> In the end, if you're going to have real pilots, it's going to be very very hard not to allow them to make mistakes which will destroy the craft.

Perhaps the solution is to not have pilots, and live with the loss of craft when its piloted autonomously and software hits an edge case.

There is precedent for aircraft that can't be flown without computer assistance/control: http://en.wikipedia.org/wiki/Northrop_Grumman_B-2_Spirit

Indeed.

A better example might be the the Mercury Project, which made 17 unmanned test flights before attempting to put a man into suborbital tests, and a few more before into orbit.

As for "can't be flowing without computer assistance", I think the F-16 was the first production plane of ours, in the '70s (https://en.wikipedia.org/wiki/General_Dynamics_F-16_Fighting...). Airbus's A320 the first airliner in the early '80s, albeit with some mechanical backup (in a fighter you have the option of ejecting): https://en.wikipedia.org/wiki/Fly-by-wire

There's also the relaxed stability concept (https://en.wikipedia.org/wiki/Relaxed_stability), which I assume is what you were referring to with the reference to the B-2, and per that Wikipedia article the MD-11 airliner employs it (although back when it was flying passengers if I'd had any option at all I'd have avoided flying in one, for many reasons I and others don't consider it to be a safe plane. And its parent DC-10 was the least safe wide body airliner in its days; as some people said before, and probably still now that Airbus is on the scene, "If it ain't Boeing, I ain't going").

As I noted in another posting, the new fuel used in this flight meant new patterns of vibration, some of which can only be found/tested in free flight. Sometimes you can get away with that, e.g. the Space Shuttle was never unmanned, then again the first flights had only two crew and they had SR-71 survival systems (ejection seats and intense environmental suits).

Good points. I concede the argument.