It makes forty. Someone still has to pick one.

Production used to be the whole constraint. Making the thing was slow, so being able to make it at all was the skill, and everybody organised around that.

That's gone now, mostly. What's scarce is knowing which of the twenty tabs open in front of you is the right one, and if anything that's gotten more expensive.

Forty options is its own problem

I can spin up forty homepage directions before lunch. Feels like leverage, right up until you're sitting in front of forty homepage directions.

Because forty doesn't hold more good ideas than four. It holds the same three or four good ideas wearing different hats, plus thirty-six ways to lose a week arguing about hats. The job quietly moved from making options to killing them, and killing them is the part nobody sells you a subscription for.

Anyone can add. Cutting is the thing that took years.

Knowing the spacing's off without a rule

Someone glances at a layout, says the spacing's off, moves on. They didn't consult a rule. They're matching against every layout they've ever seen, built, shipped, or watched quietly die.

Which isn't intuition so much as a compressed history of consequences. They've watched a hero like that one underperform. They've seen that exact type pairing sing in a mockup and fall apart at 390px on someone's cracked Android. It arrives instantly because the invoice was settled years ago.

And you can't route around it, because the input isn't information. It's outcomes.

The blank page is solved, the rest isn't

Nine seconds from blank page to something on screen. That's what a good tool buys you, and I don't want to be precious about it, because staring at nothing is genuinely the worst part of the job. But a starting point and a finished thing are different animals, and the gap between them didn't shrink an inch.

A starting point only helps if you can see what's wrong with it. Otherwise the speed just gets you a taller stack of plausible work, and plausible? Plausible is the most dangerous thing a design can be. Survives review. Reads as fine. Doesn't work, and not one person in the room can tell you why not.

Everyone I know getting real mileage out of these things already knew what good looked like. Faster, sure. But nobody's been made right by a tool yet.

Ship it, watch it, carry it forward

How do you actually build judgment? Boring answer. You do the work, ship it, watch what happens, and drag the result into the next one.

Watching is the load-bearing part there. Work you never shipped teaches you nearly nothing, because you never found out. Ten things that went live beat a hundred concepts in a folder, and one that failed in public, loudly, in front of people whose opinion you cared about, beats all of it.

Which is also why it won't transfer. Read anyone's principles you like. You still can't inherit the exact texture of having been certain and wrong.

Deciding what should exist

Uncomfortable version, since we're here: these things keep getting better at the part I used to bill for. Deciding what should exist in the first place? No movement on that at all.

Being able to make anything only counts for something if you know what's worth making. That was always the job. Just easier to hide behind execution back when execution took three weeks.