Anyone who's dipped more than a toe in the waters of creative writing knows that much of the craft (and art) of any writing is in re-writing. But even once you've discovered that writing 'The End' is only the beginning, it can be hard to know how to go about that rewriting which we all know is the making of the piece. I know of writers who re-write each page until it's perfect, then never change a comma. I know writers who write scenes from wherever they fancy in book in their head, then stitch it all together at the end, and writers who do the same without knowing even where those scenes come from. I know writers who revise yesterday's work by way of getting back into it, before continuing with today's, and writers who don't re-read a word they've written till the whole first draft is down. A quick dip into a random selection of the Paris Review interviews, either in the anthologies or the absolute goldmine which is the online archive, will show you that, of even the authors who are willing to say how they work, the possibilities are endless.
I know how I work, and why it works for me, and if I'm asked, I'll tell. I've thrashed it out over nineteen years, eight-and-a-half novels, various insights and aphorisms from various writer-teachers, and not a few of the Paris Review kind of interview, and it works for me. So far, anyway: since every new novel springs from a core idea which I don't know if I can pull off, I can't be sure how to set about trying to pull it off, either. So I'd never prescribe to any writer how to go about it, only offer suggestions that I know might help.
But if I were tempted to issue an absolute command to anyone trying to write their first novel, it would be DON'T FIDDLE. Don't keep popping back, changing a word here, a word there, re-reading a little bit after supper and tinkering. Especially after a glass of wine we all love the sound of our own voice on paper (except when we're hating it) and with computers it's too tempting to drop by, read a juicy bit, have second thoughts about a word, a sentence, a character. But it's usually disastrous: you get in a muddle, you don't see through changes you start, you lose track of what you've done because on a screen it always looks perfect, and you lose any sense of the larger structure. You change a metaphor to something better, because you've forgotten that you needed that metaphor much more two pages further on, which is why you changed it here in the first place. You flesh out an encounter in the most fascinating way (maybe because you've just had a similar one yourself?), which screws up the time-scale for the whole second half and two subplots to boot. Even if you don't do anything more radical than correct the odd typo and fiddle with the punctuation, if you keep popping back and reading bits your eye becomes jaded, the text shopworn: it goes dead to you, and you cease to be alive to it.
So, I would say, either leave it alone, or sit down for a solid session (however short) doing a particular job: 'beef up X's character'; 'sort out lost-letter plot'; 'revise Chapter Six'; 'check geography of Manchester chapter'. I would also say, don't forget the advantages of working on hard copy: you can read it sitting somewhere else which helps to bring it up fresh again as does the sight of it on paper; your pen-marks show where you've been and what's old and what's new thinking; you can to-and-fro, but you're less likely to lose track of what you're trying to do at the moment, not to mention the pace and structure; and having sacrificed a twig or two of the planet to print it out, you're more likely to do a thorough job of everything that needs doing, so as not to waste all that paper.
It's also well worth saving up the jobs which don't need full concentration and top-quality brainpower, too. Realistically, there'll be hungover or post-throwing-up-toddler or road-drills-outside mornings. It's maddening to feel that you've got the writing time but not the brain. An hour or two dealing with 'Change McClean to McCrumb' or 'make chapter titles italic' or 'check train times to Lands End' means that morning wasn't wasted after all. And given how bloomin' long novels take to write, don't neglect the stripy-sweater phenomenon. (The friend who taught me to knit told me to get a stripy pattern, because even though it'll take you a year to finish the thing, you can at least say quite often that you've finished a stripe) So, whether the morning's work was indeed changing McClean to McCrumb, or whether it was working out in detail a moment of absolute inspiration about how your novel should end, it's enormously helpful to be able to know that you've ended the morning with something concrete done.
Not that any writing is wasted writing, of course (though that's a whole other blog post). But spending an hour racking your brains for an alternative metaphor - which you know you had nailed a year ago - because the copy-editor points out that you've used the same one twice in two pages, comes as close to wasting writing time as you ever will.