Many years ago, when there were new and second-hand bookshops all the way along London’s Charing Cross Road , there was a memorable bookshop called Joseph Poole. Much of my reading in my late teens and early twenties originate in that shop.
They did sell new books, although I never bought a new book there. Every customer – and they had a steady stream of customers – headed to the back of the shop and down the stairs to the second-hand department. Since that time, I’ve visited many second-hand bookshops, but I’ve never seen one with the turnover of Poole’s Bookshop second-hand department. And there was one feature that made those books irresistible; such a wonderful feature I’ve never understood why other second-hand bookshops didn’t implement it.
Directly opposite the cash desk was a bookcase of some six or so shelves. On those shelves were placed all the recent acquisitions. None of those books was sorted; the only criteria was that they had just been bought by the bookshop. Eventually, the books would be shelved, but during the day the books on those shelves would pile up. Why did I always go to the recent acquisitions shelf first? Because my success rate was higher there. I found more interesting books on those shelves; those were the books other people had been reading. Rather than go to the subject shelves and have to go through hundreds of books without knowing which books might be readable or not, I simply glanced at the recent acquisitions; if it looked interesting, I bought it. It was pure guesswork. I bought books without any idea of the author or title in advance. Some weeks or months later, I would take a bagful of books back to the bookshop (most of them started, and some completely read) and the process would begin again for other customers.
Why did this system work so well? I was trusting more to the fact that other readers had been reading the book, rather than knowing anything about it. And the truth is that most books are just never read; they aren’t the ones I wanted to read. They sat on the subject shelves for months, if not years, carefully filed, but hardly ever accessed. I wanted to read the books others were reading! Fiction, non-fiction, literature, history, sociology, art and architecture, it didn’t matter. And the “success rate”, the proportion of books that turned out to be great reads, was remarkably high. Of course, the books you get right of might be the worst ones; but the sheer numbers of people bringing books in meant there was a high proportion of recently-read content.
Now, extraordinarily, I’ve discovered that the recent acquisitions shelf follows perfectly the principle of the computer cache. From a fascinating book, Algorithms to live by: the Computer Science of Human Decisions, the co-authors Brian Christian and Tom Griffiths explain how the most effective algorithm for a computer cache is to remove items that have been least recently used (LRU). In other words, a computer cache works best when it contains the files you have been using most recently. By analogy, a bookshop that displays the books that have been read most recently will be displaying the books that readers are most interested in.
The most exciting thing about the algorithms book is the way it applies principles of computer science – sorting, searching, caching – to everyday activities. That in itself reveals lots of surprising discoveries, but I never for a moment believed that a bookshop could be most effectively run by an algorithm – and that my learning followed suit. In fact the authors recommend the same system for academic libraries: that the staff should place all recently returned books to an academic library in a prominent place in the foyer. The book you want, the book that others have been reading, is more likely to be there. Perhaps when I get home, I’ll sort all my books with the most recently read on the most prominent shelves.