Online social networks like LinkedIn and Facebook regularly use People You May Know (PYMK) algorithms to encourage connectivity among their users. We argue that these algorithms have the unintended effect of making users’ interactions more visible, which can deter users from being intimate online. To test this theory, we analyze the database of a large online social network that lets users buy and exchange electronic greeting cards (eCards) with each other over the network. We find that users are more likely to buy eCards when they have more connections, but less likely to buy them if they have formed connections to friends of friends. We attribute the latter effect to the increased social visibility that comes with connecting to friends of friends, which PYMK algorithms encourage. Our study has implications for how privacy and intimacy interact online, and calls for a deeper investigation into the unintended consequences of algorithms.