One key to writing well is to read. A lot. Both bad and good papers have their uses--you can always use the good ones as models or guidelines to your own writing, and the bad ones can be examples of what not to do. I've written mock grant proposals before, but the whole process felt more like being thrown to the four winds with little feedback and even less guidance. If only someone actually offered a class that was solely focused on grant and/or journal article writing!
My formal writing training, to say the least, is a bit spotty. I took several writing classes as an undergraduate but they were for fiction. (And fiction writing, even if it's speculative fiction, is very different from technical writing.) There was one science writing class that I took--to write for the lay person--but my views on that particular class were somewhat mixed. It was good that I had two people to look over my work: a technical writer/editor and a scientist who was an expert in the subject I chose writing in. But once again, guidance on how to write--particularly in a clear style--was a bit hit-or-miss.
I find science blogging to be a self-imposed writing exercise. But even if everyone can read what I'm writing, I have no idea whether or not I'm writing well or perpetuating bad habits. On a blog, feedback for writing style is pretty much nonexistent or unreliable if you rely on blog directories with "ranking" systems.
