I got this book for free from the publisher, in exchange for an honest review.

It is probably hard to write a book about bestsellers, as obviously everyone reading it would think "Well, you know what sells. You have analyzed all the patterns, you know all the tricks; now do you follow your own advice? Does it help your own book at all?" And I have to say that the authors did a very good job here. With all the obvious caveats (they analyzed works of contemporary fiction, while their book is a non-fiction work with an academic tint) it is quite clear that they have as good grasp of what works, and how it works. They could not give their book a strong female lead, and they could not quite pick a setting for it (again, it being a non-fiction book), but they built chapters so cleverly, and found a language so clear and transparent that I read it in one day. It's the first book I ever read on a digital humanities topic, and still I read it in one gulp. I only wish other academics would follow this example, and took their popular writing as seriously.

In other words, the topic aside, just from the stylistic point of view, it is one of the best-written academic-ish books I've ever read.

But then also the topic is also extremely interesting. It is not the first book any aspiring writer should read, but it can be very useful and thought provoking. At one hand, it endorses some well-known pieces of advice (such as "show don't tell" and "don't ever replace the word 'said' in dialog with anything else), as it proves these maxims with hard data from sales analysis. Good style sells better. At another hand, it offers writers some good topics for contemplate, in terms of some potential trade-offs between sales (catering to a wider audience) and literary aspirations (writing for the audience of your choice). I would guess that about two thirds of advice in this book would just objectively make your writing better. The last one third however could require some sacrifices of accessibility over quality. And that's an interesting topic to think about.

Arguably the best part of the book, from the practical point of view, is the analysis of plot curves it offers. People analyzed plot curves for years (Kurt Vonnegut has this amazing lecture about this very topic, called "The Shapes of Stories"). Yet here the data comes not from somebody's intuitions (even if brilliant), but from a large scale data mining project. It is very illuminating, helpful, and useful.

All in all, the book is a must read for anyone interested in writing, and for anyone who (like me) never read much about digital humanities before. ( )

Fascinating, super cool, readable! Haven't read many bestseller books at all, but am now feeling inspired to check out, for myself, the evidence of the conclusions drawn by Archer and Jockers. ( )

A detailed description of a machine learning algorithm to analyze what makes a bestseller. They looked at themes, emotional plot-line, style, and characters. It includes lists of books that exemplify each of the dimensions as well as a list of the 100 "best" books according to their model.
It is repetitive in spots but overall it is a quick read. ( )

The authors make a good but premature case for their bestseller-recongizing software. There were a few interesting insights on what makes for a bestseller. The program which tracks plot progression and the resulting curves was especially fascinating, as well as the accompanying explanation as to why books like Fifty Shades of Grey are so compelling as emotional roller-coasters, even for people who acknowledge their literary paucity. Too often, however, the "insights" are banal. People like to read about "human closeness"? You don't say. The authors sometimes fail to describe how exactly non-bestsellers fail on these parameters - isn't ever single book ultimately about human closeness?

Another problem is the fact that the whole program is based on the New York Times bestseller list. Would results have looked different for another list or another country? Different books become bestsellers in UK vs. US, for example. The software can also only recognize established patterns, but not really anticipate new ones. What would happen if we used it on 18th or 19th century bestsellers? Could it predict new trends latent in current bestsellers? Sooo many interesting questions and possibilities, yet the authors prefer to stay in the domain of the banal, talking about how it's good to use active verbs. As if we didn't know.

The authors also get carried away in their research and start slipping into a category error of equating "popular" with "good". Bestselling writers are presented as geniuses who know exactly how to tap into the human psyche and to maximize the effectiveness of their writing. While it is true that these writers have a certain knack, the authors never make an actual case for the literary value of bestsellers. Their best argument is that those books keep the market afloat (sad, but true). However, they don't ask whether we really need more bestsellers, whether this will lead to more stagnation and homogeneity, whether it will stifle innovators, etc. What is a bestseller for, and why do we want more of them? is a question given far too little attention (i.e. no attention). The authors are too enamored with the technicality of their project to actually consider the consequences. They want more diverse authors and a new generation of Pattersons/Kings/Steeles, but they have no regard for literary diversity.

The percentage points accorded by the software were also kind of vague in their applicability. The book that got 100% - Dave Egger's The Circle - sold well, but it was no Fifty Shades. The film was also a flop. Obviously getting above a certain percentage gives solid predictions for sales, but the difference between 90% and 100% seems pretty irrelevant. They didn't comment on this.

I am also skeptical of the authors' wholesale repudiation of the notion that marketing and advertisement is what drives sales. They are too fixated and narrow in their discourse, taking their own limited discovery as the sine qua non of all talk on bestsellers. Yet their product hasn't even yet been tested on a bunch of new books arriving at publishing houses, or on other bestseller lists. They seem far too sure of themselves.

Overall, there are a few relevant insights in the The Bestseller Code, but it's a project that needs a lot more work and sophistication before such a publication would be warranted. ( )

Jodie Archer presents the results of her colleague Matthew L. Jockers's data-mining studies of New York Times bestseller list, which are intended to reveal what it takes for a book to make it to the top. They conclude that "bestseller DNA" is all about pacing. The two recent books that best exemplify bestsellerdom, Fifty Shades of Grey and The DaVinci Code turn out to be practically the same in terms of the curves they generate according to Jockers's algorithm.

This is a dull book on an interesting subject. There is a lot of padding. I read the first half and skimmed the rest. ( )