Control systems fail for reasons that are obvious in hindsight. For example, a model leaves out an important dynamic, or an assumption stops being valid, or the controller interacts with the physical system in an unexpected way. The cause for failure is rarely found by tuning gains alone. It requires going back to first principles and understanding what the system is actually doing. We wanted to put together a collection of stories from engineers who encountered these problems firsthand: what went wrong, why it was difficult to diagnose, and the insight that ultimately led to a working system.