Many space physics researchers engage in software development as part of their work. However, most have little or no formal training in computer science or software engineering. This contributes to common shortcomings in software development efforts, which adversely affect the reliability of the software and in turn the reproducibility of any research results that rely on it. A related problem is that researchers commonly have a poor understanding of software licenses, leading to license violations and misleading statements about how their own software is licensed. Here I lay out specific aspects of both issues, and offer recommendations for how they can be addressed through education and policy.