Agreed that there should be a minimum bar for "completed" research code such as reproducibility and a clear summary, but engineers shouldn't expect the first version of a new algorithm to be easy to understand without additional material or ready for production without a complete rewrite.