On Making Things
God didn't write a spec. We spent a few decades pretending craft could be replaced by a pipeline. It can't. It never could.
February 2026
God didn't write a spec.
There's no moment in Genesis where the design is separated from the making. No phase where the requirements are gathered, the mockups reviewed, the stakeholders aligned. There's just: let there be light. And there was light. And it was good.
Design and creation, in a single breath.
When a medieval stonemason built a cathedral, he didn't work from detailed blueprints. He had a vision, a set of principles, and his hands. The design evolved as the stone went up. He'd set a column, step back, feel whether the proportion was right, and adjust the next one accordingly. The building was the design process. Each stone was both a decision and a discovery.
When a potter throws a bowl, they don't design it on paper first. They put clay on the wheel and begin. The clay has opinions. It resists certain shapes and invites others. The potter's skill isn't in executing a predetermined plan — it's in listening to the material, sensing its balance, knowing when to push and when to let go. The bowl that emerges is a collaboration between intention and material.
When a jazz musician plays a solo, they're composing and performing in the same instant. Each note is both a plan and an execution. The musician is thinking with their fingers.
In every case, the magic lives in the unity of design and creation — what Sennett calls the intimate connection between hand and head, and what Crawford argues you destroy the moment you sever it. There's a word for this way of working: craft. The craftsperson holds the vision and the tool at the same time. They think with their hands. The quality of their work comes not from following a plan but from the accumulated judgment of someone who has made things before and learned, through making, what good feels like.
So how did we end up separating them?
Scale. When you're building a cathedral with hundreds of workers, you can't have everyone improvising. You need plans. When you're manufacturing thousands of identical objects, you need specifications. Separation of design from making is a solution to a coordination problem, and it works.
But something strange happened.
We forgot it was a compromise and started treating it as an ideal.
We built entire management philosophies around the idea that thinking should happen first and making should happen second — that the planning phase was where the real intellectual work lived, and the building phase was just execution.
The software industry took this further than anyone. We created an elaborate apparatus of roles and rituals to enforce the separation. Product people wrote specs. Designers made mockups. Engineers did the "real work" of building. Each group operated in its own room, with its own tools, and they were connected by documents and meetings and handoffs.
This wasn't just an org chart. It changed how people saw themselves. The person who wrote the spec came to believe they'd done the important work. The person who built the feature came to believe their job was to follow instructions. And both were wrong, because the separation had severed the feedback loop that makes craft possible. The spec author couldn't feel the material pushing back. The engineer couldn't reshape the vision in response to what the code was teaching them. Each had half the picture and no way to hold the whole.
The result was software that felt like it had been designed by committee — because it had been. Correct, but not coherent. Functional, but lacking that quality of rightness you feel in a well-made thing, where every detail seems considered, where every interaction feels inevitable. That quality comes from someone who was present at every level — who cared about the grand vision and the smallest detail, and who understood that they're the same thing. It's nearly impossible to produce through a pipeline. You can get correctness from a pipeline. You can get polish.
But rightness emerges from the thousands of tiny decisions that happen in the act of making — decisions too small to spec, too subtle to document, too numerous to review.
When pipeline-built products do have that quality — and some clearly do — it's almost always because someone at the top held the whole vision even as the work was divided. Steve Jobs refusing to ship until the scroll bounce felt right. The best products survive the pipeline not because of it, but in spite of it — because one person refused to let the separation erase the coherence.
For a long time, there was no alternative. The tools were too complex, the feedback loops too slow. You couldn't hold the whole thing in your head while you made it, because making it required so many steps, so many people, so many layers of translation. The separation was costly, but you couldn't build anything substantial without it.
That's changing. LLMs have compressed the distance between idea and working prototype to the point where a single person can build in a weekend what used to take a team a quarter. When that happens, the coordination problem that forced the separation dissolves. One person can once again hold the vision and the material in the same hands. They can design by making. They can think with the code.
A designer at a fintech company had spent months watching merchants struggle with something. These small-business owners would get a payment notification, and then — because the app's transaction history was hard to parse — they'd spend several minutes figuring out who paid them and for what. She'd mocked up the fix twice. Both times it went into the backlog and sat there, because the engineering team had higher priorities. So one weekend she built it herself — not a mockup, but a working feature. The code was probably a mess. But the product was extraordinary, because she'd been doing what craftspeople have always done: holding the problem and the material together, refusing to let go of either until it was right.
The plan is a guess. The making is the truth.
You don't fully understand the problem until you start building the solution. You don't know what the right design is until you feel it taking shape under your hands.
I'm not arguing against all planning. Even the cathedral builders had rough sketches. Even the jazz musician knows the key. Vision matters. Intention matters. Having a sense of where you're going is essential.
But the sketch is not the building. The key signature is not the solo. The spec is not the product. These starting points are useful precisely because they're loose — they provide direction without constraining the discoveries that happen in the act of making.
The last few decades created an anomaly: an industry built on making things where the people with the ideas and the people doing the making were different people. That anomaly is ending. And for the first time in a generation, it's possible for the person with the vision to be the person who builds it.
This brings me back to Genesis. Whatever its theological meaning, it describes a way of making that I think is fundamentally right: the creator creates. Not plans and then creates. Not designs and then builds.
Makes the thing, judges it, adjusts, continues.
Let there be light. And there was light. And it was good.
That "and it was good" is the whole method. Make something. Step back. Is this right? Does it serve the whole? Adjust. Continue. Carry what you learned into the next act of making.
We spent a few decades pretending this could be replaced by a pipeline. It can't. It never could.