Large tasks can be overwhelming. For example, many people have thousands of digital photographs that languish in unorganized archives because it is difficult and time consuming to gather them into meaningful collections. Such tasks are hard to start because they seem to require long uninterrupted periods of effort to make meaningful progress. We propose the idea of selfsourcing as a way to help people to perform large personal information tasks by breaking them into manageable microtasks. Using ideas from crowdsourcing and task management, selfsourcing can help people take advantage of existing gaps in time and recover quickly from interruptions. We present several achievable selfsourcing scenarios and explore how they can facilitate information work in interruption-driven environments.