The French air safety authority, the BEA, is beginning to put together the picture of what happened to AF447 enroute from Rio de Janeiro to Paris. Key parts of the wreckage have been found, including the flight data recorder. The BEA website contains links directing the reader to a more detailed view of many aspects of the investigation.
What has been so unnerving about this particular crash is the lack of detailed understanding of how it initiated and propagated. We know how it terminated. In particular, the flight is an example of in-flight loss of control of the aircraft. By all accounts, the airplane was in good working order and well equipped for transoceanic flight. It had a crew that, by prevailing standards, was well qualified to operate the aircraft. How could there possibly be a loss of control that could confound this well equipped machine with expereienced crew?
The aircraft had more than one crew on board as well as a highly automated flight control system comprised of advanced navigation and communications, auto pilot, and auto throttle systems with the usual redundancies. Yet with all of the human resources and automation, and with a century of aircraft design knowledge behind it, this passenger aircraft managed to take an excursion into uncontrolled flight and impact the ocean.
The BEA has disclosed a timeline of events in the cockpit as well as a description of the flight attitude of the aircraft. What I find interesting are the control inputs made by the PF (pilot flying). In the face of indications of a stall, the PF primarily tried to pull the nose up. This is the wrong control input for a stalled airplane. What makes the incident worthy of note is the interaction of the crew with the automation and sensors. Aviation Week and Space Technology has a good article worth reading on this very topic.