I'm being a little presumptuous with this post. My book, Computing for Ordinary Mortals, won't appear until the fall. And it's my first book. So I might end up retitling this post "How to Write a Popular Science Book that Nobody Reads," or (the happier but less likely case) "The Secret to Writing a Popular Science Best Seller." We'll see.
Here are a few things I kept in mind as I was writing.
Choose your audience. When I sent the completed manuscript (or so I thought) to my editor, she said, "You need an introduction to tell people what they're getting into." Won't readers find out by reading the book? Sure, in principle, but you want to set up the right expectations. You've probably come across one-star Amazon reviews that say, "I picked up this book thinking it was about this, but it was really about that!" or "This book would be okay for beginners, but it's pretty useless for experts," or "The author didn't explain enough." That's part of what an introduction should do--tell readers about where the book is going, and explain who might enjoy the journey. I'd had an audience in mind the entire time, but I needed to tell potential readers thatthey are part of that audience.
Find a topic with enough room for a new book. Maybe the world needs another book on string theory or undirected evolution or cognitive science and design, but Brian Greene, Richard Dawkins, and Don Norman have set a very high bar for books on these topics. Mark out your own territory. Ideally, you want someone to say, "If you read only one book on this topic, this is that book." When I started writing my book, in 2009, I found almost no competition on the popular science shelves.
But it's now 2012... Here are a few titles that have been released in the meantime:
Yikes! And they're all good, readable books. Fortunately, they're not my book. A couple are for younger readers (or adults like me who enjoy good children's literature); a few frame a selection of ideas in the context of historical and current progress; some focus on specific areas of computing; some balance the ideas of computing with the practicalities. There's enough room, I think, for a book that tries to pull together the grandest, most important concepts in the field, to produce an understandable big picture.
Try to write well enough to carry the reader through the difficult bits. My favorite popular science books are written by scientists, but many great books have been written by journalists. Think of James Gleick. Or Bill Bryson; his early books are about travel and the English language. Obviously, there are two words in "popular science", and they're equally important. Writing popular science is partly about getting the science right, but it's also about making it understandable and enjoyable to read.I think of a Washington Post blurb for one of John Barth's collections: "...the quick march of his verbs..." And George Orwell's rewriting of Ecclesiastes:
Objective considerations of contemporary phenomena compel the conclusion that success or failure in competitive activities exhibits no tendency to be commensurate with innate capacity, but that a considerable element of the unpredictable must invariably be taken into account.
Orwell was writing about political language, but typical academic language has many of the same flaws. Much of science (especially computer science) is about abstraction, and yet the language describing it should be concrete and vivid.
Get the science right. This is hard, for what may be a surprising reason. It's relatively easy to make sure that you get the facts right (even though I've discovered that my memory has played me false in some cases). The hard thing is ensuring that you convey the right intuitions when you explain complex ideas, even if you can't afford to go into all the details. For contrast, one of my friends writes (slightly tongue in cheek) that the ideal reader of his latest book has "a solid grasp of the foundations and development of mathematics, classical and relativistic rational mechanics, quantum theory, logic, probability, the theory of computation, psychology, artificial intelligence, and mathematical, psychological, and political economics." A popular science book needs to be somewhat more accommodating of the wide range of knowledge and experience of its readers. It's not always easy to make sure that your high-level overview isn't leaving out something important.
Be ambitious. I mean "ambitious" in a specific sense. The best popular science books introduce readers to new ideas, and if some of those readers are scientists themselves, the ideas may change the way they think about what they do. Valentino Braitenberg's Vehicles: Experiments in Synthetic Psychology, for example, sparked research on the emergence of interactive intelligence from the low-level behavior of vehicles; James Gibson's The Ecological Approach to Visual Perception established the field of ecological psychology; Richard Dawkins's concept of memes gave rise to a new (if controversial) field of study. My book isn't really ambitious in this way, except perhaps for an argument I make only implicitly: This is what everyone needs to know about computing. That's probably worth talking about.
Realize that you haven't written the last word. In talking with other people in my field, I realized that if they'd started out with the same goals that I did, each would have written a different book. We all make judgments about what's interesting and important, and opinions differ.
Of course, I should generalize this last point--I've just described my own thoughts about writing. I'm sure there are other ways of going about it.