How do you keep your own question card current between rounds of a multi-round loop?

## Why the card has to move A loop is not four independent conversations. What round one tells you changes what is worth asking in round three, and the questions you generate during the day are usually better than the ones you wrote the night before, because they come from information rather than from public research. The maintenance loop is small: **two minutes between rounds**, three actions — strike, add, re-rank. ## Strike Draw a line through every question the last round answered, on every card it appears on. This is the mechanical defence against the worst version of a long day: asking the fourth interviewer something the first one already told you. That does not just waste a slot; it suggests you were not paying attention earlier, and it is entirely avoidable with a pen. Striking is also what keeps a live top line at the top of every card. When the handover comes and you are tired, you want to read, not recall. ## Add Write down the questions the round created. During a peer round you might learn that the mobile team ships on a fixed cadence and that a chunk of the work right now is paying down a migration. Neither fact was on any card that morning; both generate strong questions for the manager round — what the migration is costing in feature time, who decides when it ends. Add them as questions, not as topics. "Migration" written alone will not survive to the next handover; the full sentence will. ## Note, lightly Alongside a struck line, write a keyword or two of what you were told, plus who told you. Two reasons. First, routing: if the recruiter said the level was set by the scope of a specific surface, that is the sentence you build the manager's question from. Second, the decision: at the end of the loop, the card carries what each person actually said, which is far better than trying to reconstruct four conversations that evening. Keep it light. A candidate transcribing answers is not making eye contact, and the interviewer will feel it. Keywords, not sentences. ## Re-rank Before the next round, re-order that round's four against the person coming in: promote what only they can answer, demote what the day has already partly covered, drop anything the last round made irrelevant. If the day has answered enough that a card is down to two live lines, top it up from the added list — you want four prepared and two asked in every round, including the last one, where fatigue makes improvising most likely. ## A note on repeating a question deliberately Sometimes the same question genuinely belongs on two cards because two people would answer from different vantage points. That is a deliberate choice you mark on the card, not an accident of never having updated it. The distinction matters: a repeat you planned reads as thoroughness, and a repeat you forgot about reads as inattention. ## Where to keep it One card, or one note, for the whole loop, with a block per round. Splitting it across four documents means the strike in round one never reaches round three. If the loop runs across several days, re-read the whole thing before each day — not for the answers, but to rebuild who said what. ## Worked example across a day Morning, recruiter round: four prepared, two asked, two struck. You add one question the call created — the team has been shipping fixes rather than features for a while, so "what would need to change for that to flip?" goes on the manager's card, ranked first. Midday, peer round: the release cadence question gets answered in passing before you ask it, so it is struck; the engineer mentions a migration, so a new question goes on the manager's card. You still ask two. Afternoon, manager round: your card now leads with two questions the day generated and neither was written the night before. Those are the two you ask. End of the day the card holds sixteen prepared lines, eight asked, most struck, and a keyword beside each. It is the only place the loop exists in writing, and it is what you read before deciding anything. ## The tell it produces Interviewers late in a loop can usually feel whether a candidate has been listening all day. A question that starts "earlier today someone mentioned X — how does that look from your side?" is only possible if the card moved. That is the real return on two minutes between rounds.