NOT A KNOCKDOWN PUFF. The latest coverage of the Columbia post-mortem is looking at causes other than tile damage account shedding of external tank insulation on takeoff. "[Shuttle program manager Ron] Dittemore said that during Columbia's final minutes, the autopilot was causing the craft to rapidly move the control surfaces and to eventually even fire small rockets in a losing effort to gain control of the yawing motion of Columbia." Couple that with the slight rise in temperature on the port wing and you have something akin to a sailboat on a broad reach getting hit with a knockdown puff. In a displacement sloop, you simply ease the jib and the main and keep going. In a planing catboat, such as the Laser, where there is only a mainsail, that is a very dangerous strategy (it took me a few swims to figure it out) as easing the main has the potential to create a positive feedback loop leading to the sail rolling the boat to windward: the death roll. (Best have the cunningham lead made fast, as that's the only thing holding the mast to the boat as it turtles.) The proper response to the knockdown puff is to trim in and enjoy the ride! (I have some silverware attesting to how enjoyable THAT can be.)

Knockdown puffs don't happen at the edge of the atmosphere, according to a meteorologist I asked. "At that altitude, they wouldn't have come into contact with stronger winds. They'd have to get another 50,000-100,000 feet lower for that. Also, the wind gradient in the jet stream up there is never that sharp; since the friction of the earth's surface is zero for all intents and purposes, the only way there could have been anything remotely wrong is if they flew over a really tall thunderstorm. Obviously, the weather was fine where it happened. Clear air turbulence is not a factor up there, either." So, what happened to provide signals to the guidance system to make corrections that did not provide negative feedback?

UPDATE: My brother the King Air pilot provides the following "no s***, this really happened" moment: "I, for one, have experienced failures of the autopilot on the King Airs that I (used to) fly in which the autopilot was working perfectly well and suddenly decided to send some odd command or another causing one of those brief moments of sheer terror during the hours of monotony that we quip about. There have been cases in which the attitude indicator from which the autopilot takes its input tumbles and the autopilot "corrects" and places the aircraft in an unusual attitude. However, with all of the back up systems on the space shuttle, I should think that a tumbled gyro or flaky autopilot would be taken off line and replaced by one of the redundant ones.

It will be interesting to see what NASA learns from this accident."

Quite.