Why Isn't More Research Code FOSS?

Academics produce software and tool prototypes and research code to supplement their research papers. This code dies because academia rewards the papers and not the packages. Most of this research code is gone, frozen, or unbuildable within a year of publication. This talk diagnoses why and and ends with concrete fixes for both researchers and the FOSS community.

1. I start with three of my own papers and trace what happened to the artifacts built for each. This grounds the central tension: artifact evaluation creates a real incentive to build working code at submission time, but nothing downstream rewards keeping it alive. Publication, graduation, hiring, and tenure all reward the next paper, never the maintained package.

2. I show the failure modes concretely. A short before/after demo of my own artifact-evaluated tool makes the "verification tax" specific to programming languages and formal methods visible: proofs and builds that break when a prover or a dependency updates.

3. I close with the payload: concrete asks for three audiences. For academics: cite software, count maintenance in hiring, fund the maintenance year, design for handoff from day one. For the FOSS community: mentor researchers in, treat a solid artifact as a real contribution, and adopt promising research tools instead of reimplementing them. For students: a genuine on-ramp from class project to maintained project.

I close with the argument that software sustainability is itself a research problem, and should be thougt of as a shared responsibility between two communities that currently misunderstand each other.

1. Why research code so often dies and the incentive structure behind dead repos.

2. What artifact evaluation does and doesn't guarantee, and why a reproducibility badge isn't a maintenance commitment.

3. How verified/PL/FM code rots in its own particular way when provers and dependencies move underneath it.

4. A concrete action list for academics, FOSS maintainers, and students.