Book recommendations for people we don’t know well can be tricky. When we recommend books to friends, we have a lot of information to go on. We know their likes and dislikes, political and religious beliefs, family histories. We’ve discussed books, so we know their favorites, right down to their favorite parts (that part with the butterflies!) and, more importantly, their least favorite parts (that scene in the bathtub!).
Book recommendations are part of my library job, a very enjoyable part. But making book recommendations is one of those things that, like figure skating, looks easy until you try to do it. A good Book Whisperer’s skill is honed by years of practice and reading.
It’s also necessary to listen for the nuances of a reader’s request. For example, when a lady asks for a “nice book” I know that’s code for no sex, violence, or upsetting elements. That hardly means a trite or uneventful read, as fans of Maeve Binchy, Joanna Trollope, or Alexander McCall Smith know well.
Recommendations for kids call for even more nuance. In just a few moments, I must get a feel for the kid (dragged in or enthusiastic), chat with him or her enough to get a feel for reading level and interests, and very delicately, balance the parent’s opinions and family standards (“No vampire books!”)
That’s why I’m dismayed by the addition of a new feature to our library’s catalog. Now appearing under the usual information about library holdings is a You Might Also Like selection of books generated by Goodreads. I have nothing against Goodreads or other computerized book recommenders. Their crowd-sourced suggestions are often good, but sometimes hit or miss. Lately the misses have outnumbered the hits.
Yesterday, I checked the library catalog for my blog mate Linda Rodriguez’s novel Every Last Secret, winner of the Malice Domestic Best First Traditional Mystery. It’s a compelling novel about “Skeet” Bannion, a half-Cherokee woman who makes a new life for herself as chief of a Missouri campus police force. The software suggested some other novels I might like based on my interest in Linda’s traditional mystery.
First was Steven F. Havill’s series set in Posadas County, New Mexico featuring a female sheriff protagonist. Good match.
Another featured read alike? Stieg Larsson’s The Girl With the Dragon Tattoo, which I seem to remember was set in Sweden. Sure, TGWTD is compelling, but its blend of sex and violence hardly makes it a traditional mystery. That comparison isn’t apples to oranges; it’s apples to flamethrowers.
I am not saying these recommendations are not often useful – they are in many cases. It’s just that when they go wrong, they can go really wrong, especially for kids.
A young reader wanted the latest of Rick Riordan’s grade school fantasy novels. When I checked the You Might Like section, books by Lori Armstrong came up. The first of her books in the list? Bitten, an adult vampire novel. If you scrolled down the listings, you’d find another of her books, one that was written for children, but the way the computer program worked, Armstrong’s most recent titles were listed at the top. And kids don’t scroll. Children have a more trusting relationship with computers than adults do. Adults have an easier time determining when to trust and when not to trust what a computer “tells” us. Kids, the digital natives who have grown up in the blue light of a computer screen, sometimes lack this ability.
A coworker shared a story about a little girl who searched the catalog for what I’ll call Book X. The girl kept saying “but I don’t want to read Book Y.” We checked the computer screen. She thought that because her book was not available, she had to take the next book the computer listed in the You Might Like section.
Until book-matching software improves, I might like it better if libraries stopped using crowd-sourced book recommendations.
Writers, have you been pleased with the titles that book recommendation sites like Goodreads match to your book?