If you are a casual follower of the story of the crashes of two Boeing 737 MAX airliners, you are probably aware that a sensor malfunction caused an automatic software program to mistakenly point the plane’s nose downward to prevent a stall, overriding the pilot, and sometimes putting the plane on a crash trajectory. If you read a little deeper, you know that pilots often had less than a minute to take back control from the flawed automatic system, and were not adequately trained in how it worked.
Yet another case of allowing businesses to regulate themselves, resulting piles of charred bodies.
The AirBus is higher off the ground meaning easier to land. In traffic when planes are within 1000 feet the AI system is required thought to be more reliable than pilots. Pilots are trained today to trust the AI and use it often. It was AI that crashed the planes with one pilot 30 and one 25. Trump is a symptom of the disease.
Mr. Kuttner -
Good topic. Thanks. It’s about time we looked critically at delegation of authority.
Regulatory capture - check.
Bipartisan, going back to the late 70’s - check.
This high level of delegated authority was mandated by Congress in FAA reauthorization bills - check.
This regulatory drift is a predictable failure of neoliberal orthodoxy - let markets solve all our problems - government is the problem - check.
The connection to Donald Trump - ummmm, check, although we would be right back here if Hillary Clinton had won.
For the record, angle of attack sensors are needed to fly the airplane. All airplanes have them. They are not part of regulatory capture. It’s jarring and distracting every time you mention them.
If you can, please go back and delete every mention of the sensors, starting after noting that sensors failed in 3 three flights in question.
We have a Republican Duopoly in this country.
When will We, The People have the courage to vote 3rd party?
Kuttner closes with:
“The growing and accurate sense on the part of ordinary people, that the rules were rigged on behalf of the wealthy and the powerful, led a lot of voters (who knew little about the details but experienced the stunting of their own lives) to vote for You Know Who.”
Kuttner will never acknowledge that his Democrats, also “rigged on behalf of the wealthy and the powerful,” made Trump possible by selecting the inspiring-to-no-one H Clinton in 2016, who spent her campaign traipsing around with Goldman-Sachs.
Just for another check. As far as I know (please correct me if I missed this) no official determination has been made of the causes or cause of either crash. And I am someone who was into the practice of “safe software” in the late 1970’s or so when it was clear software would be controlling more and more machines (at that time I think my attention was on controls for building elevators although fly-by-wire was already part of aviation, especially with high performance military aircraft).
That is important because jumping on what looks like an obvious cause (as it seemed/seems to me also, right from the start) and fixing what you think is the problem will lead to the problem still lurking and ready to strike, if the cause was somewhere else.
An example I remember as the main programmer in the mid 1990’s for a program converting CAD files was a seemingly obvious fix for an object positioning problem (matter of a minus sign that should have been a plus sign - or vice versa, don’t remember which now). I made the fix and that seemed to work for a while until a different drawing showed the seemingly same problem. So did that customer have the previous edition of the software or did we somehow lose the fix? This time I found what turned out to be the final fix in a deeper layer of software. So I removed the previous fix (+ or - sign) and made the fix at an earlier stage in processing the blocks. That fixed it for all example files and no other customer ever brought this up again.
That said, when I saw the existence of the MCAS and the reason for the MCAS (maneuvering characteristic augmentation system), as a software guy, I knew that software was being used not as part of the design but as a kludge, similar to using a shim to fit parts instead of making them to the right size to start with.
That calls into question basic judgement calls regarding this design. So, while I’m noting that we should not go off too far assuming we already have a solid hand on the reason for those two crashes, I still regard the use of software to skirt around good physical design to be questionable, regardless of the final determination of cause(s).