The more books there are to choose from, the harder it becomes to pick new authors and novels to try. That too much choice makes it harder to choose is a well known phenomenon, so the sensible thing to do would be to restrict choice in some way thus making it easier to pick out books we want.
One way to do that is simply to ask my friends to tell me which new books and authors they like, but their taste isn’t necessarily mine, an issue highlighted by the fact that I seem to be alone in my active dislike of Philip Pullman’s His Dark Materials trilogy. Another way is to ask a computer, but online recommendation engines don’t seem to do much better.
Amazon has a massive catalogue of books and a huge amount of sales data, but its recommendation algorithm is hit and miss at the best of times. Partly this is because it relies on user reviews and star ratings and, in all honesty, most users seem incapable of spotting a dud when they read one. I recently bought an ebook that had nearly 40 excellent reviews, all with four or five stars, and was rewarded for my efforts with a heap of turgid dross.
Indeed, it’s been such a long time since I’ve trusted Amazon reviews that I decided to search for other book recommendation sites to see if someone was doing a better job. This was partly inspired by the discovery of the Book Readers Appreciation Group (BRAG) and its list of self-published books. BRAG says that “Our mission is to recognize quality on the part of authors who self-publish both print and digital books” so I was hopeful I’d find some gems. Its list of 17 “medallion honorees” contained three books about dolphins which, I will admit, gave me pause. But ignoring that, I picked three recommendations to explore further and went over to Amazon to read excerpts. Two were execrable, the third was merely very bad. Any diamonds in that list are sadly well buried by the manure.
Valioo looked more promising. “Never pick the wrong book again!” says the website, promising a “quick, fun and easy way to rate your books and receive highly personalized recommendations”. The service hasn’t been launched yet, but I signed up for the mailing list a while ago and was happy to get an email with a link to their beta Android app.
Now, I’m a fan of the developer motto ‘release early, release often’, but there is such a thing as releasing your app too early. Most of the Valioo app doesn’t work — you can only scan your books in by barcode, you can’t search for them by author or title, nor can you edit books you’ve added. The recommendations functionality doesn’t work yet, although it hints that I might get some recommendation emails at some point. Whilst Valioo might eventually develop into something useful, I’d give it several months before checking back.
On to LibraryThing and GoodReads, which both work roughly the same way: You catalogue your library and rate the books you’ve read, and they use that data to figure out your preferences. In the case of LibraryThing, your recommendations are simply a long list of book titles which is easy to skim but doesn’t give you enough information for you to start winnowing further. No cover art, no synopses, no key reviews.
GoodReads at least gives you either a grid of cover art or a list, with the list view the more usable of the two. Books are grouped by genre, and you can see the cover, rating, publication date and the first few lines of the synopsis. You can also narrow down the recommendations by picking which genres you’re interested in, although it takes a long time for that to percolate through to the recommendations page. Of the two, GoodReads seemed to produce a better list more closely aligned to my tastes.
BookLamp takes a totally different approach. Says Mashable:
BookLamp measures the story components of a book (characteristics like history, domestic environments, physical injury) and how it’s written (density, pacing, dialog, description, motion).
It uses these descriptions to suggest books you might like based on a book you’ve liked in the past, turning up books that match the actual style and content of the text rather than books people like you have purchased in the past. The objective is an improved online browsing experience.
BookLamp requires only the name of a favourite author or book to start the process. Once BookLamp has completed its analysis, you can whittle the list down by selecting which genres you’re interested in. The list is pretty minimalist with just the cover art, a link to get an overlay of the publisher’s information, and info about the book’s ‘DNA’. Again, that makes it hard to get a sense of what might be interesting.
Using yet another different approach is WhichBook, which asks you to pick four pairs of words, like ‘funny/serious’, ‘happy/sad’, ‘safe/disturbing’ or ‘beautiful/disgusting’. You move a slider to the preferred spot between those two extremes and WhichBook then recommends books that most closely match your selection. Whilst it’s an interesting idea, it doesn’t really work that well if you’re looking for something that you think you’ll actually like, but it would probably suit someone looking for books they have never heard of, providing you’re willing to take a punt.
YourNextRead.com also only needs the name of a book to begin and produces a ‘map’ of similar books. Unfortunately, for Neil Gaiman & Terry Pratchett’s Good Omens that just meant a list of Gaiman books, but for Nick Harkaway’s Gone Away World it produced much more interesting results. I think. With only the cover art to go by it was hard to tell.
Neil Gaiman (Photo credit: Wikipedia)
Finally, I looked at Hunch, a general recommendation engine. Hunch asks you loads of questions about yourself which it then uses to match you to holiday destinations, movies, music, books and a load more things. It’s quite fun answering Hunch’s questions, but their book recommendations page is really hard work to navigate. Again, it’s just a grid of cover art, with precious little extra information. There are no options to reduce your recommendations by adding favourite genres, all you can do is add keywords like ‘fiction’, ‘fantasty’, etc. Click on a book and all you get are recommendations made by other Hunch users, if there are any, but no publisher information or synopsis. Disappointing, really.
One problem with almost all of these recommendation engines was the lack of information about the books being recommended. That made it very difficult to assess the quality of the recommendations without searching Amazon for excerpts. Many don’t have ‘read inside’ enabled on Amazon, which is always a shame, because I really want to see a snippet of a book before I buy it as I can tell within a paragraph or two whether the writing is terrible or not. Whether relying on an algorithm or users reviews and ratings, all of these sites failed to convince me that I had found my next read.
It’s clear that book recommendation sites still have a very long way to go before they’ll be useful. In the meantime, it’s no surprise that the people find out about new books and authors through word of mouth and browsing in both online and bricks-and-mortar bookshops. Anything else is just too much hassle.
