Most incidents are going to happen at the two parts of the flight with least margin for error, take-off and landing. Even if something bad does go wrong in cruise, you are very far from the ground and glide well, you have lots of chances to fix it and not die.
So, sure, it might not be a coincidence, but "it happened near take off" is not surprising.
Sure, but this plane also showed extremely erratic vertical speed readings, just as Lion Air did. The pilots appeared to be fighting the plane before requesting an emergency turn-around.
I wonder whether those flights will happen on those planes. There are only a few hundred MAXs in existence, and now two new ones have failed. Grounding the fleet wouldn't feel like an overreaction to me.
I'm not sure why you say "however" or think we disagree. Maybe you misread my "wouldn't feel like an overreaction" as "would feel like an overreaction"?
I agree with grounding the fleet, especially if it's the case that an MCAS software update was already being prepared but not yet deployed.
My condolences to your family. But seriously, this is concerning and I feel terrible for the people that lost their lives and their families. What a terrible way to go. I cannot even imagine.
That should already have been the case after the first incident. I would guess and hope they'll be grounded until MCAS is significantly reworked, if it was the cause. This level of risk is completely off the charts for modern airliner aviation.
And most crucially, this system can't be overridden using stick inputs whereas previous systems on the 737 were. And this change amongst others was kept relatively hidden by Boeing as a cost-savings measure, so that a 737 pilot can fly a new MAX without needing a new certification.
That is probably true. Airbus pilots know and expect the computer to have the last word. Pilots of all planes are expected to know how to engage alternate law (more direct control) or disable specific parts of automation when necessary. I was going to say they are not doing too well at that, but then you only hear about it when they failed. In the same vein you don't hear much about automation saving lives.
I have a friend who is a commercial pilot and who doesn't like Airbus planes because he doesn't want to be overruled by a computer... Accident statistics are a wash between Boeing and Airbus, though.
Giving the infrequency of airplane crashes, it probably isn't coincidence.
Very unlikely for a brand new plane to suffer issues with altitude shortly after take off.
Even less likely for it to happen twice in a span of a few months.
However, aviation history has more than a handful of samples where a new design suffered multiple crashes shortly after introduction and the cause was ultimately a design flaw.
Someone is probably pulling all of the flight logs for the 737 MAX right now and is going to see how many other non-fatal cases of altitude issues after take have been reported.
That depends on the size of the effect. The total loss of a modern airliner due to a control issue is so rare that we might indeed be there now, statistically.
No. If you take a small sample size and you find the problem twice then it warrants a great deal of investigating. This is common in auditing for example.
I was going to try to tell you that airplanes are extremely safe, that two highly publicized crashes in 8 months should be compared to a far larger number of unsurprising, non-newsworthy auto crashes...but there are only 330 total 737 Max planes flying, and two have crashed. Those aren't the numbers I expected.
My wife, a civil engineer, loves all kinds of new things, especially new car models.
Me, the software engineer, am extremely paranoid with newly released models of any kind. I'd rather wait a few years for it to be tested, never being an early adopter for it.
I predict that this is going to end up being a software error related to the angle of attack sensor (Boeing reported possible problems with the AOA on 11/7/2018)
The incident definitely bears a lot of similarities to the recent Lion Air crash that prompted that report (control issues shortly after takeoff, followed by a high energy impact).
In response to future 'trim system/AOA sensor' comments: Transport aircraft have a procedure to cut out a runaway/malfunctioning trim or stick pusher. Usually, this procedure is to press and hold the autopilot disconnect button on the yoke/stick or the trim switch and then, later, deactivate the system. The pressing of the autopilot disconnect button or trim switch is a 'memory item', meaning that it should be performed immediately, without a checklist, upon recognizing that there's a problem. Previously, this malfunction has been a rare occurrence, and so can be startling to pilots or not immediately recognized. I've seen pilots fail to apply the procedure and crash many times in the simulator, even after being told it was coming. It could be said that the pilots of the Lion Air aircraft and, potentially, this Ethiopian aircraft should have known the procedure and reacted properly, however, for Boeing to put them in this situation is, in my opinion, also quite hazardous. With so many hastily trained and low-competence airline pilots flying around, should we really rely on them to understand the technology and react properly?
I don't think that's avoidable. The demand is too high right now. Personally, I'd like to see increased automation and "low-pilot" aircraft enter service.
Note that we have decades of experience and skill at easily teaching basic stick and rudder flying skills, however deploying independent systems/automation analysts under emergency conditions is highly problematic and will likely result in more tragedies.
I'm just saying its simpler, cheaper, and safer to deploy people who can troubleshoot simple flying systems rather than to deploy people who will have to debug untested and complicated automation systems on the fly. It would be the most extreme field service posting, ever.
What we have here is an over optimization problem; we "have to" build unflyable aircraft to keep up with everyone elses level of recklessness to squeeze a couple more hyper-financialized pennies out of the system or be replaced by someone reckless enough to do it. Traditionally regulation helps in these situations. The aircraft of the future is far more likely to have networking and interconnected processors prohibited by law for safety reasons than to be an internet of things self-flying appliance. That would result in a nice simple, admittedly lower performance, far more reliable overall system. Non-networked non-interconnected control microprocessors seem safe enough so far, but making an infinitely complicated hyper optimized system seems in practice to result in planes falling out of the sky too often.
You could make the same faulty argument to obviate the need for all safety features on everything. Safety is about defense in depth. Ideally we'd have all planes flown by perfect platonic human pilots, but such a thing does not actually exist. All people are flawed (some more than others), and we should design technology to be resilient in the face of the inevitable human failings to the fullest extent possible, especially when lives are on the line. Boeing has simply failed to do so here.
You could make this argument for a toaster or a word processor. If you’re talking about a flying vehicle carrying 158 people it’s no more than a poor excuse. The pilots ought to be competent and properly trained and only then you can start blaming the equipment.
If this accident is a second trim related incident due to their MCAS system, it is damn near criminal negligence at this point and they probably need to ground the fleet. From an earlier article, pulling back on the yoke would disable the system.[1] They removed that functionality. Not easy to teach people new habits or proper procedures when the intuitive solution worked. And if the intuitive solution worked is that not really the proper procedure despite what engineers on the ground think.
If MCAS is the culprit here, Boeing clearly released this functionality without proper vetting and testing. It’s a sad day for safety and for automation. These accidents will likely slow the advance of BVR for drones[2] and more autonomous flight for cargo.[3] The fact these accidents are occurring overseas where the pilots have less training will also buttress the decision for the 1500 hour rule.[4] Like I said, if MCAS is suspected here, probably time to ground the new 737 despite what it will do to Boeing, because right now it looks like they cut some corners, the FAA let them, and families are mourning. Pilot error isn’t really the issue, it’s engineering, process, and training.
Doesn't this mean he/she was more used to the old functionality, which didn't work and perhaps hadn't been sufficiently retrained on the new operating procedure?
BTW "they" is a gender-neutral pronoun that has been used as a singular pronoun for over 600 years (longer than singular "you"), is less awkward than "he/she", and is inclusive of non-binary people.
So if the cause is what the parent comment suggests that just adds weight to the notion that it's a stupid procedure. I'm in favor of waiting for the final report before getting all agitated.
>because right now it looks like they cut some corners, the FAA let them, and families are mourning. Pilot error isn’t really the issue, it’s engineering, process, and training.
That doesn’t seem like engineering, process, nor training.
That looks like something connected to large scale corporate profit.
How did Boeing manage to gain self reporting to the FAA?
> If MCAS is the culprit here, Boeing clearly released this functionality without proper vetting and testing
From what I read MCAS is not even a true safety system, much more like a piloting assist, and should not have had the authority to override pilot input.
It appears that the true stall avoidance system is a "hard" safety system and works independently with much safeguards, and a lot of "overrideability" just for such moments of sensor failure.
So, they added that dangerous system to just let pilots to have the same feel and experience as in old model - an absolutely noncritical issue.
Boeing deserves to be sued into oblivion for this, if it turns out to be something related to Lion Air. It’s a horrible feature and they are being obstinate about it simply because they are afraid of liability. Meanwhile hundreds have died already.
> for Boeing to put them in this situation is, in my opinion, also quite hazardous
I'm sure this is an edge case given all of the successful 737 MAX 8 fights, but it's a catastrophic one.
To go from smooth flying conditions to having the nose unexpectedly dive and having seconds to save your life and everyone on board is almost more dangerous than difficult flying conditions because it's so unexpected.
I'd be surprised if Boeing continues to treat this as a training issue. The plane needs to be better.
Right, but how many successful test flights and real world flights where this never happened?
2 accidents is 2 too many, but it's not like every MAX 8 or anywhere close to the majority or even a large minority are experiencing catastrophic failure.
Sure, it's far higher than what's acceptable and the industry norm, but I'd imagine it's enough of an edge case that it snuck past testing and 99%+ of real world flights.
I just read that the software update for MCAS (the Lion Air crash cause) had been delayed and isn't rolling out until next month. If that's true, seems to raise the probability of this being a similar MCAS issue.
The parent comment merely said delayed MCAS training would make it more likely that it was the cause. That seems like an obviously true statement... If the training had happen, that would obviously make it less likely.
Given the number of other similarities to the Lion Air crash, I think the parent comment was being pretty conservative in their speculation (at this point, I'd be surprised if it _wasn't_ MCAS related).
Quick clarification: it's a software update to the plane's MCAS (I think automatically disabling it when the pilots are trying to adjust trim themselves) that's been delayed, not training.
So, I wonder how much credence my comment would get if I proclaimed:
"My guess is both pilots fell asleep right after takeoff but before they could get the autopilot engaged, so the plane plummeted to the ground because no one was awake at the controls."
Probably very little, but there is nothing stopping you making that statement. However HN usually has a decent signal/noise ratio so it's worth discussing here at the very least.
theses high Flights numbers seems to be incoherent with the fact that Max is a new plane with small number delivered, it is surely a mistake with the 737NG (next gen) introduced in 1997
Honestly, I know people often jump in here with technical comments on stories like these, but posting this stuff on HN always just feels like rubbernecking to me.
If this is a software error should someone be held responsible? Developer? QA? Manager? Did they follow their process? Kind of scary for us to think of...
91 comments
[ 5.8 ms ] story [ 148 ms ] threadI hope it's a coincidence, but that can't look good on Boeing.
E: Removed the word 'sure'
So, sure, it might not be a coincidence, but "it happened near take off" is not surprising.
- the one that avoided flying MAXs and it turned out to be something else
- the one that avoided flying MAXs and it turned out to be a problem with the MAX
- the one that did not avoid flying MAXs and it turned out to be something else
- the one that did not avoid flying MAXs and it turned out to be a problem with the MAX
I think it is much safer to be in the first two and potentially being labeled paranoid later on worst case.
I agree with grounding the fleet, especially if it's the case that an MCAS software update was already being prepared but not yet deployed.
Previous crash was involving a stall prevention software overwhelming the pilot input.
It was said to be added to new 737 due to it getting slightly tail heavier, and as a safeguard to pilot overreacting to it feeling unusual.
It may be the case that Airbus is simply better at building automation into the fabric of their aircraft from the ground up than Boeing.
I have a friend who is a commercial pilot and who doesn't like Airbus planes because he doesn't want to be overruled by a computer... Accident statistics are a wash between Boeing and Airbus, though.
https://www.king5.com/article/news/local/us-pilots-want-more...
[0] https://www.seattletimes.com/seattle-news/faa-faulted-for-ou...
Very unlikely for a brand new plane to suffer issues with altitude shortly after take off.
Even less likely for it to happen twice in a span of a few months.
However, aviation history has more than a handful of samples where a new design suffered multiple crashes shortly after introduction and the cause was ultimately a design flaw.
I don't think that's how statistics work. You need larger sample sizes in order to make such a statement with high confidence.
I am not flying any leg on the 737 Max now.
Me, the software engineer, am extremely paranoid with newly released models of any kind. I'd rather wait a few years for it to be tested, never being an early adopter for it.
basis: none whatsoever; just a guess.
[1] https://www.businessinsider.com/airlines-that-fly-the-boeing...
I'm just saying its simpler, cheaper, and safer to deploy people who can troubleshoot simple flying systems rather than to deploy people who will have to debug untested and complicated automation systems on the fly. It would be the most extreme field service posting, ever.
What we have here is an over optimization problem; we "have to" build unflyable aircraft to keep up with everyone elses level of recklessness to squeeze a couple more hyper-financialized pennies out of the system or be replaced by someone reckless enough to do it. Traditionally regulation helps in these situations. The aircraft of the future is far more likely to have networking and interconnected processors prohibited by law for safety reasons than to be an internet of things self-flying appliance. That would result in a nice simple, admittedly lower performance, far more reliable overall system. Non-networked non-interconnected control microprocessors seem safe enough so far, but making an infinitely complicated hyper optimized system seems in practice to result in planes falling out of the sky too often.
Until we teach spherical cows to fly planes, that is...
If MCAS is the culprit here, Boeing clearly released this functionality without proper vetting and testing. It’s a sad day for safety and for automation. These accidents will likely slow the advance of BVR for drones[2] and more autonomous flight for cargo.[3] The fact these accidents are occurring overseas where the pilots have less training will also buttress the decision for the 1500 hour rule.[4] Like I said, if MCAS is suspected here, probably time to ground the new 737 despite what it will do to Boeing, because right now it looks like they cut some corners, the FAA let them, and families are mourning. Pilot error isn’t really the issue, it’s engineering, process, and training.
[1] https://news.ycombinator.com/item?id=19077371
[2] https://www.altiuas.com/bvlos-mean-commercial-drone-users/
[3] https://www.bloomberg.com/news/articles/2018-10-10/will-you-...
[4] https://www.forbes.com/sites/marisagarcia/2018/07/27/a-perfe...
That doesn’t seem like engineering, process, nor training.
That looks like something connected to large scale corporate profit.
How did Boeing manage to gain self reporting to the FAA?
From what I read MCAS is not even a true safety system, much more like a piloting assist, and should not have had the authority to override pilot input.
It appears that the true stall avoidance system is a "hard" safety system and works independently with much safeguards, and a lot of "overrideability" just for such moments of sensor failure.
So, they added that dangerous system to just let pilots to have the same feel and experience as in old model - an absolutely noncritical issue.
Truly, "UX" people...
I'm sure this is an edge case given all of the successful 737 MAX 8 fights, but it's a catastrophic one.
To go from smooth flying conditions to having the nose unexpectedly dive and having seconds to save your life and everyone on board is almost more dangerous than difficult flying conditions because it's so unexpected.
I'd be surprised if Boeing continues to treat this as a training issue. The plane needs to be better.
There are 350 737Max in service. It's a brand new airplane - it went into service in mid-2017.
For 2 brand new (months old) 737Max to fall out of the sky is not an "edge case".
And Boeing’s procedures on how it disseminates information should be held under a microscope and punished severely.
2 accidents is 2 too many, but it's not like every MAX 8 or anywhere close to the majority or even a large minority are experiencing catastrophic failure.
Sure, it's far higher than what's acceptable and the industry norm, but I'd imagine it's enough of an edge case that it snuck past testing and 99%+ of real world flights.
Given the number of other similarities to the Lion Air crash, I think the parent comment was being pretty conservative in their speculation (at this point, I'd be surprised if it _wasn't_ MCAS related).
"My guess is both pilots fell asleep right after takeoff but before they could get the autopilot engaged, so the plane plummeted to the ground because no one was awake at the controls."
It was clearly marked speculation. If we are not allowed discussions about possibilities we can kill off comment fields right now.
Boeing and Regulators Delay Jetliner Fixes Prompted by Lion Air Crash
Software update, initially expected in January, now likely pushed until April or later
http://www.airsafe.com/events/models/rate_mod.htm lumps max together with other 737 models, which of course turns out to be very safe.