I can imagine someone writing a new classic novel and they turn in their first draft of their next draft to their publisher, and their publisher says something like, "This sentence structure...studies have shown that it's a bit too complicated for most readers to parse on the first read, and they can take 3 or 4 times reading it before they understand what you were trying to say. Try to simplify it or break it up into multiple sentences."
That's a goal of dumbing down books to get stupid people to understand them. That's why people cite 1984. Orwell's newspeech is also about dumbing down intellectual discourse.
Is a book provides the reader an intellectual challenge that's not supposed to be a bad thing. Authors of serious fictions do have a license to provide their readers an intellectual challenge.
I don't have a problem with an author making a stylistic choices to use simple language, but I wouldn't want to create a complex system that enforces a dumbing down of literature.
That's a rather mild example of how this sort of data would come into play, but I guess the examples I think of are less, 'Shelf full of Twilight novels' and more 'Same variety of books we have now, written with structure that's more in tune with how people read and think.'
That depends on the quality of the computer algorithm. If you have an algorithm that can tell a publisher the percentage that a specific book is going to be the next Twilight, that publishers might start making publishing decisions based on that number.
There are monetary pressures.
In the New York Times journalists have to suddenly care about readers interests via hard numbers. If you want to read about what damage that dynamic did to journalism read Ryan Holiday's "Trust me, I"m lying".
On HackerNews, this article was linked. The general idea is that companies are studying what people like to read, to help authors produce books that people like to read.
Now, for me, when I look at this idea, I see some down sides, but I certainly see some benefits as well.
Almost none of the commenters on NYTimes seemed to see any benefit whatsoever to studying reader behaviour. There were a few who saw the downsides as more mild than the other commenters. But most of the commenters basically saw this technology as some sort of 1984-esque idea that will turn all books into uninteresting, unimaginative pieces of paper that would better serve as a door stopper than as something for literary consumption. Out of 50 comments that I've read, only one person has said something along the lines of, 'This technology can possibly offer something to help authors improve their books'.
Is this just technophobia? Or am I missing something, and this really is a horrible, evil technology that should be avoided at all costs? [That's a rhetorical question -- I'd be surprised if even one LWian held that position]
I guess what I'm asking is, what are the psychological roots for the almost-unanimous aversion to this attempt at gathering and using information about what people want?