Flaps1Pls
Our environment is stochastic, with a bucket of inputs and factors that impact every moment of our interface with the job at hand. from noise of heavy rain impact, accelerations at the seat from turbulence, ATC interactions, cockpit coordination, cabin needs, flight path management and systems. In this case, yon driver is levelling off at the same time as assessing weather on the radar and manoeuvering. Assuming the APLT is engaged at first, then if the LH driver is handling, when he decides to turn the plane to avoid weather, he needs 3 hands. His left hand is on the wrong side of the tracks; his right hand is on the throttles guarding them because that's a thing we are all going to do... and his 3rd hand is selection the heading bug on the MCP, while he looks at the heading relative to his EHSI radar overlay. Having 3 hands, he is now going to identify the APLT mode has gone from speed or VS or VNAV to ALT, and the ATO has changed from THR to SPD. And nothing goes wrong, the clutch dropping a TL (if that has happened) gets to be detected immediately, as the pilot is as sharp as everyone else that ponders their eventual fate.
Almost all adverse outcomes, from our aircraft accidents to Chernobyl, TMI, Apollo 13, even arguably Challenger, and even Columbia, Nimitz fire, ANZ901, AA905, EA401, KE801, the Bali B737 submarine, the Chuuk B737 submarine, AS214 at SFO, AFR 447, AIRBUS A330-300, Bombardier test accident, Gulfstreams test accident, CAL006, SIA006, etc etc etc, all have their roots in a loss of SA, either someone somewhere sometime didn't recognize an unexpected difference between expectations, and reality or didn't comprehend the significance of that having recognized it, or were unable to project the consequences that issue would result in later.
ASSumptions
Computers and automation bring improvements in some matters of reliability of outcome but make the SA issue harder for the operator. AA905 insert a DCT Rozo, "R" and assume the algorithm is going to give the closest WPT, and by a quirk of the NDB, questionably non-compliant with the standard, the WPT that is the top of the pile just ain't so... and the plane does an unexpected left turn... thereafter the crew busy in their activities are surprised to see the plane turning away from where they expect it fo go, and recover, in a way that reduces their time to recognize and recover the SA loss they have identified. and bonk into the top of a hill. The hill has been there for a few hundred million years So, as 905 taught, automation expectation can snowball quickly, and recovery from that can be hard. With 3 hands, 2 heads, per person, then things may be better. In the absence of those enhancements, mnemonics like OODA are a good way of remembering, like Caesar that thou are mortal. Shakespeare would have had some pithy line for that for sure, Mel Brooks will do though.
I would contend that few pilots will pick up a fault and respond correctly every time, in fact, I would go further, I'm not sure I have ever seen a pilot in a critical event respond exactly per the checklist or the training. That includes my direct observation, auditing, safety reviews, accident investigations.... pretty much the whole bunch. Maybe, It just doesn't come to mind. We can do something for 10 years in the sim exactly as written, and won't do it on the plane...
OTOH, have seen some innovative solutions to stuff like takeoff warnings, USA's etc, that are at least entertaining if there is no blood spilt.