Hey Reader,

Over the last two weeks I posted how WhatsApp works, in four blocks. This is what I see when I ask someone to build it in an interview.

I sit on the other side of that table often. What follows is a composite. Not one person. A pattern.

It does not matter what I ask

Design WhatsApp. Design a parking lot. Design a URL shortener. The question changes. The failure does not.

The candidate nods. Then, inside the first two minutes: load balancer. Database. Cache. Queue.

Each word is said and left there. None is explained. Nothing has been sent, parked or shortened yet. There is no user. We have a cache for a thing that does not exist.

The part that never comes

I wait for the moment the candidate says "and now the app grows, so we need X". The moment where the system evolves from one user to many, one problem at a time.

It never comes. They jump from a blank page straight to scale and reliability. The basics are skipped, not failed.

Here is the rule I say in my head every time.

There is no point doing optimisations on a functionality that does not exist.

It is the fastest way to get rejected in my room. Not a wrong answer. A skipped one.

What I am actually grading

Boring things. One phone sends. One server accepts. One phone receives. Now a problem shows up. Phone B is off. So you need to hold the message. Another problem. The server is busy. So you need a queue.

Each box appears because a problem appeared. That is the evolution I am grading. Not the boxes. The reason each box exists.

The four blocks I posted are that path, in order. Accept the message. Deliver it. Hold it if the phone is off. Then, and only then, make it fast.

Why smart people skip it

They think the basics are too basic to say out loud. They think I want the big words. The opposite is true. Big words with no path make me doubt the small things too.

Do this in your next round

1. Before you name one component, say how one user does one thing, end to end.
2. Grow it one problem at a time. Say the problem. Then the box.
3. Only when the basic path works, ask "what breaks at scale". Now your cache has a reason.

Reply with the system you are most scared to be asked. I will reply with the first three boxes I would want to see.

Next Thursday: the call I got a few months into a new job, and why I said yes in two days.

Either way, keep going. The effort you're putting in will compound.

TheGuidedGrowth.