I think the most useful point is the first one, which I would reword to say that learning-by-doing is the way to learn to write. Reading is helpful too, and reading in particular with an eye toward the structure of the argument rather than just to the substance.
I would add three other bits:
11. Even if your paper is not as good as the last paper you read by Heckman (or the alternative famous and brilliant scholar of your choice) it can still be a really useful scientific contribution. There are plenty of great papers out there that never see the light of day because their authors compare them to the best one percent of papers and find them wanting. That is not the relevant standard. The relevant standard is: does this paper add to the existing body of knowledge?
12. I find that giving talks is really helpful in writing papers in the sense that I find it easier and more natural to develop the "story" aspect of papers in the context of a seminar presentation than when writing. Once I have the story, the writing is much easier.
13. Blogging is really good writing practice.