missing ideas, not missing habits.

"The best preparation is to read more lessons." Reading does not train the clock or your voice. Lessons take 90 of 296 minutes.

"A mock needs an expert interviewer." A friend with the six questions finds many mistakes.

"After a bad mock, I should take another at once." Without a review, a second mock repeats the mistake. Write the rule first.

1. Name the three parts of a mock run and the six review questions.
2. You have 4 hours (240 minutes) a week. A mock with review takes 75 minutes, 3 drills take 21 minutes, and review cards take 50 minutes. How many 45-minute lessons fit in the rest?
3. A mock log: requirements 5, entities 2, API 4, data flow 3, design 16, closing 3. How long are the deep dives? What rule do you write?