The Long Grind of Writing a Book
Posted by Tom Leinster
I’m using the quiet of Christmas to finish writing a book, Basic Category Theory. It’s nothing revolutionary: just a short introduction to the subject, based on courses I’ve taught. But the process of book-writing is demanding and maddening enough that I wanted to take a moment to reflect on why that is — and why you hear authors complain so much.
Put another way, I’m taking a break from the tedium of writing a book to write about the tedium of writing a book. I hope it’s not tedious.
I want to try to articulate why writing a book is so much more painful than writing a paper. This isn’t something I’ve thought through; I’m typing this off the top of my head. But I’ll see if I can gather some reasons.
First, let me confess that I’ve been surprised by just how demanding it’s been. I’ve written one book before, and that was an extremely stressful experience. But this one is a venture of a completely different kind: it’s half the length, it’s a textbook rather than research (and therefore not nearly so personal), I already had what I thought was a nearly-final draft when I approached the publishers this time, and I’d been polishing the notes up, on and off, for the previous twelve years. What’s more, I’m older and, I hope, more able to cope with stress: just as carpenters get calloused hands that make them insensitive to small abrasions, I like to imagine that academics get calloused minds that allow them not to be bothered by small stresses and strains.
So, I went into this aware of the potential stress. I think I successfully removed just about all of it. But what I hadn’t bargained for is that when you remove all the stress, what’s left underneath is… boredom!
Let me qualify that. When you’re writing just about anything, there’s an intensely satisfying period when it’s all coming together. That’s great. But after that, towards the end — and that’s the stage I’m at now — there’s an awful lot of grind.
Let’s start with the obvious. A paper is long if it’s 50 pages; a book is short if it’s 200. But the crucial thing is that the pain does not scale linearly.
For instance, you have to check that you’ve used notation consistently throughout. The time it takes to check this for each piece of notation is proportional to the length of the book — but so too is the number of pieces of notation. So, the time needed to check consistency of notation is proportional to the square of the length.
You also want to make sure you haven’t repeated yourself. (In an earlier draft, I told/reminded the reader what the discrete topology was three separate times.) This amounts to checking that line is not too similar to line for all lines and , which again means that the time you need is proportional to the square of the length.
What’s more, a book feels different from a paper. Books tend to get more publicity, and people engage with them in a different way. My experience is that if I tell a mathematical friend that I’m writing a book, they’re pretty interested; but if I solemnly informed them that I was writing a paper, they’d look at me like I’d told them I had two legs. (I do.) So when you’re writing a book, you know that what you’re doing is likely to come under more scrutiny. This will bring out all the perfectionism in you.
In my case, this was a big effect. I’ve taught master’s-level category theory courses several times, and I had a polished set of Latexed notes for them. I thought it wouldn’t take much effort to turn them into a book. What I hadn’t realized is the extent to which I was speaking to my students — not particular individuals, but the generic Glasgow master’s student that the notes were addressed to. To adapt them for an unknown anyone-in-the-world reader, I’ve needed to examine and undo a lot of assumptions, and that’s taken a lot of work.
On top of all this, there are things you need to do for a book that you don’t need to do for a paper. One of them is indexing, which has to be up there among the most boring tasks in academic life. I actually have a book on indexing (excerpted from the Chicago Manual of Style), which I bought when I was indexing my last book, in Chicago. It’s 65 pages long — and yes, it has an index.
Anyway, I’m happy to say that I’ve very nearly finished. My deadline is 31 December. Although these deadlines seem to be almost infinitely elastic, I intend to meet this one. After that, there will be a whole lot of to-ing and fro-ing with Cambridge University Press, who are publishing it, and I hope it will be on the shelves some time in the middle of 2014.
Merry Christmas!
Re: The Long Grind of Writing a Book
Regarding consistency of notation, what I have found helpful is using latex macros. For example, when writing about a random process
X^\epsilon
I defined\newcommand{\resamplede}{X^{\epsilon}}
so that I didn’t have to remember whether the epsilon was sub- or super-script. You can also use this technique for more complicated examples.