Ahhh, the 'ole "coputers are safer" argument...

Yeah, and pilots are supernatural, infallible beigns... You do read NTSB's preliminary investigation report digests, don't you? Not many navigation equipment breakdowns, plenty of pilot judgment errors.

In other words, had I let the "computer fly it", both my brother and myself would be dead.

OTOH, had those Ukrainian pilots over German/Swiss border listened to their TCAS and not the human controller, 70-something childern would be alive...

Fly-by-wire is fine, but you DAMN well better have a hard-wired safety override. A good pilot's "feel" and knowledge of a plane is often WAY better than an engineers.

I follow this kind of discussion in AW&ST and Flying for ages. Pilots on your side of the fence would say 'So what if I overstress the plane a bit in dangerous situation? Better wrinkled wing skin than hole in the ground.' Yes, but why are they so sure they would 'just wrinkle the wing', not break it, or enter accelerated stall, or stay too far from the limit and fail to use all of the plane capability?

Engineers don't feel the airplane, they model it (then measure its actual performance, then refine their models...). Enourmous amount of feedback from development, testing and early customer pilots is taken (mostly regarding engonomy, consistency of symbology etc). The result is something like A320 where, for example, a pilot deciding a go around is called for just yanks the joystick fully backward: the plane applies TOGA power, raises the gear, retracts the flaps and tries to achieve the best angle of climb airspeed pulling up at the edge of stall (or something like that). A friend who transitioned from ancient B-737-200 to A320 tells me he would not change back for anything.

It is in principle possible to make software as close to bug-free as one wants; it is just a matter of time and resources, that is, money. OTOH, it is impossible to eliminate human error beyond some level quite far from perfect.