The engineer's questions, and the bench

Open. About the past or the present, answerable by anyone who runs the business. "Which systems does one sale touch, from the till to the bank to the ledger?" It costs three to five minutes, it uncovers the facts behind its door, and it earns a point of trust, which is the currency the sensitive facts are priced in. Every open question in the bank carries a teach-back line.

Closed. Yes or no. "Is the close on the fifth working day?" One or two minutes, one fact at most, and only the fact you already had the shape of. Cheap, and useful for confirming, but a meeting made of them learns nothing you did not walk in expecting.

Vague. "What does good data look like to you?" It sounds open. It is not, because it gives her nothing to be specific about, and what comes back is a generic: accurate, on time, the usual. Three minutes for a sentence you could have written yourself, and on this bank that one reaches no fact at all.

Jargon. "Do you have change data capture on the POS, or should we batch?" She has to ask what you mean, so it costs the question plus an explanation, five minutes, and she is a little less sure you understand her business. On the bench that is a point of trust lost, and it is the only kind that loses one. Engineers pay this cost more than anyone, because the vocabulary is the job.

Leading. "The POS is the source of truth, right?" She says "I suppose so", because the question told her the expected answer and she is being polite. Nothing true came back. Something false went into your notes, and it costs later, when you build a pipeline that forces the ledger to agree with the till. On the bench a leading question records a belief with a cost attached, and the fast interview records five of them.

An engineer's output is a pipeline other people trust. The facts that decide whether it is right are every system a number passes through, who can log in to each, which copy is declared real and how often each is read, and they live behind door 4 of the Six Doors corridor. On Adeline's sheet that door holds six facts worth 19 of 68. One sale touches the till, a bank settlement, a ledger run by an outsourced bookkeeper who exports once a month by contract, and a stock spreadsheet that exists three times, with the copy Priya treats as real sitting on her laptop. And one fact, worth 4, is priced at trust 3: Kenneth, the only IT person and the only POS admin login, is leaving in January, and she has not told the team.

The five open questions that open that door, each with the teach-back line you say out loud:

• g08 "Which systems does one sale touch, from the till to the bank to the ledger?" Teach-back: "Every hand-off between systems is a place a number can change. I am counting the hand-offs."
• g09 "Who can log in to each of those, and pull an export?" Teach-back: "Access is the first blocker on most data projects, so I ask early and I ask by name."
• g10 "Do the Singapore and Malaysia outlets use the same product codes? Who decides a new one?" Teach-back: "Two names for one thing is the most common reason a total is wrong. It is a naming problem before it is a data problem."
• g11 "Where do the stock numbers live, and which copy is the real one?" Teach-back: "A file that exists three times has no real copy until someone declares one. I need that declared before I move it anywhere."
• g19 "How often does each system get exported today, and by whom?" Teach-back: "How often a system is read today is a ceiling I inherit until a contract or a login changes it."

The tuned order does not open with them. It opens with three quick questions from other doors: g01, "Tell me about the last time two numbers did not match", then who decides at month-end, then show me the file. They earn three points of trust and put the auditor's letter, the day-five deadline and Priya's six hours on the table before a single data-state question is asked. Then it goes through decision and the current file, and opens the data-state door widest: five of its first eight questions are door 4, and by the time g09 asks who can log in, six open questions have gone before it and the trust Kenneth's fact needs is already there.

Thirty-seven questions: twenty open (two to six per door), four vague, five jargon, five leading, three closed. The open ones are the list you built in session 2 with Adeline's answers already attached. The others are there because they are what gets asked in real meetings, and the bench needs them to show what they cost. The five jargon questions are the ones an engineer asks another engineer: change data capture, the SLA on the ledger sync, medallion layers, how refunds are modelled, the idempotency of the export. Every one is printed with its kind, its minute cost and the facts it reaches in the bank file the page loads; the widget below draws its buttons from that file, so what you click is what it counts.

Six Doors in printed order gets 47 of 68 with no false beliefs and every door opened. Tuning the order for an engineer, three trust-earning questions and then the data-state door widest, adds four points and, more usefully, puts 24 in your notes by minute 20 instead of 20. A search that is allowed to read the fact sheet finds an order worth 62, and it opens with "what happens when Priya is away" before the interviewer has heard the name Priya. Nobody in a real meeting can read the sheet. Optimal against a known answer is not the same thing as good; the framework is the thing you can do on Tuesday, blind, and get 47.

Two hundred shuffles of the same bank average 27.6 and range from 5 to 45. The average is the meeting most people run when they have a good list and no order: some of the jargon lands early and costs trust, a leading question or two gets recorded, and a door stays shut, often the one with the login behind it. The spread is the argument for an order at all. A method that scores 47 every time is worth more than one that scores 45 on a good day and 5 on a bad one, and the bad days are the ones with the auditor's letter due in eight weeks.

Five technical questions in a row spend 25 of the 45 minutes and take trust to the bottom of the meter. The obvious cost is the minutes. The one that matters is what happens to the open questions that follow: the sensitive facts stay unsaid because she is no longer sure you understand her business, and the whole meeting ends with five facts and two doors, the lowest score on the bench. The technical opener is not wrong because the questions are bad; change data capture, the ledger sync and the idempotency of the export are all things you will need answered. It is wrong because they were asked of the wrong person, first. They belong to Kenneth, in week two, and before January.

A closed question that gets a "no" has taught you something. A leading question almost never gets a "no", because it carries its expected answer and the person across the table is being courteous, busy, or both. Fitzpatrick's rule for customer conversations applies without change: people will say yes to be polite. What comes back is not information about her systems; it is information about how agreeable she is, filed as if it were the first thing.

The bench prices each false belief at what building on it wastes. "The POS is the source of truth" costs 4: a pipeline that overwrites the ledger, when the gap was settlement timing and refunds all along. "Pull everything daily" costs 3: month-end needs the number by day three and Monday needs it weekly, so a daily job buys nobody anything. "The vendor API is already in use" costs 3: nobody has ever touched it. "Finance's number is simply wrong" costs 3, because refunds only exist in accounting. "Stock lives in the POS" costs 3: it lives in three spreadsheets. Five leading questions in five or six minutes, and the meeting starts at minus 16 before the first true fact.

Closed questions are not the villain; leading ones are. "Is the close on the fifth working day?" is closed and gets a real fact for one minute, and the bench scores it so. The rule is about sequence: confirm with a closed question after an open one has shown you the shape of the answer, never before. And the anti-lever carries its own limit: a fast interview that asked only closed questions, no leading ones, would score higher than 17 and lower than the framework, because it would still learn nothing it did not walk in expecting. On this sheet the technical opener scores lower still, at 15. The bench does not pretend that speed is the sin. Assuming is.