Jack Ganssle's 2002 top-ten list of why embedded projects fail remains startlingly current — unrealistic schedules, vague requirements, and optimistic code still sink teams that should know better.