Significant Function Logic

The significant function logic is intended to let users weed out insignificant functions from further analysis. Some find this confusing and think that that it could cause the missing of important failure modes since failure effects and consequences have not yet been defined. Those who use it respond that it is only used to elimate cearly unneceassary functions and that if there is any question that a conservative default to significant is used. In the NAVAIR process this step is clearly optional. Lately there has been discusion favoring removal of it. Any opinions out there?