A couple of years ago, Amazon ignited the ire of independent booksellers when it launched its Price Check app, which allowed users to compare prices between the online megastore and whichever shop they were standing in. Rather brutally, it included a launch-day special giving a 5% discount to anyone who bought this way. Many were up in arms, and whatever your opinion of Amazon, it seemed a little callous, even if intended to be used more against supermarket chains than local bookshops.

Since then, many of us have been told off in bookshops for taking out a smartphone, on the assumption we're price-surfing online: finding the book we like in stock, but buying it from Amazon instead. Those of us with access to a good independent bookstore might scoff at this, knowing the real value of a bookshop isn't in its pricing, or even really in its range, but in its curation and the knowledge of its proprietors. Good recommendations are the one thing algorithms can't crack – or can they?

Last week Random House launched BookScout, a Facebook app that recommends users books based on their timelines: the things they like, the things they discuss, the things their friends like. A recent Cambridge University study showed that researchers could predict a Facebook user's gender, race, age, politics and sexual orientation from even a few known "likes". While some might consider choice of reading material to be even more personal than any of these, with enough data a few literary leanings can't be that hard to discern.

What is perhaps most interesting about BookScout is that it doesn't just recommend books from Random House, but from all publishers. This may seem like an obvious step, but it's one that most publishers have missed in the past, concentrating on competing with one another rather than addressing the far greater threat of domination of the industry by tech companies. BookScout is a small, but welcome, redress.