Hey, there's a lot of middle ground between pre-planning a book (by outline, synopsis, or summary) and the complete random flakiness of a divining rod--I think I'm detecting just a little bit of a pre-planning bias here! :-)
I don't plan my novels at all--I write them. I'll have an idea, and maybe an opening (and maybe I'll rewrite the opening a few dozen times--or not), and then I'll go in and write the book. Which is what I did with the book I finished a few months ago: I jumped in, wrote a draft (and found out where the underlying tensions of my stories lay), threw much of it away (but kept the tensions), wrote a second draft that was closer, and a third draft that was closer than that, and so on and on until, by a series of successive approximations, I got a book that worked for me.
I once broke down my messy drafts process something like this:
Draft 1: Write the wrong story. But sort of kind of get a feel for what the right story is about. (The exploratory draft.)
Draft 2: Write the right story. But with all the wrong words, muddled arcs, and not enough sensory vividness.
Draft 3: Get something approaching the right words. Only with lots of the wrong words still mixed in.
Draft 4: More right words. Not so many wrong words. Much tighter.
Draft 5: Polish until my teeth hurt.
Any one of those stages can get repeated multiple times, though I think for my recently-finished project it really was five drafts, which is pretty much a bare minimum for me. Between the first and second drafts I took a research trip, so the changes between drafts were pretty significant--though they usually are anyway, with at least three quarters of the first draft changing completely these days.
That first exploratory draft is crucial, though. It's where I find the story, and for my process, the final book would be much weaker--if it managed to happen at all--without it.