One for the Books: A Good Book is Hard to Find

Tracking down a good book can be a challenge.

Most avid readers have a handful of reliable resources—certain best-seller lists, respected publications, and reviewers who have a history of sharing their personal tastes. Such methods, however, have their drawbacks, the most prevalent of which is predictability.

Currently, the top two books on the New York Times’ paperback nonfiction list are by Barack Obama and two of the top five paperback fiction best-sellers, The Reader and Revolutionary Road, have been made into Oscar-nominated films. This is not to say that such books aren’t worthwhile: if film adaptations and, heck, even historic elections are necessary to alert us to the presence of good literature, so be it.

Still, sometimes a bookworm gets an undeniable urge for an undiscovered good book, one that isn’t prominently displayed on the fiction table at every Barnes & Noble or Borders. When this happens, we turn to personal recommendations.

Having worked at an independent bookstore for years, I can attest to the popularity of this approach, but the method is less than scientific. Results are easily skewed by outliers. If the answer to that all-important question—“What’s the last thing you read and liked?”—is unusual, it can lead to some wildly subjective recommendations.

Luckily, a handful of mathematically minded bookworms have decided to confront this challenge, setting up databases that use your reading preferences to provide a list of lesser-known options that might appeal.

I stumbled upon this a few weeks ago in the New York Public Library. Killing time in the Microform Room by browsing the databases, I came across “Fiction Connection,” which promised “titles similar to books you already enjoy.” I was intrigued. I typed the name of a favorite short story author and hit enter.

The bounty! I came away with a list of authors, many of whom I’d never head of before. I felt as though I’d hit the literary jackpot. There must be others, I thought, and thus the quest began.

Fiction Connection doesn’t list its methods, but from what I could tell, it was all about the keywords. Library of Congress subjects, subdivisions of genre, and aspects of plot like time and setting provide connections between books—if titles share several of these, they’re put on the list.

Other resources, like Book Lamp, take an even more objective tack. The site describes itself as a Pandora for books, and it claims to skip the less quantifiable aspects of books in favor of mathematical counts. If you want a descriptive book, Book Lamp searches its database for books with a high density of adjectives. Other categories include tone, tense, perspective, action, and dialogue, as well as user-selected preferences like length.

Book Lamp is, to put it bluntly, a literature geek’s dream. It breaks down the basic elements of writing to deliver a book that should suit you perfectly, with graphs showing each aspect’s distribution over the course of the book.

But I found myself a little disappointed with the site, and so have others. The limited number of books are almost uniformly science fiction, and, as one blogger pointed out, the machine-generated recommendations can be odd: the top choice for fans of George Orwell’s 1984 was the PATRIOT Act, a 98 percent match.

As time passes and Book Lamp progresses beyond the beta release, these kinks will probably be ironed out. In the meantime, resources like Fiction Connection can help readers discover writers that might otherwise have remained in relative obscurity. But I find myself reluctant to state with certainty that books can really be defined by linguistic algorithms. Maybe some things are better left subjective.

Rebecca Evans is a Columbia College junior majoring in English and Creative Writing. One for the Books runs alternate Mondays.