And here's a link to the website in question:
It appears to this writer that using the power of digital analytic tools can be a very good idea for a publisher (but what do I know, eh?) whereas it might be a little less useful for a writer. After all, one needs a lot of data to analyse and arguably only a publisher will have enough data in a back list of books to be able to discern trends. A writer is a little mired in actually writing the stuff.
But once the trends have been identified, could it really be possible for publishers to commission books made to suit a winning formula? And what will be the result? Identikit vanilla pap (as the reactions to the Publishers' Weekly article would suggest) or, as hoped by the website creators, genuine best-sellers?
So, what will it be? Opinions, please!
And if the tools are that good at predicting success, how soon before we have an entirely computer generated novel? Could we be witnessing the awakening of Frankenbook, or the dawn of a new Digital Age?
Is there much difference from what already exists, on the other hand? Books are often written to suit a formula. Romances, police procedurals, westerns ... all may fall into a "formula". Similarly, these formulae are exploited by publishers that specialise in one genre or another. Data collection as to what is a winning trend is not new either. It's what agents and editors already do. And all one has to do is read the comments left on sources such as Amazon or Goodreads to have enough data for analysis. But this could be a slightly more precise method, a little more scientific. It will be interesting to see what happens with this new development.