The amount of information available to individuals today is enormous and rapidly increasing. Continued progress in science, education, and technology is fundamentally dependent on making sense of and finding insights in overwhelming amounts of data. However, human cognition, while unparalleled at discovering patterns and linking seemingly-disparate concepts, is also limited in the amount of information it can process at once. One promising solution to this problem is through social collaboration, in which groups of individuals collaborate to produce knowledge and solve problems that exceed a single individual's cognitive capacity. Emerging online paradigms that aggregate the efforts of many individuals -- such as Digg, del.icio.us, Wikipedia, and Mechanical Turk -- are existence proofs of the power of collective intelligence. However, what makes these systems successful, and what will the next generation of systems look like?

Here I describe a series of studies examining harnessing the power of the crowds for complex and creative information processing tasks in Wikipedia, Mechanical Turk, and beyond. I also present research into visualization and machine learning tools aimed at increasing the effectiveness of these systems. Finally, I discuss early forays into extending social collaboration to support insight and discovery.