The Second Draft Is Where the Book Begins
Every writer I know can describe their first-draft process. Almost none of them can describe their revision process, which is odd, because that is where the book actually gets made.
I have run the same structural pass on five novels now. It is not elegant and it takes about six weeks. This is the whole thing.
Rule one: no sentences
The single most expensive mistake in revision is polishing prose that is going to be deleted. So the second draft has a hard rule: I am not allowed to change a sentence for its own sake. Not one. If a paragraph is clumsy and the scene survives, the clumsiness stays until the third pass.
This is harder than it sounds. Fixing sentences feels like work and produces the pleasant sensation of progress. It is the writing equivalent of reorganizing your desk.
The reverse outline
I read the draft once, quickly, with a notebook, and I write one line per scene:
- Who wants what.
- What stops them.
- What is different at the end.
- How many pages it took.
Nothing else. No commentary, no ideas, no this is terrible. Four facts.
When the list is done, the shape of the book is visible in a way it never is inside the manuscript. On the last novel this exercise produced eleven consecutive scenes where the third column said “nothing”. Eleven. Ninety pages in which the situation did not change once.
Reading the outline
Three patterns show up almost every time:
- The false start. The real first chapter is chapter three; the first two were me warming up. This is true of every first draft I have ever written and I no longer take it personally.
- The sag. A run of scenes where the want has quietly gone missing, usually because the protagonist has become a tourist in their own book.
- The rushed turn. A major reversal handled in half a page because I was tired when I got there. These are almost always the scenes readers later say they wanted more of.
The cut list, before the fix list
Deletion first. I make a list of everything that goes, total it up, and only then decide what to write. On the last book that was 38,000 words out before a single new word went in.
You cannot see what a novel is missing while it is still full of what it does not need.
The cuts are easier if you keep them. I move everything to a file called offcuts.md and I have never once gone back to it. Knowing I could is enough.
Then, and only then, new material
The new scenes go in to answer specific questions raised by the outline. For each one I write a single index card: what changes here, and why it cannot happen anywhere else. If I cannot answer the second half, the scene is not a scene, it is an explanation looking for a home.
Roughly a third of my new material fails this test and dies on the card, which costs an afternoon instead of a fortnight.
The order of operations, in full
- Read fast, reverse outline, no fixing.
- Diagnose: false start, sag, rushed turn.
- Cut list. Execute it. Total the damage.
- Index cards for new scenes.
- Write the new material badly and quickly — it needs to match the draft’s roughness, not exceed it.
- Read fast again. If the outline now reads cleanly, stop. Structural work is over.
- Only now, sentences.
Step five is counterintuitive and I fought it for years. Polished new scenes stand out against a rough draft like a new brick in an old wall, and you end up making decisions based on the shine rather than the structure.
How you know it worked
The second draft is finished when you can say what the book is about in one sentence, and the sentence is true of the pages rather than of your intentions. Mine was wrong on four novels out of five at the end of draft one, and right by the end of draft two every time.
That is the whole method. It is boring, it is repeatable, and it has never once failed to make the book better — which is more than I can say for inspiration.
