AbstractModern research is inescapably digital, with data and publications most often created, analyzed, and stored electronically, using tools and methods expressed in software. While some of this software is general‐purpose office software, a great deal of it is developed specifically for research, often by researchers themselves. Research software is essential to progress in science, engineering, and all other fields, but it is often not developed, shared, or stored in a sustainable way. The following paper presents findings from an ethnography of two research software projects that have, over the last 10 years, cooperatively organized development efforts to produce important software enabling scientific breakthroughs in both astronomy and macromolecular modeling. The work of these two projects are framed in terms of James Carse's model of finite and infinite games. I argue that by incentivizing institutional governance that resembles the design of an infinite game, funding agencies can increase the sustainability of research software and improve various aspects of data‐driven scientific discovery.