Second biggest hole in the cheese... (IMHO)
Well, first off let me say that I am not a pilot. But I have written quite a lot of software that interacts with humans.
After reading through all the posts in this thread, imho, for the last 500 posts or so the biggest hole in the cheese has been largely neglected.
I understand very well the thinking that changing the flight software to make decisions "for the pilots" is not necessarily a clever idea, that pilots should be kept "in the loop" instead.
I think the second biggest hole in the cheese was that the "Retard, retard, retard" warning ceased right after first TL was pulled to reverse. After viewing the video of another A320 landing posted here, it is quite obvious that it is NORMAL for a pilot to hear two or three "retard" shouts during flare. It is normal, and it is expected. Part of the "culture" of flying this plane. Now, during the accident flight the pilots heard "retard, retard, retard" shouts probably exactly as many times as they have ALWAYS heard. It provided absolutely no new information to them. It did NOT tell them that PF had forgot to pull back the TL2 lever. It basically told them that "you are supposed to pull the levers just about now", instead of "you have forgot the levers". Then PF pulled the lever (mistakenly only one) and the warning stopped. Perfectly normal, it seemed, business as usual.
Only after not feeling retardation and PNF callout "no spoilers, no deceleration" it became clear that something was very wrong. But by that time it was actually quite difficult for them to find out what was wrong:
1. the "retard, retard" call came and went as normal.
2. the PF remembered pulling the levers (by that time he certainly didn't know/remember that he pulled only one), as he should.
3. quick look at the throttles probably showed what he expected to see: one lever in the max reverse, the right lever substantially forward of the left one, hand holding the left lever. Given the shock of "no deceleration" a second or so before, I don't think it would be easy to notice that the right lever is a bit MORE forward from the left lever than it should be.
4. and for the PNF scanning the instruments it would be very difficult too because of the new software version not being installed and hence no warning about right lever being in the wrong position.
*IF* the plane had continuously shouted "retard, retard, retard" at them while they were frantically thinking about what the hell is wrong, why isn't it stopping, then I don't think it would have taken too many seconds for them to see the problem. I would bet the "retard, retard" calls would have actually reached their minds quicker than the written warning on displays, but either way, both would have eventually gotten to them.
As it was I think they went over the end blissfully unaware of the right lever being well forward of idle detent.
Frankly, I can't see ANY good reason for a "retard, retard" calls to stop after one engine pulled to reverse. From CVR it is perfectly clear that it did.
* * *
Besides that, as a secondary thought I think the interlock preventing forward thrust during at least one engine reverse (and ALLOWING spoilers) and/or vice versa: preventing reverse (and spoilers) during one engine forward thrust - makes perfect sense and should be introduced. First off, if even Boeing drivers who hail their planes as epitomes of manual flight consider such interlocking prudent, then it should have been incorporated on AB software even before the plane left the drawing boards. Because AB philosophy is to be MORE active at protecting the pilot from his own mistakes, not less.
And having forward thrust on one engine and reverse on the other IS a mistake, any way you cut it: when on a runway a plane should either be taking off (or going around, almost the same thing) OR stopping, braking. No inbetween, no hesitation. The current setup allows the plane to do neither of the two. Nor accelerate nor decelerate. That should NOT be possible, no matter what. Whenever the plane would do that (i.e. do nothing, no accel, no decel), the result is 100% certainly catastrophic - or at the very least
shockingly close to a nasty incident IF the pilot was quick to realise his/her mistake.
And from Airbus manuals and apparently from common knowledge: when engine(s) have gone into reverse then you ARE staying on the ground, no other choice - then this actually means that from the moment engine(s) have gone into reverse there remains no ambiguity whatsoever about the pilots intentions: he IS going to land. Full stop. When both squat switches are on, wheels are spinning and the reverser has deployed, there's no way to go around anyway (or at least it supposedly isn't), so
one throttle forward should NOT prevent activating the most important braking aid there is - spoilers.
* * *
Of course all this doesn't mean this logic caused the accident - the biggest hole in the cheese was of course a human mistake. But lack of "retard, retard" shouts *has* to be the second biggest, preventing them from diagnosing the nature of the problem quickly.
And no, I don't think AB human interface design philosophy (including the throttle design) is somehow flawed or worse than Boiengs. Implementation, yes, seems to have some flaws, which seem to even have already developed cures, albeit not installed on this particular aircraft because of various parties not respecting the gravity of those shortcomings - even in light of previous accidents.
And no, of course I don't have too many credentials to write about all that. A mere software developer (occasionally SLF) as I am. And a 5 year lurker on this great forum.