The other day, the Wall Street Journal reported on the various ways eBook providers were gathering information on reader behavior. This is not really a surprise but as I have noted before, the notion that publishers may learn more about what readers want has generated undue alarm in some circles. This latest discussion is on precisely how publishers might take new information on reader behavior and change their practices.
Here is the starting premise:
In the past, publishers and authors had no way of knowing what happens when a reader sits down with a book. Does the reader quit after three pages, or finish it in a single sitting? Do most readers skip over the introduction, or read it closely, underlining passages and scrawling notes in the margins?
Think about that for a second. “No way of knowing.” So when we had physical books, publishers had no ability to find out what readers did? This is surely an extraordinarily wrong statement. Now, to be sure, publishers may not have known but that is different from not having ways of finding out. For instance, they may have asked readers or conducted something we like to term research. It is not outside of the realm of belief that publishers could have worked out whether people read introductions or not and when they might quit reading.
To be sure, digitisation makes that process potentially more cost effective but I think we are kidding ourselves when it is stated …
… that nonfiction books tend to be read in fits and starts, while novels are generally read straight through, and that nonfiction books, particularly long ones, tend to get dropped earlier. Science-fiction, romance and crime-fiction fans often read more books more quickly than readers of literary fiction do, and finish most of the books they start. Readers of literary fiction quit books more often and tend skip around between books.
If this is news for publishers then this is deeply troubling because it isn’t news to readers.
But more might be learned from readers using analytics for eBooks. For instance, when those books are presented in ways consumers might make choices (the WSJ article discusses instances where readers can choose where a plot goes), they learn what consumers might like. They can also put out multiple versions of a book and see which ones get read the whole way through. Notice, however, that this is well beyond taking analytics from book purchasers. Those purchasers are already biased: they bought the book. In order to truly learn what readers want you have to run experiments with a sizeable random element to them. But that isn’t spying on everyone, that is having a set of focus groups; with very different implications for privacy.
What is more, is it really the goal of publishers to make books readable the whole way through? To be sure, that makes sense but I can imagine situations where books that drive you to commit more of your attention — say, by being more addictive — aren’t ones you want to buy. Anyone who has read a Dan Brown novel knows you are going to be played and will find it hard to put down. But are you always going to pay more for books you can’t put down? Who has the time for that?
The subtle thing is not to learn how people read books but how people spend their time. The analytics coming from devices will be able to build up a bigger picture of the allocation of individual attention. Book publishers, for instance, may not care at all if someone buys a book but only gets part way in before purchasing yet another book a reading half of that. Thus, the main danger is that they get a ton of information about some aspects of reader behavior without really thinking about what behavior they want.