Rotating schedules are commonly used in a number of industries and public s
ervices where employees work round the clock seven days a week. Several rul
es govern the design of such schedules. While this problem can be formulate
d as an integer linear program, this methodology is often too rigid. In thi
s paper I argue that to obtain rotating schedules that are acceptable in pr
actice, rules often have to be broken so that designing good schedules is m
ore of an art than a science.