A peer engineer says code reviews here are pretty fast — how do you follow up in that round?
answer
- Climb down one rung at a time
- Adjective, instance, typicality, mechanism
- Two follow-ups, then move on
- The hedge itself is data
- Ask a second interviewer the same wording
basics
~20 sStep down one rung, not sideways: ask how long their last change waited for a first comment, then whether that week was typical. Two follow-ups, then move on and ask another interviewer the same question rather than pressing.
solid answer
~40 sI climb down from the adjective to an instance to a mechanism. First: 'How long did your last change wait for its first comment?' That converts an impression into a fact without challenging them. Second, if the number is surprising in either direction: 'Was that a normal week?' Third, if I still have room: 'Who reviews changes on this team — a fixed set of owners, or whoever has capacity?', because the mechanism explains the number. Then I stop. Two follow-ups on one topic is the limit before it reads as an interrogation, and a peer round is scoring me too. If the answers stayed vague, I note the hedge and ask a second person on the loop the same question; two accounts that disagree tell me more than pressing one person ever would.
go deeper
Learn one follow-up and use it: after an adjective, ask how long their last change waited for a first comment. Getting one concrete instance is already better than most candidates manage.
Be able to run the ladder in order, from adjective to instance to typicality to mechanism, and to explain why the mechanism predicts whether the number holds when someone is away.
Show that you read hedges and cross-check. Know when to stop pushing, record the exact wording of the answer, and ask a second person on the loop the same question rather than pressing the first one harder.
Own the tradeoff between the information and the relationship. You are collecting evidence from people you may soon lead or depend on, so decide what a vague answer is worth to you rather than trying to force a clean one.
## The situation You asked a peer engineer about code review and got an adjective back: reviews are pretty fast. The temptation is either to accept it and move on, or to push until the answer improves. Both are mistakes. The first wastes the question; the second turns a conversation into a deposition with someone who is also writing feedback about you. The move is a short ladder, taken one rung at a time. ## The ladder **Rung 1 — adjective to instance.** 'How long did your last change wait for a first comment?' This is not a challenge; it is a request for a memory, and most people answer it happily. You will usually get a real number or an honest 'I don't remember, but the one before sat a while' — which is itself an answer. **Rung 2 — instance to typicality.** 'Was that a normal week, or was something going on?' This catches both the flattering outlier and the unfair one. A peer who says their last change got a comment inside a few hours but adds that the week before it sat 31 hours has just given you the range, which is worth more than either endpoint. **Rung 3 — number to mechanism.** 'Who reviews changes here — a fixed set of owners, or whoever has capacity?' The mechanism predicts the number's stability. A team where two people must review everything has a turnaround that collapses whenever either is on leave, whatever this week's figure says. Three rungs is the whole ladder, and you will often stop at two. ## When to stop Stop after two follow-ups on the same topic. Stop immediately if the peer's tone changes, if they hedge in a way that looks like discomfort rather than uncertainty, or if the answer starts to be about a specific person. Nothing you gain from a third push is worth what you lose: the peer is a future colleague and a current evaluator, and 'grilled me about our review queue' is a note that gets written. Stopping is not giving up. It converts the question into a different instrument. ## Hedges are data, not obstacles When a peer cannot give you a number, listen to the shape of the deflection. 'It depends on who you ask' points at variance between sub-teams. 'It's got a lot better recently' names a problem in the past tense and invites the obvious follow-up about what changed. A laugh followed by a topic change is not proof of anything, but it is worth writing down. Record the hedge and move on. You are not owed an answer in that moment, and the round is not your only sample. ## The cross-check, which is the actual senior move Ask the same question of a second person on the loop, in the same words. If one peer says a first comment usually lands within about 5 hours and another says their last change waited into the next day, the disagreement is the finding. It might mean the two work in different parts of the codebase, that one is much more senior, or that the answer genuinely varies — and any of those is a real thing you now know about the team. A cross-check costs nothing, needs no confrontation, and produces better information than any amount of pressure applied to one person. Keep a note after each round with the exact question and the exact answer, because by the fourth conversation of a long day you will not reliably remember who said what. ## Tone that keeps the door open Give half a sentence of reason when a follow-up could sound loaded: 'I ask because I've worked somewhere reviews stalled for days and it changed how I planned my week.' That reframes you from auditor to someone with a preference, and it usually causes the peer to volunteer more than the question asked for. React to the answer as a colleague would, rather than nodding and firing the next question, and be visibly fine with an imperfect answer. ## What this is not This is a technique for eliciting specifics inside a round, not a verdict on the company. One slow review week is not a conclusion, and deciding what a pattern of vague answers should mean for whether you take the role is a separate judgement made later, with the whole loop in front of you.
- What if the peer genuinely does not remember how long their last change waited?Take it at face value and move to the mechanism instead: who reviews changes on this team, and what happens when the usual reviewers are unavailable. A structural answer is easier to recall than a duration and often predicts the turnaround better than one remembered number would.
- How do you use two interviewers giving you different answers to the same question?Treat the gap as the finding rather than an error to resolve on the spot. Ask the later interviewer neutrally whether experience varies across the team, which surfaces sub-team differences, seniority effects or genuine variance, and note both answers verbatim so the comparison survives a long day of rounds.
- Can you push a third time when the answer still has not become concrete?I would not. Past two follow-ups on one topic the exchange reads as an interrogation, and the peer is both a future colleague and someone writing feedback. I note the hedge, move to another question, and get my second data point from a different person on the loop.
saying these in an interview costs you the question
- Pressing the same point three or four times until the peer commits
- Treating a hedged answer as proof of a problem
- Asking leading questions that only invite agreement
- Following up in a tone that audits rather than converses
- Failing to write down the exact answer before the next round