Velocity and craft

The tools are fast. Anyone who has used them for a week knows it. You can design a user flow in a day and build a component library in a week.

That is true, and it is the least interesting part. The tools speed up production and nothing else. The person looking at your work still takes it in at the same speed they always have. Everything that separates one body of work from another lives in that gap. What I want to figure out is how to take the speed without losing the judgment.

What craft consists of

Production used to be a real skill. It stayed scarce for as long as it stayed hard, and it stopped being hard for everyone at once.

A skill everyone has tells no one apart. Whatever is still scarce becomes the thing that does.

Right now that is craft. A model gets you a competent first seventy percent. Everyone gets the same seventy percent, so the last thirty is the whole margin.

Craft is what consequences leave behind. You ship a decision, it turns out slightly wrong, someone feels it without being able to name it, and you remember. After enough of those you start catching the wrong decision before it ships.

The thing doing the catching is taste. I mean trained perception here, which is different from preference. It is the ability to see a difference that looks like nothing to everyone else, and to explain why one side of it is better. You cannot build what you cannot see is missing.

You cannot speed perception up, because consequences set its pace. A pattern shows it was wrong when someone misuses it six months later. An architecture shows its limits when the fourth feature arrives and has nowhere to go. Those clocks do not care how fast you shipped.

The trap

A lot of the friction these tools remove is the friction that forms perception. Being stuck is miserable. It is also when you look at a problem closely enough to learn its details.

If your taste is still forming, the tools ask you to review output with judgment you do not have yet. Reviewing well takes the discernment that doing the work used to build. So the tool that needs taste to use well gets sold as a replacement for the work that builds taste.

That loop will not break on its own. The only way out I can see is to keep choosing to do some things slowly, on purpose and at real cost, even though the tool could do them now.

Systems that multiply speed

Once your taste is formed, the tools are just good. You can see what is wrong with the output, so you throw out the bad parts and keep the speed.

The bigger win is leverage. You can write formed taste down. It becomes a component, a token, a convention, a line on a review checklist. Judgment written down like that applies itself faster than any hand could, and this is where craft actually speeds up.

It cuts both ways. A system spreads the judgment that made it, shaky parts included, and it spreads them exactly. A shaky assumption gives no warning. It shows up in forty screens at once, and then people defend it as consistency.

So put more of your attention on foundations than feels reasonable. They are the one place where iterating multiplies a mistake instead of fixing it.

The exchange

Everyone can ship now, and that is not going back.

So move quickly. The speed is real, and turning it down proves nothing. Just be honest about what you are trading, and protect the slow work that builds the judgment everything else rests on.