Hacker News Reader

Comment by phire | Hacker News Reader

Certianlly not just a single flaw. They found and fixed multiple other flaws while checking though everything for re-certification.

In my opinion, the 737Max is probably one of the safest planes right now, simply because of the amount of scrutiny it's been though.

View on HN

Boeing whistle-blower says proposed 737 Max fixes aren't enough.

179 points by bookofjoe 50 days ago

https://news.ycombinator.com/item?id=24543411

Unless, of course, it was fixed and scrutinised by organisations which still have the same systemic flaws as the first time around.

Until the sociotechnical systems that do the work change, they will keep overlooking problems and making similar mistakes, from a high-level perspective.

Do you mean the same regulatory organizations that have reduced the US airline fatality rate by 50x over the last 50 years to the point where it's now over 100x safer per mile to fly than to drive?

I mean the FAA certainly has some flaws, but it's important to keep things in perspective.

Don't get me wrong. If I'm going to trust anyone to learn from their mistakes, passenger aviation are the ones.

I'm just saying that when it comes to complex systems, it's not evident that someone who said "I've looked at it and it checks out", the thing fails, and then the same someone said, "Okay, now I've really looked at it and it checks out again" that it's now safe. There's more nuance to it than that.

Whether or not that nuance has been taken in consideration by the FAA this time I really am not in a position to say.

A caveat here is that the modern FAA is very different than the FAA of even 20 years ago. In response to decades of stagnant funding, 'the beast' is fundamentally unable to fill the same regulatory role it was intended to do - to all of our detriment [0].

[0] https://www.washingtonpost.com/investigations/how-the-faa-al...

I think the passenger miles statistic is baloney (calculated by multiplying the number of passengers by the number of miles travelled), and I'm a bit sick of seeing it trumpeted everywhere to claim flying is vastly safer than driving. The collective risk of everyone on board the plane is meaningless to calculating risk to your personal safety. At least for me, the fatality rate per journey, per hour of travel, or at the very least per mile I myself am travelling, is far more relevant. According to one source, flying is actually three times deadlier per journey than driving (fun fact, this is the metric aviation industry insurance companies use), but about four times safer per hour travelled. So flying certainly can be safer than driving in the right context, but these crazy "100x safer than driving" stats are kinda useless and take a lot of "creativity" for airlines to arrive at.

source: https://en.wikipedia.org/wiki/Aviation_safety#Transport_comp..., citing Beck, L. F.; Dellinger, A. M.; O'neil, M. E. (2007). "Motor vehicle crash injury rates by mode of travel, United States: using exposure-based methods to quantify differences"

Why would you care about per-journey statistics? Surely all that's demonstrating is that cars are often used for very brief, non-highway travel.

When the question is of how to travel between cities or states, why do you find the relative safety of low speed non-highway driving to be relevant?

Maybe because the choice I make as a consumer isn't between going to Hawaii by flight or by car -- it's whether to make a plane journey to Hawaii or a car journey upstate.

So maybe per total hour from door to door is the relevant metric: you tend to make decisions based on comparable travel times, whatever the reason for travel or mode eventually selected.

> Do you mean the same regulatory organizations that have reduced the US airline fatality rate by 50x over the last 50 years to the point where it's now over 100x safer per mile to fly than to drive?

As a layman, per-mile normalizing of the safety statistics feels wrong to me, given the speed and distance disparities between car and air travel (for many trips, the two modes aren't really substitutes).

IOW, 100x safer per mile doesn't look as good when contemplating a plane trip 200x longer than your daily commute.

I'm not sure what would be best, but either time-based or per-flight normalization seem like they would be an improvement. Or perhaps per-person (eg. x people fly per year, y have been in an accident).

Well the engine placement hasn't been changed and that was the underlying problem that lead to everything else.

To be a bit more nuanced, it's not that the engine placement was the problem, rather it was that Boeing made software adjustments to make the plane fly like a regular 737 without training the pilots in these changes.

The engine is the start of the problems

The fundamental problem is a lack of effective redundancy for bad sensors. They haven't fixed that. Software alone will not fix it. They've just implemented superficial procedures that don't address the real safety issue.

> rather it was that Boeing made software adjustments to make the plane fly like a regular 737 without training the pilots in these changes

No, this is a persistent and pernicious myth on HN – that the MCAS was some sort of “737 Emulator” intended to allow the plane’s customers to avoid pilot training.

In reality, during testing it was discovered that the engines’ new position caused the stick forces to invert in some situations during flight. “The stick forces invert” means that it becomes easier to pull into an aerodynamic stall than push away from it.

This fails a Fundamental Airworthiness Requirement that all commercial aircraft are required to conform with for certification.

Boeing’s “solution” was MCAS – using the stabilizer trim to automatically push the plane’s nose down as it approaches a stall (and therefore putting additional force on the stick that keeps it at all times easier to push away from the stall than pull into it).

The pilot training thing was just that they concealed the existence of MCAS from pilots out of fear that it would prompt calls for additional training.

tl;dr the problem was that the engine aerodynamics made the plane fundamentally uncertifiable without a software hack.

I mean it still only adheres to safety standards of the 70's by using a loophole in the regulations...

It is methodologically unsound. The airframe is unstable. The fixes are software and training, human factors engineering, and hence don't address the underlying problem which is one of physics. I am sure that the measures taken will reduce the rate of crashes, but doubt that they will eliminate them, as you can't exclude effects without first excluding causes.

Reply on news.ycombinator.com