The second draft is where the argument shows up

The first draft finds out what I think. The second finds out whether it holds. These are completely different activities and I spent years trying to do both at once, which is the slowest possible way to write anything.

What the first draft is for

Discovery, and nothing else. I do not know what I think about a subject until I have written four pages about it badly, and no amount of outlining substitutes — the outline is a record of what I already believed, which is precisely the thing worth testing.

So the first draft is allowed to be shapeless, repetitive and wrong. Its only job is to produce a claim I did not have before I started. If it does that, it has worked, however bad the prose is.

A first draft that reads well usually means I wrote down something I already believed. Those are the ones I throw away.

Where the argument shows up

The second draft asks one question of every paragraph: is this load-bearing? Most are not. They are throat-clearing, restatement, or a point I made better two paragraphs down.

What is left after that cut is the actual argument, and it is usually about a third of the length and in a different order. Three things reliably happen.

• The real opening is somewhere on page two. It almost always is.
• At least one paragraph I was proud of turns out to be decorative, and cutting it is the moment the piece gets good.
• A gap appears — a step I had assumed rather than argued — and filling it is the only genuinely new writing in the second draft.

Why not merge them

Because editing while discovering kills the discovery. The sentence that would have become the actual point gets deleted at birth for being badly phrased, and you end up with a clean draft of a thought you had before you sat down.

Two passes is not slower. It is the same work, separated, with the difference that each pass has one job and can be finished.