Long before the shuttle Columbia was destroyed on re-entry last month, NASA scientists had considered literally hundreds of problems that might threaten the craft's safety and decided to launch it anyway. Columbia had accumulated a thick sheaf of what in the rocket business are called safety waivers problems that NASA had noted but decided posed too small a risk to bother with. "That's a pretty deep stack; it really is," a member of the Columbia Accident Investigation Board told TIME. "A lot of these [waivers] are legitimate every launch is going to have them but others are things you've learned to live with." The board is still weeks away from determining what technical failure caused Columbia's crash (investigators are examining foam that struck the left wing shortly after launch and super-hot gasses that entered the left wheel well during reentry). But the overuse of safety waivers has emerged as a sign that some blame may lie with NASA's approach to risk assessment. Many waivers, from snapped pins to broken foam, had been carried over from flight to flight, becoming virtually institutionalized weaknesses during Columbia's 22 years of use. "It's vitally important for us to understand why engineers are raising the same concerns they were years ago," says Sheila Widnall, who is on the investigation board. "I sympathize that you can't hold up a launch every time you have a concern, or you'll never fly. But there needs to be a way to put those concerns into an actionable form."