Marissa Lingen writes up her experience of writing fiction and writing non-fiction, and how the two are different (and, also and importantly, how the two are alike). It’s an interesting read and you should check it out.
For me, the major difference between my writing fiction and writing non-fiction is that with non-fiction I generally write from structural outlines (i.e., I know what every chapter is supposed to be about, and within each chapter, what I’m writing and what needs to be addressed), whereas with fiction I tend to wing it and make it up as I go along. Which is to say the fiction writing process is inherently more creative than they non-fiction writing process. But this is not to say my process for fiction is better than my process for non-fiction; both are excellently suited (for me, anyway) to the goals of the writing.
There’s also something Mris says in the entry that I agree with even though it’s not technically correct for my own writing path. She writes: “If you haven’t written a lot of fiction, you probably can’t write good fiction right off the bat.” I think this is generally true because generally speaking no one is good at anything without putting in a considerable amount of time at it.
That said, when I wrote Agent to the Stars in ’97, I had written almost no fiction at all once I got out of high school, the exception being a three stories in college (one actually started back in high school), and a couple of three-page aborted attempts at novels in the early 90s. So Agent was my first completed work of fiction begun since 1987. The reason I think I managed it was a) I had a job as a film critic, so I spent several years evaluating other people’s story structures (and dialog, and everything else), b) I was writing every day for a living, c) I had worked as an editor, talking people through the potholes in their own work. All that compensated for not actually writing much fiction first.
On balance, however, I think it’s easier for most people just to write a bunch of fiction and get up to speed that way. Which goes to Mris’ point.