I was honored today with a lengthy response to a recent Shatzkin Files post on the Digital Book World blog from Neil Balthasar, who apparently uses techniques similar to those in a forthcoming book “The Bestseller Code: Anatomy of a Blockbuster Novel”. My post had been a response to a PW article announcing the upcoming publication of that book. I reacted strongly to this sentence near the top of the story:
“In the forthcoming The Bestseller Code: Anatomy of The Blockbuster Novel (St. Martin’s Press, Sept. 20), authors Jodie Archer and Matthew L. Jockers claim they created an algorithm that identifies the literary elements that guarantee a book a spot on the bestseller lists.”
It was the “guarantee a book a spot on the bestseller lists” that got my juices flowing.
In his response to me, Balthasar moves the goalposts completely. To him, textual analysis is just one of a number of inputs he uses in his company (called Intellogo, an entity with which I am not familiar at all.) In fact, he says “The Bestseller Code” does not claim what the sentence I quoted above clearly does claim. And then he goes on to suggest that my post suggests a lack of appreciation for “machine learning” and “big data” and succumbs meekly to the “romance of publishing”.
It seems pretty clear that he doesn’t know much about Pete McCarthy or me, nor is he much aware that we have spent our careers arguing to the romantics in publishing that they need to be more data-centric.
Balthasar claims an overlap in our viewpoints but creates a total straw horse by saying “…I agree in theory with Shatzkin that an algorithm alone cannot predict whether a book will be a bestseller or not, that isn’t precisely what The Bestseller Code claims, nor what our experience working with machine learning at Intellogo defines”. (Of course, it is precisely what the sentence quoted from the PW story does claim for the book!)
While we apparently agree that big data is an essential analytical tool for publishers marketing books today, where we emphatically part company is on the relative importance of textual analysis. Compared to research into the audience, segmenting it, understanding its search behaviors and social activities, and understanding the competitive environment for a book at the moment that it is published, the analysis of the book’s content adds very little, even when it is deeply analyzed and bounced against other sources.
Or, let’s put it this way. We do lots of projects designing digital strategies for books without performing textual analysis. Maybe some of those plans would be improved if we also used a book’s text as seed data for portions of our analysis. But there’s no way we’d try doing any meaningful marketing planning without the other things we do, no matter how rigorous or skilled a textual analysis was.
I’m glad a fan of “The Bestseller Code” is moved to put the textual analysis in the category of “among the things we do” rather than “we can predict a bestseller from the text”. But that wasn’t the proposition I was reacting to when I wrote the post that provoked his response.
When Balthasar says (as he does), “Imagine a day when we take all our data about what people are reading and provide publishers (and authors) ideas of what people want to read, where to find those audiences, and better ways to reach them”, he is pretty much stating the nature of our work at Logical Marketing. We do precisely what he’s suggesting today for a wide variety of clients. Textual analysis has almost nothing to do with it.