There's a version of writing that's just transcription — you know what you want to say, and the job is getting it onto the page in a readable order. This post isn't about that kind. I'm interested in the other kind: the one where you don't actually know what you think until the sentences are out.
the test
The test is whether, halfway through a draft, you ever catch yourself writing a sentence that surprises you. If the answer is no, you weren't thinking; you were typing. The surprise is the signal that the writing has done work the thinking couldn't do on its own.
I write entirely to find out what I'm thinking, what I'm looking at, what I see and what it means. — Joan Didion
why it works
In your head, an idea can survive indefinitely by being vague. Everything is roughly consistent with everything else because nothing is fully specified. The moment you commit to a sentence, you have to pick. You have to say which part of the idea is load-bearing, and which is a rhetorical flourish you were pretending was an argument.
Most of what I cross out in a first draft isn't wrong, exactly. It's just revealed, by the act of writing it, to be something I didn't actually believe once I saw it on the page. That's the value.
a practical note
Writing-to-think only works if you're willing to throw most of it away. If the cost of a bad paragraph is high — if you're writing into a venue where polish is required up front — you'll unconsciously pre-censor, and the surprise won't happen. This blog is, in part, a low-stakes venue designed to make the cost low enough that the surprise can.