Originally Posted by
Mac the Knife
One things that stands out to a non-pilot high-hours PPruNer is that how many of these nasty incidents involve:
1) Autotrim, and (to a lesser extent),
2) Autothrottle
Mac
These highly automated aircraft are not adequately annunciating to the operator
mode-changes; their reason, and consequences/effect on other systems.
:-(
In hindsight you can often find issues with annunciations, but it is a highly non-trivial problem. Cognitive overload is a huge issue, and a lot of effort is put into answering the questions of what information to present to the pilots at what time. Lots of pieces of information are intentionally withheld in certain phases of flight in order not to distract the pilots unnecessarily. In many cases with highly integrated electronics, a lot of this information can later be recalled on a computer display in the cockpit to assess which system are affected by a certain failure.
After the event it is often easy to say "If only the pilots had known X when Y happened! Why didn't the aircraft tell them?"
To appreciate the near-impossibility of the task of designing cockpit information systems, try to make a list of parameters which may become supremely important to know at some point. Then make a list of critical scenarios and try to cross-reference the parameters with the scenarios. You will most likely end up with dozens, if not hundreds of parameters and small events that
might be extremely important to know for almost all scenarios. But with humans being cognitively limited, it is impossible to present all of them, or even a majority. A decision to limit display (or aural or tactile annunciation) to a handful of parameters must be taken, also taking into account the probability that any of these parameters may be wrong.
These day such tasks involve cognitive psychologists, human-factors and human-machine-interface specialists, perhaps even sociologists, besides engineers and pilots. But by the very nature of the problem, these decisions
cannot be perfect.
Bernd