It has been argued that distributed hash tables, or DHTs, are a useful substrate for the construction of wide-area applications. Yet to date there exist remarkably few deployed applications built atop DHTs. We believe that the difficulty in deploying a DHT is one factor contributing to this paltry success. Moreover, since many DHT applications can be built without modifying the DHT itself, we argue that a public, shared DHT deployment would increase the range of DHT applications in common use. In this talk, we demonstrate that the range of applications that can be built on a simple put/get interface to a DHT is larger than commonly assumed, and we discuss research challenges involved in sharing a DHT such as resource allocation. We will also present some of our experiences running a shared DHT on PlanetLab over the last few months.