While the NTSB findings are not yet published on its website, press reports (New York Times and Washington Post and excerpted below) of yesterday's board hearing indicate the following:
1.) The board said that the copilot pushed the rudder too far to one side and then overcorrected by pushing hard in the other direction, and that the crash would not have occurred had he not touched the rudder. The five-member board agreed unanimously on the causes of the accident, but disagreed over whether the training or the tail sensitivity should rank second.
2.) The co-pilot had attended American's advanced training program, which sought to teach pilots how to better recover after an unexpected, dangerous incident. In one scenario similar to the Flight 587 accident, American's training video prompted pilots to respond to dramatic wake turbulence after following a 747 into the sky. American's simulation taught pilots to use the rudder, which controls the vertical tail's back-and-forth movements, along with the wings to stabilize the plane. But the safety board said the training was inaccurate because it taught pilots to associate wake turbulence with using the rudder. The co-pilot incorrectly perceived these reactions of the aircraft as being caused by outside turbulence rather than his own actions.
3.) The safety board concluded that the Airbus rudder was more sensitive than those on comparable aircraft and that its performance differed dramatically on the ground, in the air and at high speeds.
For all airliners there is a "maneuvering speed" above which rudder use is unsafe, but the A300's system for limiting rudder use as speed increases is unusual. On most planes, as speed increases, the pedal must be pushed farther and farther to get a given amount of movement out of the rudder. On the A300 and the A310, the system simply limits how far the pedal can move. The result is that very small pedal movements produce relatively large rudder changes. Since pushing the rudder produces a delayed result, pilots may push too far when they get no immediate response, then overshoot.
Dr. Malcolm Brenner, a psychologist with the board, called it "an inherently unfriendly design."
The Board recommended that the Federal Aviation Administration require manufacturers and aviation operators to provide better pilot training on rudder capabilities and proper rudder use in flight. It also recommended the FAA review the rudder design of Airbus A300-600 and the similar rudder on A310 aircraft. Airbus said it would work with the FAA if the agency decides to review its rudder design.
4.) The board voted to discuss developing more formal standards to have the FAA receive reports of overseas safety incidents that might help prevent similar incidents in the United States.
The issue arose after investigators said they were "disappointed" that Airbus and American were less than forthcoming in providing information about the rudder performance of another A300-600 involved in a non-fatal accident four years before Flight 587 crashed.
American alleged that Airbus withheld information about its rudder system in the 1997 incident, in which the plane stalled and its rudder was damaged; pilots regained control of that plane. The safety board ruled that incident was unrelated to the rudder-design issues raised by Flight 587.
5.) In the board meeting, staff members and members of the board engaged in extended discussions intended to disprove the idea that the crash was caused by sabotage.
6.) The board also rejected the idea that the tail failed because it was made of composite materials.
7.) The "tragic coupling of both the pilot and the aircraft really was the precipitating event that took down Flight 587," NTSB Vice Chairman Mark V. Rosenker said.