In the perennial argument about whether you do research before, during or after you write a novel, one answer is that you do it when the children are away, the hangover's wearing off, and the bailiffs are out of the house: in other words, whenever you can. But given the choice, I found myself saying the other day on a forum thread, there are some kinds of research you have to do first, so that you've got something 'to start thinking against'.
Something to start thinking against. It was one of those ideas I didn't know I had till I saw it on the screen. So much of what non-writers ask us about research is about 'getting it right', about correct facts and authentic detail. That's where non-writers assume the challenge lies, especially when they read historical fiction, and boy, are they (am I!) critical if they spot an error. I'm told the naval history buffs are the most terrifying. Of course incorrect facts matter if they'll trip a reader up and shake their faith in you as a storyteller. That must include subtle facts of manners and morals, which is usually where weak writers get it wrong, as well as convict ship journey times and period underwear. But, as Rose Tremain says, you have to leave the research behind.
Storytelling works by taking elements of the real world and spinning them into something else: into a new pattern that the listeners haven't seen before. Even ancient myths grow and change as they're handed down, accreting wheels and city-states from the world of later generations. The only difference between writing fiction and telling other sorts of stories - legend, history, biography, politics - is that fiction is acknowledged never actually to have happened.
If people reading my fiction want history, they can go and read a history book. If they want to know the truth about hilariously drunk twenty-somethings trying to stumble into the arms of Mr Right, they'd do better to leave the bookshop and go and hang out in Covent Garden on a Friday evening. Our stories are stories because they're not simply these facts. 'What if... instead?' we ask ourselves. 'How would it have been if...?' The facts are a solid mass, and we have to think against them, break them up, rearrange them, add things and take others away.
But, unspoken, there's a pact between writers and readers about what you can and can't invent: in most genres you can invent a small African country, but not move the continent to be north of Russia. Which facts you can chip off the mass and rearrange or discard varies from genre to genre, and along the spectrum from literary to commercial. But some facts are like cliff-faces: unavoidable, indestructible. You can create mythical cities and fabulous beasts, even self-willed luggage with legs, but as a writer you can't breach the laws of aerodynamics, or the evolutionary biology which dictates that reptiles have pentadactyl limbs. In other words, even your dragons have to be believable.