The logic in book recommendations isn’t always algorithmic

26 July 2010

The algorithms used by online book-sellers still have some way to go when it comes to guessing — with any real accuracy — what we might like to read next.

Our prevailing mood, or state of mind, rather than the title we read last, or whatever we’ve read in the past, is what actually determines our next book title choice.

“You can’t recommend books to strangers without asking personal questions,” Stein told me. As he pointed out, what we want to read is often pegged to transitory moods. The same book may not thrill the same person at every point in his or her life. “I don’t think people read ‘for’ pleasure, exactly,” he went on. “Of course there is pleasure in reading. But mainly we do it out of need. Because we’re lonely, or confused, or need to laugh, or want some kind of protection or quiet — or disturbance, or truth, or whatever.”

Originally published Monday 26 July 2010, with subsequent revisions, updates to lapsed URLs, etc.