Open source software communities depend on diverse groups of maintainers, contributors, and users working together to develop tools. In this talk, I introduce the idea of a ComposableCommunities dataset, discuss key incentives issues in further developing the package community.
While the bus factor may be one of the most popular (and organizationally political) metrics in open source software development, it is only a partial view of package community health, and often interpreted without regard to community anti-patterns like gatekeeping. Julia is unique in the extent of its use of GitHub as the user interface and ‘operating system’ for Julia package development. Good package infrastructure alone, however, does not guarantee healthy communities. Julia’s package infrastructure and particularly its use of git and GitHub, mean that data about the package development process and package (as well as maintainer) dependency structures is available in unusual quality and granularity. I propose a new piece of ‘community infrastructure’ with which the Julia community can introspect, defined shared metrics and goals, and via which the economics of open source, in particular incentive misalignments like topic squatting, contributor churn, and zombie packages might be explored quantitatively.