Nicholson Baker talks about the way in which the most successful nonfiction books are those that can be boiled down into an argument so that everybody can wade in with an opinion without having to undergo the inconvenience of having to read the book itself. The more you can condense it, the better.

I think, about the distinction between fiction and nonfiction. Fiction is not really about anything: it is what it is. But nonfiction - and you see this particularly with something like the BBC Samuel Johnson Prize for Non-Fiction - nonfiction we define in relation to what it's about. So, Stalingrad by Antony Beevor. It's "about" Stalingrad. Or, here's a book by Claire Tomalin: it's "about" Charles Dickens.

The process of book writing for me is entirely one of trial and error.

Have more than one idea on the go at any one time. If it's a choice between writing a book and doing nothing I will always choose the latter. It's only if I have an idea for two books that I choose one rather than the other. I ­always have to feel that I'm bunking off from something.

One of the reasons so many nonfiction books are so boring is because what they've done, very diligently, is fulfill the terms of their proposals. They've written up their proposal, long-form, and often what this does is then set up a sort of serial deal, where the whole book can essentially be reduced back to the size of the original proposal!

For so long I didn't have any kind of readership at all - I'd get published, but not read - the idea of writing for an audience is so anathema to me, it's never bothered me.

People say it's not what happens in your life that matters, it's what you think happened. But this qualification, obviously, did not go far enough. It was quite possible that the central event of your life could be something that didn't happen, or something you thought didn't happen. Otherwise there'd be no need for fiction, there'd only be memoirs and histories.