We present a new interface for exploring and navigating large-scale discussions on the internet. Our system, tldr, focuses on three primary user goals: finding, navigating and filtering content based on the users' own relevance criteria. We employ a progressive design approach that combines user research from the popular web site Reddit.com and visualizations from existing discussion data. Based on lessons gleaned from previous research and our own user studies, we argue that the tldr interface can lessen the problems of information overload and inefficient navigation in large-scale discussion spaces. Current implementation and next steps for wider deployment are discussed.