This is an extended analogy:
Sites like Pinterest, Twitter, Tumblr, Instapaper, Snip.it, Clipboard, and Curisma, among others, all allow their users to decide what aspects of the web (text, media, etc.) are worth saving and sharing, instead of browsing the web from Google, or even Facebook for that matter. Because many of these networks have asymmetric follow/follower models, and because users can “tune” whom they are following, users’ feeds could increase in relevance as items are retweeted or repinned. -Semil Shah, The Rise of Pinterest and the Shift from Search to Discovery
The implication of “search” is that there’s a huge volume of stuff, through which you sift. It’s a bad model for the educational aims of libraries. For most of library history, libraries were not based on “searching” for materials. Libraries were and are assembled at great expense and effort, by human beings, so that there can be a place where community can form around an evolving body of knowledge.
Discovery, as Shah uses it above, emphasizes the role of human agency, both in shaping a body of knowledge itself, and in being able to select what voices one pays attention to. In order to meet the lofty goal of being communities of discovery and education, libraries, like Pinterest, need to emphasize curation and discovery over “search.” When you’re searching, you look for a hypothetically known thing, when you’re discovering, you learn about something that you didn’t know before. It’s something algorithmic search design does poorly, because they rely on known patterns matching what’s out there. If you really want to discover something, you need to involve other people. Pinterest, like libraries, puts the people first, and the algorithm second. That being said, a body of knowledge will always reflect its creators, so see below: