A New Path

Twenty years ago I walked into a college career office with a graphic-design course load and a stack of psychology credits I wasn’t going to finish. I walked out with a part-time job shooting product photos for an ad company. The job also needed a print catalogue and a website that changed every week. So I did those too.

That is the whole career, in miniature. Nobody handed me a lane. I took the work that was there and learned what it needed.

For the last ten of those years the work was writing web applications. Now I’m turning toward deciding what they should be — a smaller change than it sounds, and I want to say why.

The deciding was always the part that mattered — what to build, for whom, in what order, what to refuse. The typing was how you found out whether you’d decided well. Now the typing is going to the machine, and the deciding is what’s left in my hands. So I’m reaching for it on purpose.

Three ways of seeing one thing

Three things taught me how to decide. For years I mistook them for three careers.

Story. Writing first, then film — six sets in front of the camera, half a dozen behind it, grip to operator to director. Film teaches you that there is a story in everything, and that the idea behind a product has a story of its own: where it came from, what it cost, what it had to give up to get made.

Interface. Design, then front end. An interface is not decoration laid over a system. It is where the system and a person meet, and both of them show what they are. And the interface is a function of the data — decide the shape of what is stored and what is derived, and you have already decided the product; the screen only reports the decision. A good interface also knows the path you took to reach it. It opens holding what you did last time, because whoever laid it out had walked the same road and left the door open. A pocket of order, set into the noise on purpose.

Data. Ten years full stack, and the lesson the other two couldn’t teach: it starts with the shape of the data. What exists, what relates to what, what is kept and what is thrown away, what must never be deleted. Get the shape right and the product falls out of it. Get it wrong and no interface will save you.

These are not three trades. They are one way of seeing, turned to catch three faces of the same object — and they are, exactly, what deciding a product asks for: a sense of what it is for, what it owes the person who reached it, and what it is.

A different axis

There is a fourth thing, and it is not a fourth trade. The three above are what a product is. This one is how it gets made — the process, the operations, the plumbing of shipping something and keeping it alive. I set it apart because it sits on its own axis. And because it is where the ground actually moved.

The next layer

Computing has only ever done one thing. It builds the next abstraction on top of the last, and hands it to you through an interface. Assembly hid the machine. The compiler hid the assembly. The framework hid the compiler. Every layer was a new interface onto the one below, and every one let you decide more and type less.

The language model is the next layer. It hides the framework. New hardware is shifting the floor beneath it, and plain language has become a way to program — so the interface for making software has changed as much as the software has. Nothing is ending here. The abstraction took one more step, and this time it stepped onto the interface itself: the thing I spent a decade learning to read.

The shape underneath

Watch one shape appear three times.

An application’s meaning is never written down. It is cut out by use. A person acts, the system answers, the person acts again, and a thousand turns of that loop wear the thing down to what it is actually for. Nobody authors it. Use does.

A working life goes the same way. The elements accumulate — story, interface, data, the craft of shipping — and you run them through enough years that something you never set out to learn emerges from the pile: judgment about what to build. It arrives right as the work begins to need it.

And computing goes the same way. Patterns accumulate, get reused and repurposed, and a new abstraction emerges from the heap. Stock becomes flow. The new thing then becomes the floor the next thing is built on.

Same shape at every scale. I did not decide to see the world this way; it is simply how it looks to me, and it is why this turn feels like arriving somewhere rather than leaving. The elements were the stock. Product judgment is the flow they finally became.

The position

So: product — engineering, development, management, whichever word the room uses for the person who decides the shape and answers for it. Not a plan. A description of where the work already went.

The machine will take the typing. It will not decide what the data should be, or whether the thing built is the thing that was wanted, or what to refuse. That is the seat I want: the one where a person decides and stands behind the call, and the machine does the pushing. If that is the seat you need filled, tell me what you’re building.