What’s better: computer-generated book recommendations, or the recommendations of a librarian?

Jessica Leber, a blogger over at Fast Coexist, contemplated this question in lieu of a new, free service promoted by Brooklyn Public Library, BookMatch, which involves filling out a web form that includes questions such as: “Are you looking for any specific reading recommendations? If so, please explain below. Otherwise, please share titles, authors, an/or types of books you enjoy, and why.” Or: “Are there any authors and/or types of books that you don’t like? Why?”

“The fact that I was so delighted [by the concept of BookMatch] was exactly what was dismaying,” Leber wrote, detailing her confusion over whether suggestions from librarians are better than those from Amazon, which suggests books based on data collected on millions of customers. And, as Leber explained, after the launch of the new service, the librarians at Brooklyn Public Library branches could not keep up with the requests in the span of one week, as initially planned—taking several weeks to respond, instead—implying that librarians may not be able to keep up with computers in providing quality advice.

Ultimately, it’s hard to know. Some librarians aren’t relying on their knowledge alone, but using algorithmic analysis and recommendation software, such as NoveList, to write their recommendations. But Leber makes an important point: “The real difference…may be less about the method and more about the motive. One is commercial, the other is not. Both want to recommend books people will like—but the commercial interest is to sell more books and the calculations behind that are not transparent to the user.”

Co-director of the Harvard Library Innovation Lab, David Weinberger, argued that librarians are generally more on the side of the reader than a computer is: “If you’re Amazon, you want to recommend books that people like, so they’ll come back. But if you’ve read Gone Girl, then Amazon is likely to recommend books that are very much like Gone Girl in all the best ways they can find. A librarian is likely also to recommend things very much like Gone Girl, but maybe she’ll also suggest books that are not exactly like Gone Girl and are more of a risk.”

“One of the great virtues of libraries is that they have a reader’s interests at heart,” Weinberger said. “That doesn’t necessarily mean that all humans trump all algorithms.”