What teaching kids Pong taught me about onboarding

Three tulips, two red and white and one yellow, in front of a brick house with striped awnings

When I teach kids to code, I start with Pong. Two paddles, a ball, a score, and a game most of them have never seen but understand instantly. It is old and simple, and within ten minutes something they typed is moving on the screen. I have taught enough sessions to believe the first ten minutes decide whether the rest of the lesson happens.

I think the first week of an engineer's onboarding works by the same rules.

Something real has to move early

A kid who has a ball bouncing before the first break will sit through the explanation of what a variable is, because the variable now controls the ball's speed. A kid who gets the explanation first, with the ball promised for later, is already looking at the clock.

I think the goal of an engineer's first week is a real change, in the real codebase, through the real pipeline, into something that runs. A one-line fix that ships is worth more than a week of context, because the context now has something to attach to.

One idea at a time

Pong is a sequence of single ideas. Draw a rectangle. Make it move. Make it stop at the wall. Make it bounce. Each is small enough to get right and makes the next one necessary.

Onboarding tends to do the opposite. It front-loads everything, the architecture, the history, the tooling, the team norms, all in the first three days. I think the better shape is a sequence of small real tasks, each introducing the one new idea it needs, with the architecture talk arriving when a task requires it.

Breaking it is the lesson

The best moment in any Pong session is when a kid changes a number to something absurd, the ball disappears, and they have to figure out why. That is the first time they are debugging, and nobody told them to.

An engineer needs the same thing, a way to break things safely in week one. A local environment that runs, a test suite they can watch fail, a branch where nothing matters. I think how fast a new engineer can break the system and recover is the best predictor of how fast they will become useful, and it is almost entirely under the team's control.

A person, not a document

I have never had a kid learn Pong from the handout. They learn it from the person next to them, who answers the question they have instead of the one the handout anticipated.

Documentation has its place, but I think the first change should be made with someone, pairing, in real time, with the new engineer typing. It is the fastest way to transfer the unwritten things, and the unwritten things are most of what a team knows.

The sessions that go well start with motion, add one idea at a time, let people break things, and keep a person close. I have not found a reason those rules stop applying when the student has a job title.

Photo source: https://photos.robertstowe.com/flowers