Best practices in software citation
Software is both a foundational tool and a primary output of modern computational research, yet citation practices for software remain inconsistent, incomplete, and rarely machine-actionable. Existing infrastructure designed for paper and data citation does not adequately serve the distinct needs of software citation, leaving a gap that impedes reproducibility, misattributes scholarly credit, and obscures the labor embedded in research pipelines. Drawing on a NASA-funded community workshop held in April 2026, we present an analysis of four interconnected themes: (I)~the cultural barriers to consistent citation practice; (II)~the need for clearer community norms and conventions; (III)~gaps in existing technical infrastructure and workflow; and (IV)~the emerging challenges posed by AI-assisted research. For each theme we identify targeted interventions and assign responsibility across stakeholder groups. We conclude that meaningful progress requires simultaneous action on technical and cultural fronts. Journal editors and publishers represent the single highest-leverage point for accelerating this change, and correct citation must become the path of least resistance within researchers' existing workflows.