1 in 100: How do you know?

NASA said the chance of failure with the shuttle launch (that was, in fact, successful YAY!) is 1 in a 100. How do they know that? According to the New York Times, NASA
" combines the findings of flight experience, computer simulations and expert judgment to assess how the shuttle's millions of parts will work or fail in varying situations."

Now this is just bunk. Sure, NASA can test systems and subsystems. Test parts for integrity under varying conditions, but it cannot test the entire shuttle until it flies. There have been 113 flights with two (2) catastrophic failures. It is impossible to quantify just what the chances are (probabilities, if you will) for failure. There just are NOT enough tests to know. That said, NASA still has to decide if the launch should go or not. In the end it's a decision based on little more than guess work.