Abstract: The present application is directed to methods and systems for providing social interaction within a customized media streaming service. In one aspect, the present application is directed to methods and systems for automated playlist generation based on social metadata. These systems allow for an internet media delivery service to learn about a listener's preferences, and changes in those preferences over time, without the listener being required to tell the service directly, but rather through social networking profiles and relationships to one or more similar or related users. This allows the service to immediately, without any input from the listener, play programming that is likely to be enjoyed by the listener. It also allows the service to learn about changes in a listener's preferences over time without requiring the user to actively express the preferences.