Lion Air crash: Jakarta Boeing 737 'had prior instrument error'

[julianhigginson]

Absolute most drastic case, MCAS will be removed and pilots will require re-training to deal with the different handling characteristics of the max. Some people here seem to fail to comprehend this, it is not unstable or badly mannered, it simply handles differently than the 737 classic and thus requires the MCAS in order for it to be flown by pilots certified on the 737 classic. The upward pitch when thrust is increased is a characteristic of ALL airliners with engines slung under the wings. The difference here being the max exhibits this to a greater degree.

My reading of the fundamental issue that the hidden software bodge was designed to overcome, is the 737 MAX8 not only exhibits regular upward pitch when thrust is increased, but far worse - it also exhibits more upward pitch increase the more the upward pitch is increased.... ie the upward pitch on the 737 MAX8 has a positive feedback element that is not present on other 737s (or pretty much any other commercial plane ever) and the plane is aerodynamically unstable for that reason.

[Brumby]

.... and the plane is aerodynamically unstable for that reason.

How wrong can you be?

IF it WAS aerodynamically unstable, then how come there have been thousands of successful flights?

Such illogical statements don't add to the discussion, but make a mockery of it.

[Kleinstein]

.... and the plane is aerodynamically unstable for that reason.

How wrong can you be?

IF it WAS aerodynamically unstable, then how come there have been thousands of successful flights?
......

Most planes get  aerodynamically unstable at to high an AoA - that is what the stall problem is about. The problem with the 737max can be that high high thrust the critical AoA can be lower than normal. So adding thrust can be even worse.

With a working MACS (e.g. no sensor failure) the system seems to work OK and they even got through with only minimal extra training.

I am a little skeptic about the option to only give only better instructions to the pilots on how to disable the MACS. They would ad least need quite some extra (e.g. simulator) training on how to fly with the MACS deactivated.

My guess is Boing will have to update the software and give some training / info to the pilots, so that the system is less likely to fail (e.g. use both sensors, limit the power) and the pilots know better how to handle a possible upset. So it's more like a question on how long the planes stay at ground. If thing go well maybe before the summer season, would be nice to be ready before the winter season.

[julianhigginson]

How wrong can you be?

IF it WAS aerodynamically unstable, then how come there have been thousands of successful flights?

Such illogical statements don't add to the discussion, but make a mockery of it.

I dunno... how wrong can you be?  It looks like you are trying to make it a competition or something. those 3 sentences have an amazing hit rate for wrongness. I'm not sure I'll be able to keep up but hey, I'll give it a red hot go:

Now. In absence of you going and reading up on control theory or even just feedback in systems I'll break what I said down into smaller steps for you, and try to explain the terminology so you can keep up with the discussion and stop resorting to derailing the discussion and making a mockery of it by acting abusively.

So, what I'm talking about with this new plane, is there's a problematic state variable in a flying plane "system". This state variable is either "pitch" or "angle of attack" (whichever it happens to be, the particular one is unimportant for this discussion.. possibly it's also tied to a second state variable, which I'd expect might be "thrust", maybe there's even more associated variables.. but I'm leaving that out for the sake of simplicity.. only one contains the feedback mechanism when feedback starts, so it's the critical one here)

1) This state variable, for some range of values, has a positive feedback loop. That is - for a range of values of the state variable, with no control operating on that variable, that state variable will increase without bound (well, until hitting a natural limit of the system, stalling the plane, and killing everyone onboard)

2) The positive feedback effect on that state variable is literally a case of instability for that state value. Other 737s in flight do not have this positive feedback on this state variable and therefore this state variable isn't unstable for other 737s in flight.

3) Instability in any one (or more) state variables of a system literally defines that system to be unstable. that's, like, the definition of an unstable system...  Now, other planes that don't have positive feedback in any state variables don't meet this definition. This plane does, though.

4) Now, we need to take a moment and back off from the definition above and realise that a system being "unstable" doesn't necessarily mean the whole system just oscillates chaotically in the full range of all its state variables and is impossible to control in any way from the instant it starts in motion (that would be your worst case for instability in a system)

5) In the case of this plane, instability most likely makes it harder to control the plane when the system hits a state variable value that brings in positive feedback... Which is exactly what the software bodge is there to try and avoid in this plane, but seems it did that in a terrible way. The very existence of this software bodge tells us explicitly that Boeing knew that not only is the flying plane unstable in this variable, it's bad enough that the instability needs to be countered (or masked) You don't bodge software like that up and put it inside a life critical control system just because you're bored on a rainy weekend, you know.


PS, I'm not an expert on aerodymanics  here, but it's my understanding that a lot of fighter planes are deliberately aerodynamically unstable in different ways because it allows them to do fast manoeuvres they couldn't do if they were stable... BUT fighter planes have ejector seats for that reason, and aren't full of passengers for that reason (OK, and probably other reasons...)

Anyway, here's some fun discussion on fighter plane stability I googled for you. Enjoy. But please be very careful not to sign in and abuse the people on this stack exchange about how they are stupid and not contributing to a discussion on fighter plane instability and control, because if fighter planes were unstable how could they ever fly, and thousands of fighter plane flights happen every day.
https://aviation.stackexchange.com/questions/8049/are-fighter-jets-designed-to-be-so-inherently-unstable-that-a-human-cant-fly-on

[ogden]

5) In the case of this plane, instability most likely makes it harder to control the plane when the system hits a state variable value that brings in positive feedback... Which is exactly what the software bodge is there to try and avoid in this plane, but seems it did that in a terrible way.

Thank you, well explained. Chain of events for many may seem like: high thrust of engines pitched plane up, MCAS activated and decided to kill everybody by trimming horizontal stabilizers into deadly nosedive. Definitely it was not that simple. Problem could be that positive feedback pitch-up is non-intuitive for pilots that are not trained for such. They may see MCAS actions (trim) as "this f**ing plane does not let me gain altitude as fast as I am used to". If they counteract MCAS trim using other control surfaces - it will make things worse. How else we can explain stabilizer trim jackscrew to be in full nosedive position while plane was gaining altitude & overspeeding till last moments? It may not be faulty AOA instrument. It could be so that pilots were successfully operating older 737 planes closer to critical AOA than treshold is set for MCAS. Suddenly MCAS decided that what they do is not safe, started to trim, pilots did not know what the plane is doing, was trying to compensate using wrong controls/approach. When pilots counteract stabilizer trim during speed increase at high thrust, sooner rather than later trimmed for dive stabilizer will be more effective than all the other controls combined that are in the pilots hands (joystick).

To me it seems like "pilot not trained for particular (type of the) plane", like this airplane crash. They were taking off with parking brake activated...

[Nusa]

of which over a hundred planes just like the 2 crashed ones are out there in the world by now

376 delivered to customers as of last month.

The blame here has already been established: 737M8 and M9 are grounded worldwide. Simple as that.

Such things have happened before. The DC-10 fleet was grounded by the FAA (they even pulled it's type rating) early in its career for a series of accidents. Yet it went on to a successful passenger career. A bunch of them are still flying in cargo service (most are now called MD-10's after a glass cockpit upgrade that allows them to share a type rating with the MD-11). Fex-ed operates most of them.

The 737 classic had rudder failures earlier in its career that caused fatal crashes and control problems. Took them years to figure that one out and years more before they were all retrofited. There were so many in the sky that I don't think anyone even considered grounding the entire fleet. It would have crippled airlines around the world.

And that's just off the top of my head. The other guys are right, the 737-MAX is going to fly again, and probably fairly soon. I'd put money on it.

I dunno... how wrong can you be?  It looks like you are trying to make it a competition or something.

You might be winning this competition.

PS, I'm not an expert on aerodymanics  here

That's real obvious. Yet you're acting like an authority with your conclusions.

[ogden]

PS, I'm not an expert on aerodymanics  here

That's real obvious. Yet you're acting like an authority with your conclusions.

What he tells about positive feedback is not conclusion, but well known fact:

The engines were both larger and relocated slightly up and forward from the previous NG CFM56-7 engines to accomodate their larger diameter. This new location and size of the nacelle causes it to produce lift at high AoA; as the nacelle is ahead of the CofG this causes a pitch-up effect which could in turn further increase the AoA and send the aircraft closer towards the stall.

All this adds to pitch-up effect from engine thrust.

http://www.b737.org.uk/mcas.htm

[Nusa]

PS, I'm not an expert on aerodymanics  here

That's real obvious. Yet you're acting like an authority with your conclusions.

What he tells about positive feedback is not conclusion, but well known fact:

The engines were both larger and relocated slightly up and forward from the previous NG CFM56-7 engines to accomodate their larger diameter. This new location and size of the nacelle causes it to produce lift at high AoA; as the nacelle is ahead of the CofG this causes a pitch-up effect which could in turn further increase the AoA and send the aircraft closer towards the stall.

All this adds to pitch-up effect from engine thrust.

http://www.b737.org.uk/mcas.htm

This is not a software analysis that's simulated to guaranteed failure in 0.01 seconds. There's real physics and time and pilot opportunities to correct involved. What an automatic system can do automatically, a pilot can do manually, if necessary. In any case, the engine placement is NOT the direct cause of these accidents. Focus on what actually went wrong.

[ogden]

This is not a software analysis that's simulated to guaranteed failure in 0.01 seconds. There's real physics and time and pilot opportunities to correct involved. What an automatic system can do automatically, a pilot can do manually, if necessary. In any case, the engine placement is NOT the direct cause of these accidents. Focus on what actually went wrong.

Even Boeing admitted that most likely source of crash is (possibly inferior) MCAS system which were introduced for which exactly reason? - Engine placement. So, please, don't.... Also you have no authority to tell what we shall focus on or not. On the other hand I agree to you that most likely they will solve potential issues w/o scrapping MAX or reverting it to NG engines.

[tooki]

My reading of the fundamental issue that the hidden software bodge was designed to overcome, is the 737 MAX8 not only exhibits regular upward pitch when thrust is increased, but far worse - it also exhibits more upward pitch increase the more the upward pitch is increased.... ie the upward pitch on the 737 MAX8 has a positive feedback element that is not present on other 737s (or pretty much any other commercial plane ever) and the plane is aerodynamically unstable for that reason.

The thing is, having an inherently aerodynamically unstable airframe and using software to make it what I'll call "apparently stable" to the pilots is a completely valid design approach. (Think of the Segway scooter, which is only possible because of software.)

So, operating under the assumption that MCAS is what failed, my take on this is that Boeing screwed the pooch in the implementation of MCAS.

1. It should not have been allowed to make such drastic corrections. The maximum correction per intervention event is already far more than they'd originally planned — and had submitted to the FAA — but the real issue is that if a pilot overrides a full-scale MCAS intervention, even by the tiniest amount, MCAS will re-trigger soon and can apply full-scale correction again, resulting in MCAS being able to push the elevators all the way nose-down. This is many, many, many times more correction than it had been described as being capable of, and I'm fairly confident this was not by design. (It should only be able to apply a certain amount of total cumulative correction, rather than only being limited within a single intervention, but unlimited in the number of interventions it can make.)

2. The other issue, arguably the bigger one, is that MCAS seems to use only the input from a single AOA sensor, which is insanity IMHO, especially given that the plane already has more than one such sensor! I can't imagine any reason why you wouldn't use the inputs from hardware that's already in place. IIRC, the MAX has two AOA sensors, which of course means that in the case of disagreement, you can't immediately know which one is wrong. So I totally agree with the others who've said that simply, there should be three, so that you can do a "best 2 of 3" analysis and thus have a reading with a high level of confidence.

Both of these flaws have to be in place at the same time for the crashes to happen in the way we suspect as of right now. Fixing either one would have prevented the crashes. And both are comparatively easy to fix, so if I were Boeing, I'd fix both.

[Kjelt]

As I said before, I am not sure this plane will fly again with the same name.
It has a bad reputation and people don't want to fly in it anymore.

Today the first airliner cancelled their order of 49 MAX planes .
Sorry only dutch but translated: the reason for the cancelation is that their customers have lost all trust in this plane and the airliner would like to order a different type of plane.

https://www.nu.nl/economie/5804268/vliegmaatschappij-garuda-indonesia-annuleert-bestelling-49-boeings.html

[tooki]

As I said before, I am not sure this plane will fly again with the same name.
It has a bad reputation and people don't want to fly in it anymore.

You've said it before, and I'll say again that there's a zero percent chance this will happen.

The DC10, for example, had some design flaws that caused a few crashes early on. It had a similar image problem. But the flaws were identified and fixed, and the DC10 went on to become one of the most successful airliners of its era. Consumers absolutely did not dwell on its early reputation.

The fact is, most people don't even give a second's thought to what aircraft type they're flying on. They assume (correctly) that the airlines have a vested interest in not having their planes crash, and thus that they're airworthy.

[julianhigginson]

That's real obvious. Yet you're acting like an authority with your conclusions.

Most of the conclusions I'm talking about are already discussed by people who do have expertise..

I'm not the person who invented the idea that the basic 737 shape is incompatible with the huge engines they bolted on to the max8, and that the hacking on of engines that are too big caused the plane to become unstable.

I'm not the one who first explained how the way this particular plane was allowed into the market and the skies seems to be shady as all hell due to the above 2 points and the apparently secret/hidden fix that was bolted into the plane's software system to try and patch it but apparently spoken about very little.

that post I made that you quoted half a line of out of context was just dealing with someone who wanted to have a go about things and didn't even know the fundamentals of system stability.

my main point in this thread is dealing with armchair pilots who think that despite ALL WE HEAR about how irregular this project was, and how different the plane physically is compared to the 737 version it's pretending to be the same as, that just because one pilot is known to have been able to save one of these planes with the fault from crashing once, that it must mean that the other 2 flight crews whose planes crashed, killing everyone on board, must have just been incompetent and the planes are fine.

Maybe they will fly these planes again, because sure, rich people hate losing money, and I bet there's a bunch of rich people directing lackeys right now to make sure they lose the least money possible over this... And sure, maybe it is just going to be a control system update...

Whatever they do, I just want a serious investigation into the suitability of whatever their final fixed version of the plane is, because the first declaration of suitability for that plane was an absolute pile of bullshit. So more of the same process is not going to cut it..  And really, moving forward, airplane manufacturers should not be self-certifying planes if stuff like this can happen.

And hopefully they'll also come to the conclusion that's its a very new and specific type of plane so people get trained properly on how not to die while flying it...

The only conclusion of my own that I've brought up here, really, is that it is looking a bit like this project contained a high level conspiracy to avoid corporate responsibility and save money, at the cost of human safety and now lives... So a thorough investigation into the decisions made by the key high-level people responsible for this project should to be a big priority, and if found to be a criminal conspiracy or even the slightest hit of unethical behaviour, they should be going to prison for consecutive life sentences for each person that died on those planes.. And I'll argue that all day, given what we've heard about the way the project certification was managed, the way new plane introduction and training was managed, and how different this plane apparently is to the plane it's certified to be..

[julianhigginson]

he thing is, having an inherently aerodynamically unstable airframe and using software to make it what I'll call "apparently stable" to the pilots is a completely valid design approach. (Think of the Segway scooter, which is only possible because of software.)

I absolutely agree... some forms of instability can definitely be patched with control systems, and the segway is a great example.... I would never claim otherwise.. but if you're going to do that, the system has to be fit for purpose...

And when we are talking about a plane where a critical error kills everyone on board, that system has to be robust and it has to be incapable of killing people no matter what you throw at it.

And when you're talking about a massive complex thing like a plane where pilots have to be able to override the system if it's going wrong, well you also have to be frank and open about what the plane is, how it actually works, and how it might be different from other planes you've ttold people it's the same thing as..

Both of these flaws have to be in place at the same time for the crashes to happen in the way we suspect as of right now. Fixing either one would have prevented the crashes. And both are comparatively easy to fix, so if I were Boeing, I'd fix both.

you'd also have to fix the documentation and the training and probably the plane type, but yes I agree, if the MCAS could be made to not fall down in a screaming heap and kill everyone on board, it could possibly be a suitable fix for the planes physical instability, as long as people piloting the plane were fully aware of what it was doing, why it was there, and had some experience flying the plane without it enabled.

Also, whoever let the plane design out with such obvious holes  in the design needs to be seriously investigated (like, did anyone even consider a basic system level FMEA on this thing? that should have pointed out the issue with only one mechanical vane sensor being used as bright as day!)

[Kjelt]

As I said before, I am not sure this plane will fly again with the same name.
It has a bad reputation and people don't want to fly in it anymore.

You've said it before, and I'll say again that there's a zero percent chance this will happen.

The DC10, for example,

Lets continue over a year and we will see.
Referring to planes of the 70s makes no sense, society has changed too much esp social media, internet....
If airliners are cancelling orders it looks bad, I can't say I can blame them.

[GeorgeOfTheJungle]

It wouldn't be the first time the sensor is ok but the sensor data that arrives to the flight computer isn't, due a software bug somewhere else.

https://en.wikipedia.org/wiki/Qantas_Flight_72#Final_report

At 12:42:27 the aircraft made a sudden uncommanded pitch down manoeuvre, recording −0.8 g, reaching 8.4 degrees pitch down and rapidly descending 650 feet (200 m) in about 20 seconds before the pilots were able to return the aircraft to the assigned cruise flight level. At 12:45:08 the aircraft then made a second uncommanded manoeuvre of similar nature, this time reaching +0.2 g, 3.5 degrees pitch down and descending 400 feet (120 m) in about 16 seconds before being returned to level flight.[14][15] Unrestrained passengers and crew as well as some restrained passengers were flung around the cabin or crushed by overhead luggage as well as crashing with overhead compartments. The pilots stabilised the plane and declared a state of alert (pan-pan), which was later updated to mayday when the extent of injuries was relayed to the flight crew
[...]
Final report
Analysis
After detailed forensic analysis of the FDR data, the flight control primary computer (FCPC) software and the air data inertial reference unit (ADIRU), it was determined that the CPU of the ADIRU corrupted the angle of attack (AOA) data. The exact nature was that the ADIRU CPU erroneously relabelled the altitude data word so that the binary data that represented 37,012 (the altitude at the time of the incident) would represent an angle of attack of 50.625 degrees. The FCPC then processed the erroneously high AOA data, triggering the high-AOA protection mode, which sent a command to the electrical flight control system (EFCS) to pitch the nose down

[floobydust]

well this is getting interesting
so AOA indicators is an option?
like a software addon in oscilloscopes? 
but they are still suppose to tell the pilot there is a stall? or it did not?

The Wright brothers had (only) an AoA sensor and gauge indicator on their 1903 Flyer. It was a piece of yarn tethered on a stick.
I'm glad Boeing remembers it's kind of important, since day one.

"[Boeing] announced it would now make standard an indicator light that warns pilots of a sensor malfunction"
"A U.S. airline source said that feature would cost roughly $80,000 extra" 
https://www.cbsnews.com/news/boeing-737-max-plane-crash-company-to-make-standard-light-warning-pilots-of-sensor-malfunction/

edit: added cost

[james_s]

And no programmer is going to work for a company on H1B for a few years.

Usually it's OPT->H1B->EB3->Green card.
If I work for a company and I don't see my I140 filed the next year I got H1B, I'll see myself out.

It happens all the time. At a previous job I had we had several developers on H1B visas for years. I'm not a fan of that system but it's not a rare situation.

[james_s]

Lets continue over a year and we will see.
Referring to planes of the 70s makes no sense, society has changed too much esp social media, internet....
If airliners are cancelling orders it looks bad, I can't say I can blame them.

Irrational behavior has become more common it seems, but I'm still quite confident the max will fly again. All new aircraft have teething problems, and occasionally this leads to disaster. When the Dreamliner battery fires were happening I saw a lot of people say that was the end of the line for those and people wouldn't fly on them yet here we are.

Once the problem has been fixed and the planes are sufficiently tested I would not hesitate to fly on a 737 max. Looking at it rationally, even if nothing were fixed and we just started flying the planes again, I'm still much more likely to die while driving to the airport in a car than in a plane crash from a statistical standpoint. Just to put this into perspective, approximately 50,000 people are killed each year in car accidents in the USA alone. That's approximately a 737 crash every single day of the year in deaths, yet people think nothing of getting into a car.

[djacobow]

The only conclusion of my own that I've brought up here, really, is that it is looking a bit like this project contained a high level conspiracy to avoid corporate responsibility and save money, at the cost of human safety and now lives... So a thorough investigation into the decisions made by the key high-level people responsible for this project should to be a big priority, and if found to be a criminal conspiracy or even the slightest hit of unethical behaviour, they should be going to prison for consecutive life sentences for each person that died on those planes.. And I'll argue that all day, given what we've heard about the way the project certification was managed, the way new plane introduction and training was managed, and how different this plane apparently is to the plane it's certified to be..

This is a whopper of a conclusion. Though I agree there should be an investigation, and that if there criminal conspiracy or negligence are discovered, there should be punishment, neither of those things are established yet, not even close.

- is the idea of putting large engines on a 737 fundamentally flawed?
- is this airplane significantly different from other 737 that it should have required a type rating?
- if this plane pitches up more at already high AOA's, does that make it fundamentally unsafe?
- is the MCAS concept fundamentally flawed?
- is the MCAS system engineered poorly?
- should it be required for two AOA sensors to agree to activate MCAS?
- did Boeing know (or suspect) that the concept (or implementation) was flawed?
- did Boeing, or people at Boeing, know the airplane was unsafe? Did management override them?
- did Boeing hide information from the FAA? Did they use political power to force the FAA's hand?

NONE of these things are known right now, and they all matter. People seem to think they know the answers to these, but they do not, not even aviation experts.

My contention on this thread all along has been that aviation accident chains are long and complex, and that just because the plane is flawed or broken in some way does not mean the pilots acted competently. I think there's good evidence (not proof) already that the pilots involved did not perform in an exemplary fashion. Was their performance within the expected bounds for pilot capability? That will be another question for investigators to determine, too.

However, that, too is a separate question, and is distinct from everything that might be bad about the airplane, process, or organization that created it.

[floobydust]

FBI have joined a criminal investigation into the 737 max. certifications.
Boeing essentially putting a "CE" sticker on their planes- they do not have the integrity to do self-certification.

We can keep haggling over the MCAS implementation but, as with many disasters, engineers were steamrollered into deploying something as quickly as possible and thus unsafe.
I think the profession has a huge problem- whistle blow and get fired, or join the corporate group think, keep your job and proceed with corrupt management's instructions.

I see it happening more with other scandals such as Volkswagen emissions , FIU bridge collapse, Theranos, Takata air bags etc.
The executives drive the company hard to maximize shareholder's profits as fast a possible, the result is unsafe products, and all the while the engineer's little voice is ignored.

We're somehow expected as PE's to act only with integrity despite our higher ups overruling that. Boeing surely had engineers that knew this MCAS rollout is shit, Arduino-crowd kind of software quality and bypassing necessary safety design and evaluations. But dare speak up about it and get thrown under the bus.

I'm not sure when or how corporate greed is going to get moderated, how many lives must be lost. Boeing took it to a new level but I can't see anything fixing this.

[rdl]

Post by Ralph Nader, who apparently lost a relative in one of the crashes.

Greedy Boeing’s Avoidable Design and Software Time Bombs

[KL27x]

Boeing: you third world airlines just need to hire armchair pilots to fly these planes. Every armchair pilot would have cut the stab trim, and they would not have wanted to know what MCAS did or why it was needed, esp whenever they made high AOA maneuvers, thereafter, for whatever reason.

[djacobow]

Boeing: you third world airlines just need to hire armchair pilots to fly these planes. Every armchair pilot would have cut the stab trim, and they would not have wanted to know what MCAS did or why it was needed, esp whenever they made high AOA maneuvers, thereafter, for whatever reason.

I'll promise not to fly any airliners if you'll promise to stay off accident investigations.

[KL27x]

You take it too personally, Djacobow.

Serious question: What it takes seconds for an MCAS malfunction to do, how long does it take to undo that after you cut the stab trim? After you cut the stab trim you can't just press the up button on the yoke anymore, right? You have cut all powered control over the jack screw and the horizontal stabilizer. It looks like you'd be turning the wheel for a long time.

As for pilot reaction: If you are used to hearing this thing clacking away during autopilot trim adjustments, then you might not notice it, at all? During AF447, the audio stall warning went off 70 times for over 2 minutes, and the black box recordings suggest that the pilots never even discussed a stall. Some studies have suggested that audio warnings don't register to the pilot under many circumstances, which is why most of the important alerts are not audio, only.

So, when the MCAS goes haywire, the pilots immediate reaction might be nothing. Then after awhile, they might have either tuned out the trim wheel adjusting altogether. Or they might realize, "hey, that trim adjustment has been going on for longer than usual." They might then first wonder why autopilot is turned on when they are sure it is off. Then they might turn off the autopilot (redundantly). Say for sake of speculation that the MCAS just happened to finish doing its thing at the time the pilot pushes the autopilot disengage button. Then he pushes the manual trim button, to get the trim back up. And it works as he expects. Problem solved... but not yet. It happens again. And pilot is on the wrong road from here on out.

Turning ON autopilot would have disabled MCAS. But it might have looked like the autopilot was malfunctioning. Simply knowing MCAS existed could make a huge difference in response.

One of your primary algorithms when flying this plane might be that if instruments are malfunctioning, you can always turn off the autopilot and fly, manually. When you think the autopilot is turning itself on/off and malfunctioning, then you might get some panic and tunnel vision, because your major "out" and feeling of control and safety has been removed. Now your tunnel vision has you locked on the autopilot disengage button, because hitting it seemed to have worked the first time, and now you think the autopilot is trying to kill you. It's obviously not a runaway trim, because it seems to have something to do with the autopilot, and it is not continous. When you press the trim up, it works. When you press trim down, that works. Trim controls are working. Training was maybe very specific under what condition you cut the stab trim, and in the moment the pilot is probably reverting to training and not able to think so clearly due to being so close to death at a fairly low altitude. He's not able to reason that cutting stab trim will prevent autopilot from using it, too (or MCAS, secret autopilot that is the "manual" autopilot).  He is following his checklist.