Goodreads wants to tell you what to read next
What book should you read next? Goodreads thinks it has some answers.
The online site, a book-focused social network, now provides Netflix-like recommendations based on readers鈥� previous picks and pans. Once users have rated 20 books, the system will start suggesting new possibilities to try. The algorithm is based on one from Discoverreads, a smaller company purchased by Goodreads in March, according to .
Computerized recommendations are a big change for a site that鈥檚 previously been about what your friends are reading and praising. The early buzz is that the change is a plus, though, and that it鈥檚 about time the site turns the data from its 6 million members into a boon for readers rather than just advertisers.
鈥淲hile peer recommendations are important, it's hard to argue against math,鈥� . thinks the recommendations will be a boon to book groups, always facing the question of what their members are collectively most likely to enjoy.
I gave the service a trial spin. For beginning users, it is of course a rough tool 鈥� 20 ratings was enough to give the system a good idea that I enjoyed classic children's literature, but not enough to translate that to new releases I haven鈥檛 yet encountered. I鈥檓 sure it means something that I鈥檓 among the few living readers not to appreciate Elizabeth Gilbert鈥檚 鈥�Eat Pray Love,鈥� but I think it鈥檒l take Goodreads more data points to figure out just what that says about my tastes.
(The system did suggest I try Gabrielle Hamilton鈥檚 鈥淏lood, Bones, and Butter,鈥� a book I have already read. I was lukewarm on it at the start but loved the last third. Hard to translate that into stars.)
By relying on ratings rather than searches or purchases, Goodreads does come up with useful data. One of the shortcomings of Amazon鈥檚 system, for instance, is that my account isn鈥檛 just based on my personal likes; it鈥檚 skewed by my one-time searches for work, by my mom鈥檚 searches when she visits our house, by gift purchases, by my kids, and so on.
On Netflix, our recommendations reflect the odd mashup of my husband鈥檚 yen for fast-paced thrillers and my 4-year-old鈥檚 love for cartoon dinosaurs. (Washington Post critic Ron Charles obviously has a similar problem. He recently that his "daughter's fondness for all things Ashton Kutcher has rendered Netflix's recommendation engine useless.")
Goodreads, however, claims that its service will be more directed, bragging that it has more data points and more nuance, using information such as how users categorize their books.
鈥淯sing 鈥楾he Help鈥� as an example, if a reader liked it because they like reading historical fiction and they also liked 鈥楾he Guernsey Literary and Potato Peel Society鈥� and 鈥楢 Tree Grows in Brooklyn,鈥� then a great recommendation for that reader is 鈥楾hese Is My Words,鈥欌�� . 鈥淲ith Amazon, the focus is on other best sellers so someone buying 鈥楾he Help鈥� would get recommendations for books as diverse as 鈥�Water For Elephants,鈥� 鈥�The Hunger Games鈥� and 鈥極ne Day.鈥�"
For me in the end, humans are still my best source of referrals. When my mom visits from across the country, she brings piles of books to read, and we spend her days here in running discussions of how we鈥檙e enjoying them. My friend and has pointed me toward Bill Buford鈥檚 鈥淗eat,鈥� one of the great kitchen memoirs, and introduced me to Laurie Colwin鈥檚 essays. She convinced me I had to try The Zuni Cafe Cookbook and Anne Willan鈥檚 鈥淧erfect Soups.鈥�
I鈥檓 intrigued by Goodreads, enough to keep refining my ratings with them 鈥� but Mom and Nancy have a big human advantage over them: They not only know me well; they鈥檙e right there to loan me copies of the books that they think I鈥檒l love.
Seattle writer Rebekah Denn blogs at
Join the Monitor's book discussion on and .
