Title: From 0 to 100K in 10 years: nurturing an open-source software community
Abstract: The Stan project develops free, open-source tools for probabilistic and differentiable programming, statistical inference, and graphics. I tell non-scientists that it’s like Excel for applied statisticians. In this talk, I focus on what worked and what didn’t in managing the project as it evolved from an initial problem specification to functional prototype to an international group of 40+ developers and 100K+ users. I’ll touch on topics involving writing specifications, coding, testing, documentation, communication, decision making and governance, onboarding and supporting users and developers, simplified interfaces, code review, constraining scope, release management, dev ops, marketing, outreach, diversity, licensing, and funding. For background, I’ll also discuss what Stan does, how it does it, show some examples, and discuss how it’s being used by scientists and educators.