Writing by numbers

Writers spend so much time dealing with words that we often forget how useful numbers can be. Every book consists of some flexible sequence of individual chapters or scenes, and once this kind of framework exists, it can be broken down numerically in ways that often shed a surprising light on the story’s structure. I’ve quoted the young George Lucas on this point before, but I may as well do it again:

In the end I want to end up with a list of scenes. And the way I work generally is I figure a code, a general measuring stick parameter. I can either come up with thirty scenes or sixty scenes depending on which scale you want to work on. A thirty scene thing means that each scene is going to be around four pages long. A sixty one means that every scene is going to run twenty pages long…I have a tendency to work rather mathematically about all this stuff. I found it easier and it does lay things out.

In the most extreme case, screenwriters can turn into obsessive scene- and page-counters, to the point where certain screenwriting guides advise you to put the inciting incident on page 10, the “all is lost” moment on page 75, and the second turning point on page 90. And although this kind of quantitative approach may seem to have little place in writing a novel, there are times when it helps to reduce a book to the numbers, in order to guide or supplement a more intuitive approach.

As I noted yesterday, when I realized that my rough draft of Eternal Empire was about 15,000 words too long, I began by cutting it like a sculptor, looking at each page and crossing out paragraphs that seemed to break the visual rhythm of the scene, while also trimming chapters that felt too long when I was flipping through the manuscript. This top-down, predominantly intuitive method got me partway there, but in the end, I still had a lot of material to cut if I wanted to get the draft anywhere near 100,000 words. As any writer can attest, once you’ve passed a certain point in cutting, it gets much harder—the extraneous material isn’t standing there in plain sight, but is buried within otherwise tight-looking paragraphs in the form of an extra sentence, an unnecessary line of dialogue, or a clause that only repeats information that the reader already knows. To get the length down any further, I had to cut this book to the bone. And the obvious way to do this was to focus on chapters that were objectively on the long side.

I responded with what I often do when faced with a problem like this: I made a spreadsheet. I began by listing each chapter, its point of view character, its length, and whether or not it was predominantly transitional. (A transitional chapter, roughly speaking, provides a moment of relative downtime between the more intense moments and set pieces. Such quiet spots are essential, because a book that was nothing but high points would quickly become exhausting, but they also need to advance the story in some way, and should probably be kept on the short side.) Looking at the ensuing list, I saw that most of the major chapters were about 2,000 words long, while transitional chapters tended to be closer to 1,500. This, then, was my benchmark: any chapter that was longer than its characteristic length would have to be cut down. This could mean cutting fifty words or several hundred, but the result would be a draft in which no chapter was longer than its narrative peers without reason. And my spreadsheet gave me a good sense of where to start.

If this sounds crazy, well, maybe it is. (I’m often reminded of Daniel Okrent in the documentary Wordplay, in which he explains that he keeps a notebook recording all of his solving times for the New York Times crossword puzzle “because I’m an obsessive creep.”) And I would never recommend this approach to anyone whose mind didn’t lend itself to these kinds of solutions—I like putting together spreadsheets, and naturally tend to use them to organize projects. But there’s a larger point here that I think applies to every writer. It may seem ridiculous to worry about whether a chapter is 1,750 words long or only 1,500, but this approach, in general, will draw your attention to parts of the story that are relatively more likely to be in need of cutting. And any cut you can make will almost always be a good one. The real test, obviously, lies in reading the novel itself, in which an objectively short chapter may drag and a lengthy chapter may seem just right. But looking at the numbers will guide the scalpel to places that might not be obvious to the naked eye. In the end, I cut several thousand more words out of the draft, and I don’t even know where they went. The result was a much stronger novel. And I can thank the numbers for this.