Legal outcome prediction models lack explainability, emphasizing the importance of understanding precedent for human legal actors.