Graybeard;
Not picked on at all...good question and to the point of the discussion about automation, what it is to be a pilot and current trends in the industry of which this may be one iteration; the perspectives offered by a history will tell us.
Would it not be more better to say? "For whatever reason, the salient fact here is, over a time period of 100 seconds, the autopilot caused the aircraft to lose more than 40kts of airspeed in the very late stages of the approach phase, in the attempt to stay on glideslope, and neither the autothrottle nor the crew intervened.."
I understand what you're driving at and agree that the loss of speed was caused by the autopilot trying to maintain the glideslope and that neither the autothrust nor the crew intervened.
Let us take this back further, given that nothing occurs here without antecedent.
What is the antecedent cause of such non-intervention by the autothrust and the crew? In the autothrust case, it was the incorrect data from RA1. What is it for the crew?
Here, the thrust levers actually move and so provided tactile and visual cues indicating that idle thrust was being commanded. That made sense, for a while, but not forever. When the thrust levers were pushed forward for more thrust as speed decreased, they returned to the idle position and remained there for the duration of the sequence until the crew responded essentially when there was little opportunity for a successful recovery. Why?
The broader issue is of course automation complacency and is huge and growing, (not without many cautions and warnings from experienced crews).
In reference to you question about how to posit the causal factors, do we cross the line and accept that it was indeed just the autothrust which "failed" to intervene and leave it at that? I think that opens the door quite wide to those who adhere to the notion more, not less automation is "required". I have seen decisions within flight operations which, after a serious "hand-flown" incident, have required the increased use of automation to "prevent" such incidents. We can all think of some circumstances where this is absolutely the case (making things safer), but at what cost in the overall scheme? At some point,
muscles atrophy if not used...
Do we design warnings for failure of warning systems? In such case, the call for "warnings" of such non-intervention (by the autothrust or the crew) may in some way be obvious but then how far does the regression go? Is it to be "warnings of non-operating systems", all the way down? The Takeoff Configuration Warning system on the MD82 at Madrid was a single-point failure waiting to occur. Is there a requirement for warning the crew that the warning system is itself compromised? And when we get to a reasonable (vice theoretical regression!) end, what is asked for is enhanced crew attention to the matter anyway.
We know humans are poor monitors and much better "controllers" when engaged physically and mentally with the tasks at hand. I had at one time considered (philosophied) one very simplistic notion which would turn monitoring and operational tasks (flying/controlling the aircraft), around - the "airplane" (rather its autoflight system), becomes the monitor and the crew becomes the operator. Such a system "mimics what is normal in all flight regimes" in terms of human-controlled/operated flight (sampling vast amounts of "normal" data), and may even anticipate solutions to flight circumstances for comparison with what the crew does. When flight brushes up against the many, many boundaries that may be established, progressive warnings emerge as may be appropriate. I fully realize the complexity philosophical shifts necessary for such a change and it probably isn't necessary given the relative, very high success rate of the present solutions.
Returning to THY1951, the causal pathway to stickshaker must somehow be set out such that both the technical and human factors issues are addressed. The fact that this same scenario has occurred twice speaks to the urgent need to address both issues but the singular issue of automation complacency is a fundamental outcome of the direction the industry has taken. In my view, pilot Associations have had a role to play here and have not provided as robust a response as might be expected given the depth of feeling seen here and elsewhere, but the issue certainly doesn't stop there. I sense we are going to see a re-trenching by the manufacturers but given that we have gone to two-man crews, (Cost, not safety were
the original driving force for automation solutions), and have now two generations trained in autoflight priorities and operations, such a shift will necessarily have to bridge automation and human factors, (ach!, such a large term - almost meaningless), rather than pioneer new solutions. PJ2