"Columbia Disaster Underscores The Risky Nature of Risk Analysis," by CharlesSeife. Science, 14 February 2003, pages 1001-1002.
Before the recent Columbia disaster, NASA's official estimate of shuttlefailures was one catastrophic failure per 250 flights, yet the Columbia failurewas the second in 113 shuttle flights. This article describes probabilisticrisk assessment, especially how it is used by NASA. This is a fielddesigned to assess the likely failure rate of complex systems as well asidentify those components of a system that are "most likely to contribute tosuch a disaster." Both Congress and the public prefer to know one number whenit comes to risk, but the manager of NASA risk assessment says, "the relativerisks are what's really important," so that resources can be directed to areaswhere they will do the most good.
--- Mike Breen