In the light of recent events, some people have been asking why publishers don't fact-check books. It's a reasonable question. Earlier this week, Houghton Mifflin Harcourt pulled the book Imagine from the shelves after learning that the author had fabricated some of the quotations.
I want to point something out. The three features I've written for Wired are roughly 4,000 words long. And it took something like two weeks for Wired to fact-check each of them.
Fact-checking is an arduous, labor-intensive process. A good fact-checker will verify every quotation and paraphrase by calling the person who was interviewed. He or she will ask to see the source for every assertion, requiring the writer to provide a reference. My drafts are filled with footnotes citing refereed journal articles, interview notes, and reputable websites. The factchecker will examine every single one of them.
For me, it's like doing a doctoral defense combined with a colonoscopy.
But this is why I respect magazines like Wired so much. When I read a Wired story, I know that every single sentence has been examined by at least four people: the editor, the factchecker, the top editor, and the writer.
Science is scary complex, so I'm grateful for the scrutiny. It reduces the chances of my making a mistake. Not everyone may agree with my interpretation of the facts, but I can be reasonably sure the facts are right.
That's what it takes to fact-check a 4,000-word story.
Now consider that a book can be anywhere from 50,000 to 250,000 words long. Fifty thousand words is a short book, maybe 150 pages depending on font size and page layout. A popular-press science book is often about 100,000 words long. (I've written two books, Rebuilt and World Wide Mind. Three, if you count my dissertation.)
With texts that long, the economics are completely different. There is simply no way a publishing house can fully fact-check even a short book. It would need an army of fact-checkers, and the cost would be enormous.
A good editor will flag implausible claims. A copyeditor will flag typos. But detailed, sentence-level scrutiny simply isn't possible.
In the end, the responsibility for the accuracy and integrity of a book lies with one, and only one, person: the author.