Fault detection often depends on the specific order of inputs that establish states which eventually lead to a failure. However, beyond basic structural coverage metrics, it is often difficult to determine whether code has been exercised sufficiently