Narrating Your Thinking

2. Saying the Trade-off Out Loud

Most reasoning that goes unheard is a trade-off you resolved in your head. There is a form for saying it out loud, and it takes about fifteen seconds.

"Two ways to do this. A gives me X and costs Y. B gives me the opposite. I am taking A because of Z."

Four clauses: the alternatives, the cost of one, the cost of the other, the choice with its reason. Most candidates say only the fourth.

A worked version: "I could fan this out when the write happens, or assemble it when somebody reads. Writing it out makes reads cheap and makes a popular account very expensive. Assembling on read is the opposite. I am writing it out, because most accounts here are small, and I will come back to the large ones."

Beyond showing your reasoning, that tells the interviewer where to probe, which is usually somewhere you have already thought about, and it sets up the next thing you say, so the narration is doing real work rather than filling space.

One refinement worth adopting: when you reject an option, describe it the way somebody who prefers it would. Rejecting a weakened version of an alternative is a reliable signal of inexperience, and interviewers hear it immediately.

3. When to Ask and When to Decide

Candidates get this wrong in both directions. Some ask about everything, which reads as unwilling to commit. Others ask about nothing, which reads as not noticing the ambiguity. The rule that resolves it:

Ask when the answer would change your design. Decide, out loud, when it would not.

If knowing the answer leads to two genuinely different designs, ask. If you would build roughly the same thing either way, state an assumption and keep moving, because the question costs time and returns nothing.

There is a second case. Sometimes the answer would change your design and the interviewer clearly does not want to supply it, which you can tell because they deflect once. Then decide, and say what you decided and why. "You have not given me a number, so I am going to assume the larger one, because designing for the smaller one and being wrong is the more expensive mistake."

The behavior to avoid is asking a question, receiving a vague answer, and then proceeding as though you had received a precise one. Interviewers remember that, because it is what happens on real projects when a requirement was never settled.

4. Recovering From a Wrong Turn

At some point you will realize that something you said twenty minutes ago is wrong, or that the design you are building has a flaw. This feels like the worst moment of the interview. It is one of the best scoring opportunities in it.

Interviewers need evidence of how you behave when you are wrong, and they rarely get it, because most candidates either do not notice or quietly hope nobody else did. The form is short, and the order matters.

1. Name it plainly. "I have made an error. The approach I described does not handle the case where two of these arrive at once."
2. Say what it breaks, specifically rather than vaguely.
3. Say what you are changing, and whether it is a local fix or a different design.
4. Move on at the same pace. Do not apologize repeatedly, and do not restart from the beginning.

What costs candidates here is not the error. It is defending it after noticing, which turns a recoverable mistake into evidence about how you take feedback. If an interviewer points out a problem and you are not sure they are right, say "let me check that" and actually check it, out loud.

5. The Failure Modes of Narration

Narration done badly is worse than silence, in four recognizable ways.

• Narrating the obvious. Describing what a load balancer does consumes time and suggests you think it is the interesting part. Say what you chose and why, not what the component is.
• Filling silence. Thinking time is allowed. "Let me think about that for a moment" followed by actual quiet is far better than talking while you work it out, and interviewers read it as composure.
• Narrating without deciding. A candidate who names every consideration and never resolves them has demonstrated awareness and failed the part that matters. Every trade-off you open must be closed.
• Explaining rather than reasoning. Lecturing on a topic you know well is the most common way a well-prepared candidate uses their time badly. The interviewer is not assessing whether you can explain a well-known algorithm; they are assessing whether you can choose.

A useful self-check during the interview: in the last two minutes, did I say anything that involved a choice? If everything you said was a fact that any competent engineer would also state, you have been explaining rather than deciding.

6. The Senior Decision

Narration is not something added on top of the work. Done properly it is the work, made visible, and three habits account for nearly all of it.

• Announce the decision before the detail. "I am going to put this behind a queue, and here is why" orients the listener, who can then follow the reasoning instead of reconstructing it afterwards.
• Attach a condition to every choice. "...unless the volume turns out to be much higher, in which case this changes" is one clause, and it converts an opinion into a decision with a boundary.
• Say when you change your mind, and why. This is the one most people omit, and it is the one that most strongly separates somebody who reasons from somebody who recites.

All three are habits rather than techniques, so they transfer. An engineer who announces decisions with conditions attached and revises them out loud is easy to work with, easy to disagree with productively, and easy to trust with an ambiguous problem. That last phrase is the definition of the level this whole module is about.

What the interviewer is scoring: the moment interviewers most often cite in strong feedback is a candidate changing their mind, because it is rare and because it is the clearest available evidence that the reasoning was real rather than recalled. Another that recurs is a condition attached to a choice, usually quoted almost verbatim. Neither is about knowledge. Both are about making the inside of your thinking available to somebody who has forty minutes to decide whether to trust you with problems that have no known answer.