Modern scientific discovery, particularly in the geosciences, is driven by a model which entails collaboration around data and software by teams of specialized experts. Common to these collaborations is the need to share and control data and descriptions of the data, to share compute resources and tools, to share and develop code, the need to move data between compute resources and team members, and the need to save and publish data and results. Several groups have been developing technological solutions to enable data-centric collaborations based on the concept of a data commons. However, the notion of a data commons and what constitutes a data commons is not well defined. We propose a workshop to discuss what a data commons should provide for the Geoscience community based on some representative science use cases, where we are today, and what needs to be accomplished.