Framing an issue as a puzzle, problem, or mess is an illustrative approach to characterizing the issue's complexity within organizational theory and systems thinking. We use this approach to characterize the issue of designing collective behaviors for robot swarms and discuss how various research goals have shaped the current state of the field. We contextualize our discussion at these three levels by highlighting relevant literature. Our aim is to emphasize key challenges that arise in the development of robot swarms for real-world applications and to motivate further work on promising research directions.