<p>Free/Libre and Open Source software is widely deployed and serves as digital infrastructure. This software is often developed through peer production, in which contributors select their own tasks. Although much of this software is innovative and well-regarded, in some cases quality and importance may be misaligned, such that highly important software is nonetheless low quality. In this paper, we describe a generalized approach for identifying important but poorly maintained FLOSS packages (Underproduction Analysis), then apply that approach to an important body of digital infrastructure. We validate our approach using alternate indicators of importance, quality, and risk. We then test a series of hypotheses about social and technical factors that correlate with underproduction. We find that older software is more likely to be underproduced, and we are surprised to find that additional maintenance resources are not always associated with lower levels of underproduction. Finally, we examine the community’s collaboration network structure, and find that those people working on underproduced packages tend to be those who are more central to the community’s collaboration network. These results have important implications both for engineering research seeking to understand outcomes of technical and social decisions, and the software development communities seeking to maintain FLOSS.</p>