My answer got long, but I thought it might interest some people who aren't on the list.
Ooo. Good questions. And for me, the answer is, "It depends."
My process varies a *lot* from book to book. Some of them have a more or less loose outline. Some have a list of signposts. Some have random scenes written out of order in advance. Some are written in more or less strict chronological order, charge ahead, without an outline, with one exception--which I'll detail in a minute. (There are two things that I always or almost always do. Everything else seems malleable.)
My first novel, ALL THE WINDWRACKED STARS, I wrote groping in the dark. It was single-narrator first-POV, and I made a lot of mistakes with it--including not knowing enough about what was going on outside of my protagonist's head. I knew the character very well--I had a bunch of short stories about her, and I knew a lot of things, but not how to hang a story together. So it kind of meandered rather a lot, and it had that typical first novel problem of claustrophilia. No range. So that one was written classically follow-the-headlights.
Live and learn. *g*
The second novel, THE SEA THY MISTRESS, was signposted heavily along the way--I wrote a bunch of scenes in advance, and I knew the conflicts and how they developed. I knew from the beginning where it was going, and it got there, but I discovered as I was writing it that it didn't have a lot of breadth. The whole book fit in my head, which makes for a narrow novel.
(I later partially resolved the problems with these two books by frankenbooking them together, as they concerned the same characters. It's still not quite the book I want it to be, but maye in ten years I'll be able to make something better of it.)
The third novel, BY THE MOUNTAIN BOUND, had a chapter-by-chapter outline (necessary because I had three first-person POV characters, rotating in a strict order--A,B,A,C,A,B,A,C--like that) but it was a pretty vague outline--one sentence per chapter, defining the major plot point in that chapter. It still surprised me a lot, though, and had to be changed and expanded.
I still think this one is a working novel, although my sentence-level work has mproved a lot--but it's not really salable until I get ATWS/TSTM fixed up to my standards.
The major characters in these three books--Muire, Cathoair, Mingan, and Strifbjorn--and the world that they live in all date back to, oh, 1993 or so. So that's a pretty long gestation.
The fourth novel was HAMMERED, which also grew out of characters and ideas I'd had kicking around in my head since the early '90's. This book was bigger in scope than the other three--more characters, more POVs, a big sprawling political plot to tie the various personal plots together. This book, I started rather free-form, and then at about 130 pages--the dreaded 25 K wall!--I stopped dead. Completely dead. Wham!
Finally managed to break out of it by outlining the first third of the book--the part I'd already written--an extensive scene-by-scene outline that listed everything I'd established and all the questions I'd raised. Once I had that list, I set about complicating those issues and eventually resolving them.
The fifth book was BLOOD & IRON, which was then called SHADOWHAND. That one was also very much a written-in-the-headlights book. I felt at the time that I was unearthing a fossil by flashlight. Five major revisions later (including one that added two POVs and 60,000 words), it remains the hardest book I've ever written. (BTMB was the easiest.) The characters and wrold for this book are the oldest--I first came up with some of the ideas and characters behind it in... 1986. Or so. It may in fact be even older than that. Which may be WHY it felt so much like bringing up something buried--there's a hell of a lot of subtext in this book, and deep down emotional underpinnings.
The sixth book was SCARDOWN, which proceeded much like HAMMERED, except I got stuck at the end rather than the beginning. This is the one where I hit on the trick of notecarding--making a list of every outstanding item and question, putting each one on a notecard, and then arranging the notecards in piles until I hit on a logical sequence to get myself out of the hole I'd dug. This is the sequel to HAMMERED--but a much shorter gestation on this part of the story.
Then, numbers seven and eight, the two halves of THE STRATFORD MAN. I got the idea for this book while talking to a colleague of my husband's at a Christmas party in 2002. I wrote the book (it's one book in two overlong volumes) in the second half of 2003. Call it six months gestation. And heavy, heavy, HEAVY research.
All of the other books were research-as-I-go, other than basic foundation reading. SM required two months of reading before I started writing, and by the time I was done with the book, I'd read something like 1000 pages of website printouts and thirty books. Historical fiction is exhausting.
SM had a pretty firm timeline, because it's after the model of a secret history--it takes place around historical events from 1594-1605--so I had lots and lots and lots of notes. And also notecards.
However, the good news is, I got a second novel out of the same research--THE COBBLER'S BOY, written with Sarah Monette, a YA mystery. And also a novelette that sold to a pro market. So, really, time well spent.
TCB was written in a massive rush; we had talked out the plot and we knew the characters, and just drove for it. Action! Spies! Beautifullest Princesses!
So that's nine books, right? Number 10 was WORLDWIRED, which also took extensive reading--though not as much as SM. It is the conclusion of the story arc started in HAMMERED.
Number 11, just finished this month, is ONE-EYED JACK, which grew out of some short story ideas dating from 2002 and 2003. This one also took considerable research, including watching a lot of '60's television. It also occasioned at least one major, major revision (unpersoning some characters, adding a lot of motivation, etc.) This one was both previous-research and concurrent-research, and like SM, the research guided the plot rather a great deal. It also was signposted rather a lot.
Number 12, also written with Sarah, is underway. That's A COMPANION TO WOLVES, which we started last november or October after a livejournal conversation in which we got talking about some of the sillier conventions of Talking Animal Fantasy. This one hasn't taken too much research--after the first three books, and based on my college education, I have a pretty good working knowledge of Norse culture, and I know canines rather well--and Sarah's a whiz at all sorts of stuff.
There has been a lot of refreshing-myself on the Eddas and Sagas, though. *g*
Currently cooking are WHISKEY & WATER, a sequel to BLOOD & IRON, and CARNIVAL, an SF novel that has its gestation oh, about two years ago now and has been accreting bits all along.
Looking at this, it sounds like each novel is a discrete process, but really, they're not. Research for OEJ overlapped both WORLDWIRED and rewrites of B&I, SM, and ATWS, for example. Bits of ACTW were written concurrently with bits of OEJ and WORLDWIRED.
The things I do find applicable to every book, though, are these:
I need to write the first few scenes--ten to 50 pages--of every novel, and then let it sit for a good long while (six months, three years, ten years) and cook. When I'm about a third of the way done, I know how the story ends--not the climax, but the denouement. And if I'm stuck, research, brainstorming, and looking back over what I have already always kicks something loose.
Often there's a scene in a book where I explain the book to myself, which later needs to be cut. It's usually one of the few things that needs to be cut, because I am a habitual under-writer. (King's books shrink 15% on draft--Mine usually grow 15%.) I write scenes as I get them, which means sometimes out of order. I often outline, or at least take notes in advance, but really I discover a book as I go. I hate writing series synops for books not yet written. I like to get to know my characters before I write books, thus the write-a-starter-and-then-let-it-rest trick.
Too much prep work is when it gets in the way of writing the book. Too little is when you don't know what you're talking about. (The advice I got researching SM was, read until the footnotes start referring to each other. It's good advice.)
And wow that got long.