Innovation is now, more than ever before, a collaborative pursuit. Collaboration is occurring in new ways across a variety of different boundaries, including geographical, organizational and disciplinary boundaries, the public–private divide, and even the boundary between human and machine. Patents, traditionally seen as mechanisms for enhancing collaboration, may not always function as intended in these diverse collaborative arrangements. This Article focuses on three types of boundary-spanning collaboration that occur with increasing frequency in processes of innovation: (1) user–producer innovation, (2) public–private partnerships, and (3) human–machine collaboration. It identifies some of the ways in which these types of boundary-spanning collaborations challenge, and are challenged by, the existing U.S. patent law framework. It focuses in particular on the problems that arise from trying to fit a variety of different discovery processes and participants into an inventor-centric model of invention through patent law’s joint inventorship doctrine. This approach to collaborative invention ignores, rather than accommodates, the different characteristics and needs of boundary-spanning collaborations. The analysis highlights the need to take differences among participants and their relationships with each other in processes of intellectual production—what I refer to as the social contexts of innovation—into greater account in patent law as these types of collaborations become the norm, rather than the exception.