The machine's first draft is a particular comfort. It is there before you have started, already shaped like the finished thing. You read it and fix the weak sentence, and it goes out under your name. You did real work.
What you did not do was make it.
You have already run this experiment on yourself. Write a page by hand, a full one, and feel it go: the letters wander, the spelling turns strange, the hand aching by the second paragraph. Nothing dramatic happened. You typed for twenty years, and handwriting is a skill, and a skill does not keep.
Nobody warned you, because there was nothing to warn about: the keyboard was better for almost everything. The same trade is now open on the rest of the craft, and this time what is on the table is the part that notices.
Why we hand it over
Handing effort to a tool is one of the oldest human moves. The shopping list, so you need not remember. The calculator, so you need not carry the sum. The satnav, so you need not hold the city in your head. And the keyboard, so you need not form the letters. The name for the move is cognitive offloading, and it is mostly a good deal.
Each handover is rational on its own. The tool is faster and the deadline is close. The result is the same either way. Nobody decides to lose a skill. They decide that today is not the day to do it by hand, and the days add up.
The catch is what rides along. You mean to hand over the task. The practice goes with it.
What the making was keeping
The work the machine now touches was always two jobs bundled as one. Write the letter, and hear where it fails. Map the path, and see where it breaks.
The bundle was not an accident. Making was how the judging got built. You learn what a bad sentence is by writing a few hundred and feeling each one fail. You learn where paths break by drawing them wrong. Effortful repetition, with real feedback, builds a practised eye. Nothing else maintains one.
The machine has unbundled them. It makes. You judge. On most days that is a fair division of labour, and the judging is real work.
But judging exercises less than making. The draft in front of you sets the line you adjust from, and you do not adjust far. You attend to what it got obviously wrong, and skim what it got plausibly wrong, because nothing on the surface flags it.
Making walked you through every choice in the piece. Approving shows you only the ones the draft happens to raise.
The fade
Aviation has a name for what happens next: skill fade. Autopilot flies most of the flight, and the manual skills of the crew measurably dull between the moments that need them. So airlines make pilots hand‑fly on purpose. The autopilot is not the problem. The person is the backup, and a backup that never runs is not one.
The old warning about automation said the person left minding the machine has nothing to do until the moment everything depends on them. The worse half of the warning is less quoted. By the time that moment comes, the minding has worn the person down to someone who can no longer do the thing they were kept there for.
A skill fades slowly, and out of sight. There is no day you feel it go. Knowledge holds, and so does effort. What goes is the number of things you would have caught.
Attention atrophies from disuse. And it comes out of the most reasonable arrangement in the world: a good draft, ready first, every time, until the making that kept you sharp has stopped happening without anyone deciding it should.
The designer who used to rewrite
Content design is where it shows first, because content design is where the drafts are best.
Take a designer who used to write the hard letters from nothing. The ones that tell someone a claim has been refused, or that the support they counted on will not come. Built line by line, each sentence tested against the person reading it on the worst morning of their month. Slow work, and the slowness was where the ear stayed sharp.
Now a draft is open before the brief has been read to the end. The edits are real. The letters are good, and the team ships more of them than it ever has. Every measure a manager keeps says the work has never gone better.
Then one line goes out that reads clean and lands colder than anyone meant. It is caught a week late, on a read‑back, by an ear that cannot say why it let it through the first time.
The output goes up. What falls shows on no chart, until the day you reach for it and it is not there.
Nothing on the project could have caught the change, because the cost is paid in a currency the charts do not carry. You can watch a deadline slip. An ear for language thins where nothing is watching.
There is a lazy version of this argument, and it deserves to lose. It says hand nothing over and write everything by hand, out of respect for the craft. That is sentiment. A great deal of what we made by hand needed no person to make it, and handing it over is a gift. The hundredth variation of a standard notice built nobody's ear. It used one up.
The real line is narrower and harder to hold. Some of the making was practice and some of it was toil, and the two arrive in the same envelope. Hand the envelope over unopened and you lose both at once, and you will not feel which one you lost until much later.
Keeping the friction
The discipline, then: take the help, and keep on purpose some of the friction the help removes for free.
The simplest version is also the most irritating. Make your own version before you read the machine's. The rough one. Ten minutes is enough, and worse than the model's is fine. Write the letter, or sketch the path, or just name the three things the page has to do, before you let the draft set the agenda.
Having made something, you read the machine's draft as someone who has met the problem. You know where the hard part was, because you were just there, and you notice when the draft glides past it. The rough version does a different job from the deliverable. It is the exercise.
Those ten minutes are exactly what the tool was bought to save, which is what makes them hard to defend. Nothing in any schedule will ask you to spend them. The fuller the machine makes the day look, the more they look like the thing to cut, and the more they are the thing to keep.
And once in a while, make the whole thing from nothing, the way you used to, just to find out whether you still can, while the answer is still yes. A skill you never reach for is not held in reserve. It is being spent.
Your handwriting went because nothing asked you to write for twenty years. Nothing is going to ask you to keep the rest in hand either. The work will look finished either way. This one you keep on purpose, or you find out on a read‑back that it has gone.