The interview that has no answer

With a fortnight to go before a whiteboard design interview, the preparation is remarkably consistent. Twenty tools committed to memory. One diagram practised until it comes out tidy. Enough vocabulary that nothing on the day makes you stumble.

That is good work. It is aimed slightly to the left of what is being scored, and this is about where it lands instead.

Why this one is different from the rest of the loop

Every other interview has something written down behind it. The coding one has an answer, because the code runs or it does not. The behavioural one wants a story with a recognisable shape. Even the quick-fire questions everyone complains about are marked against something.

This one has no answer at all. It has a set of tools, each of which hurts somewhere, and the person opposite already knows exactly where, because they run the system you are being asked to design.

So they are not sitting there waiting for the correct tool. They are waiting to hear you notice what the tool you picked costs you.

That is the part memorising cannot reach. It is entirely possible to name every item on the board correctly and never say one sentence anybody was listening for.

A way to see the difference

Somebody is sent to buy food with a list and a fixed amount of cash.

They know what everything on the list costs. They know which trader is cheaper on which day. And they can still come home and be told they got it wrong, not because a price was forgotten, but because one item had doubled overnight, the money did not stretch, and standing there they had to decide what to leave behind.

Nobody thinks the prices were the test. Knowing the prices is what gets you as far as the stall. The decision at the stall is what is being judged, and those are two different abilities that only look like one.

And notice what gets asked at home. Not what things cost. Why that particular item is not in the bag. Answer that, and the errand was done well whatever came back.

What is actually being scored

One thing, and it is narrower than it sounds: can you hold two tools that both cost something, choose between them, and say why out loud.

Out loud is the whole of it. Not choose, but choose and name what made the other one tempting and what this one gives up.

On its own, a pick is a guess that came out well. Attach the price and it becomes a decision, which is the kind of thing somebody can trust you to repeat next month with nobody watching.

Nobody needs you to be right. Being right in a room with no answer is mostly luck. What travels is reasoning somebody can see working.

Let me be fair to the tool lists

You cannot reason about something you have no word for. Without a name for holding a duplicate of your data somewhere else, that option tends not to get raised at all. The study lists exist to close exactly that gap, and they close it well.

So a name is a price tag, and picking between two of them is the moment at the stall. Nearly every resource stops at the tags, and not from laziness: twenty tools can be written down, filmed and ticked off, while judgment has no list to tick.

The exercise, which is free

Find two tools that could each do the job. Somewhere to put the data, and a second somewhere. One approach to breaking a large program into parts, and a rival approach. Two answers to work that is queued rather than handled immediately.

Commit to one. Then say aloud what committing cost you. Not the upside. In the shower, walking home, or to a friend outside the industry, because if they can follow you it is clear enough for a room where somebody is deciding about you.

Ten of those and the room changes character.

None of this makes the interview easy, and it would be dishonest to suggest otherwise. Deciding aloud while unsure feels bad, and that feeling is not a signal about your preparation. It is what the room is made of. Anybody who looks relaxed doing it has not escaped the feeling; they have simply met it more times, usually with nobody there.

The page of pairs to practise on is here. One worked example, marked as an example, and the rest blank on purpose, because the sentences have to be yours or they are no use to you in the room.