What Lives Inside Second Draft Notes

I use AI a lot.

It gives me speed and distance from my own thinking. Sometimes it finds a sentence I would eventually have reached myself, only much faster.

For a while, I thought the best way to describe the experience was this: using AI felt like coming home after someone else had cleaned the house.

The house was unmistakably mine, simply more orderly than I had left it. But the chair was in another place, and a blanket I always left loose had been folded.

Nothing was wrong.

That was what made it interesting.

Sometimes the new arrangement was better. Sometimes I would look at a sentence AI had changed and think: yes, that is clearer. Keep it.

Other times, I could not quite explain what bothered me. The sentence was polished and said exactly what I meant, yet something in it felt displaced.

And still, I wanted to move the chair back.

At first, I treated this as a problem of AI use.

Perhaps I was letting the machine do too much. So I used it less. I rewrote more myself and became more deliberate about what I asked it to do.

It helped, but it did not solve the problem, because AI was not the only thing moving the furniture.

I had been doing that long before generative AI arrived.

A professional sentence carries assumptions about what a professional person should sound like; an argument can start anticipating objections before anyone makes them. Even a bio may begin accommodating a reader who has not appeared yet.

Sometimes an old room remains inside the writing years after you have left it. Teachers, former colleagues, whole industries with their preferred vocabulary, even people whose approval once mattered can still be in there.

And so can older versions of us.

The person who learned how to succeed in one environment may continue editing the work long after the environment itself has changed.

None of these influences makes the writing false — there is no untouched original voice underneath to protect. Language itself arrives inherited, as do many of the forms through which we learn what success should look like and who we are allowed to become.

Human beings have always been shaped from outside themselves. AI makes the process unusually visible because we can watch another intelligence enter the sentence in real time.

It did not invent the condition.

Once I understood that, a different question became more useful than whether a sentence was purely mine.

Do I recognise what is happening here, and do I want to keep it?

That question changed the way I worked.

I began looking outside writing for better ways to see what was happening inside it.

Old stories were useful for this, not because mythology contains secret answers, but because human beings have spent a very long time making forms for experiences that are difficult to name while we are living through them.

A myth can give shape to a conflict before we have language for our own version of it.

An archetype can make a familiar behaviour look strange enough to examine.

Sometimes the useful lens came from something much less grand, perhaps a conversation or the bodily resistance I felt to a phrase I had written several times without noticing.

The point was never to replace one authority with another.

It was to see more.

That became especially clear when I started looking at patterns in individual sentences.

Over-explaining was rarely only about explanation.

Sometimes the writer was protecting herself against an objection that had not been made.

A qualification could be intellectually necessary. It could also be the trace of an imagined audience standing just outside the paragraph.

A perfectly competent sentence could hide its subject so thoroughly that no one appeared responsible for what it said.

Often, the correction was for an accusation no one had made. Sometimes the whole correction was smaller than the panic around it.

This was where writing became interesting to me in a different way: the sentence was no longer merely something to improve.

It was evidence.

It could show what the writer expected from the room, or reveal that authority had moved elsewhere. Sometimes it preserved a relationship, a rule, an older version of the self — long after she believed she had moved on.

Voice, I realised, was not the final subject but one of the places where becoming became visible.

At first I called this protecting my voice. Over time, the larger pattern became harder to miss: I was learning to notice coauthorship.

That is what eventually became Second Draft Notes — not a newsletter about sounding more human in the age of AI, nor a place for sorting influences into good and bad. It begins with a harder question: how much of what is shaping us can we notice while it is happening?

Then there is what we do with what we notice.

Authorship can mean refusing an influence, or recognising that one has already become genuinely ours. At other times the most authorial act may be to postpone the decision.

We are remarkably efficient at replacing one form with another. An identity stops fitting, and before we understand why, we are already trying to name what comes next.

Now we can ask AI to do this work almost instantly, shaping the new thing before we have fully understood it ourselves.

That is useful. It can also close a question before it has had time to become interesting.

The project is organised around eight recurring movements of change, from what has stopped fitting to what has become real enough to recognise.

Then the cycle turns again.

I do not think people move through them neatly.

A person can be finished with one version of herself while still protecting the beginning of another. She can know exactly what she is doing professionally and have no language yet for what is changing elsewhere.

The movements are not a test.

They are different places from which to look.

That distinction matters to me because The Second Draft is not meant to become another system that tells people what the correct next version of themselves should be.

AI should not decide that.

Neither should a framework.

The work is to create enough room to notice what is shaping the draft before automatically carrying it forward.

Which brings me back to the house.

I no longer think authorship means restoring every room to the way it looked before someone else entered.

That would make influence itself the enemy.

Instead, I want to be able to walk through the house after something has changed and actually see it.

The chair has moved.

Perhaps I move it back.

Perhaps I discover that I prefer it here.

I want the decision to become visible.

That is what lives inside Second Draft Notes: the pleasure of noticing what has moved and discovering that the choice is still yours.