Boeing's 737 Max Software Outsourced to $9-an-Hour Engineers (2019)
industryweek.com
industryweek.com
Excerpt from the article:
Boeing said the company did not rely on engineers from HCL and Cyient for the Maneuvering Characteristics Augmentation System, which has been linked to the Lion Air crash last October and the Ethiopian Airlines disaster in March. The Chicago-based planemaker also said it didn’t rely on either firm for another software issue disclosed after the crashes: a cockpit warning light that wasn’t working for most buyers.
So it was probably non-critical, routine drudgery software which was outsourced.
As a reminder, we're talking about a piece of software that could control the plane's angle of attack, pointing the nose downwards (with the obvious risk that this poses), and which relies on a single physical sensor, a pitot tube, which is known for relatively frequent (for aviation) failures due to ice and other blockages, which is why there are two of them.
If the people who made those choices didn't understand the ramifications of their actions they have no place in aviation. At the very least they should be fired, but ideally they should be brought to justice. Alongside every Boeing pilot, engineer, manager, exec who knew about this, and lied about it.
What i was pointing out is that a lot of articles like this one have been planted in the media subtly hinting and throwing shade on these external companies (who are innocent) thus deflecting from Boeing's responsibility. This needs to be called out every time.
Wasn't the sensor an AoA vane instead of a pitot tube?
a) critical - if it malfunctions, the plane can crash
b) based on something that is known to fail often, with zero redundancy
Both of those they were mandated to disclose, and they didn't, knowingly. Why is none of these bastards in jail?
> Wasn't the sensor an AoA vane instead of a pitot tube?
Yeah sorry, mistook the sensor, indeed it's the Angle of Attack sensor next to the pitot tubes. Both are prone to frequent, for aviation, failure, so the point still remains.
That is the question, right?
I mean I get the NTSB no blame approach in progressing safety practices, but there needs to exist an authority that enforces these practices.
And the worst part is that this gives a bad rep to "IT" as a whole in the eyes of laypeople, even though IT's internal view aviation is that it's full of safety standards, redundancy, reliability and maintenance is paramount, etc.
It turns out the standards were not followed, and starry eyed junior programmers' view of aviation is just smoke and mirrors.
When will we reach a turning point?
TL;DR: Don't competitively incentivize short-term cost savings by strip-mining fundamental capabilities to deliver by throwing human capital to the wind as an interchangeable "fast-food worker" approach.
Boeing most certainly can afford to pay the price especially when human lives are at stake.
For reference, I'm a Sr aero dev pulling 150k.
But yes, the OP is very much right in that companies building actual, physical products don't have the kind of money to burn on engineers that a SV unicorn does...
That income is plenty for me not to have to worry about a thing financially, though. Housing is cheap, transport cheap-ish.
Especially when the code is directly responsible for a physical system.
In my opinion the 737MAX needs a distinct type certification. This would dramatically raise costs, so it is covered up with software. FAA and regulators in other countries don't care. This management and regulator problem is enabled by all off the passengers who pay for seats on this plane.
I don't care if Boeing decided "this revision" of the 737 Max is safe. I insist on only flying in airplanes that the pilots were trained to fly on, without a compatibility layer between the pilots' training and the airframe characteristics.
The crime of the 737Max was that the MCAS fucking shit into the wind and introduced an unbelievably dumb stochastic failure mode (every X seconds MCAS adjusted the stabilizer), and the pilots had no idea what's doing it.
I don't fly often, but I go further. I just don't fly Boeing unless it's a 747.
747 || Airbus || Embraer
My airline of choice is American. I can see what a particular route is flying prior to purchasing tickets. There are enough Airbus aircraft in their fleet for me to pull this off.
The most important is making sure the system works as intended. The code is a way to accomplish that. There are other ways to accomplish that, which frequently will include things like changes to the circuit, changes to the mechanical design, changes to specifications, operating requirements, silicon, or how the user interacts with the system in order to make sure it's working correctly.
That the code was incorrectly specced was the bug, and there it matters whether it was written by an unpaid intern, a $9 engineer, or a senior engineer who has enough system knowledge to know what the ramifications of the system interacting this way are, or allowing marketing to make it an optional fault light, or permitting only one sensor to be used as its input.
While the ultimate failure is a management one, one of the ways that management failed is they clearly didn't get the proper system-level architecture squared away.
Now, if you're reading a sensor and blindly trusting what it says, that can be a problem. And there is where it becomes important that someone with experience writes this. They understand what it means, they know the things that can go wrong.
Sending a story to India that says 'implement autopilot AOA sensor change', then that is exactly what the devs there will do.
Don't forget that the culture in India is different than here. There's no pushback. There's no asking questions. You need to be extremely explicit, and ask and ask and ask to make sure they feel comfortable asking the questions they have to you, someone they see as the one in charge.
But it highlights a general policy of outsourcing to the lowest bidder. This is generally a symptom of badly managed company placing spreadsheet profits above everything and seeing product development and Engineering as a cost center, which given the current Boeing situation, is quite telling.
There are skilled people from India, but rarely in India I'd assume. The market for software engineers in India is not great to put it politely so the talented people will obviously go where the money and benefits are which is US, Europe etc. And yes I'm sure there are exceptions, but probably not too many or do you know anyone actually buying software engineer services from India and paying western salaries?
Pop quiz: name a new program that Boeing has started in the last twenty years that was successful. Acquisitions, buyouts, remnant MD programs don't count. Now, during the post-COVID era, they're somehow spending all of their[1] cash in precisely those sectors that are forecasted to never recover - ever. Congratulations. That takes work. If you want to see this poetry in motion, take a look at what happens to any company bought by Boeing. If you could weaponize this kind of value destruction we'd figure out a way to put it in a JDAM and drop it on Hainan Island.
The MAX software issues are at the very tippity tip tail end of a single one of these chains of idiocy, so I feel uncomfortable just piling all the sins on MCAS. The fact they even had MCAS should have been a gigantic red flag, and would have been in anything like a functional regulatory regime. The FAA PNW offices were unofficial Boeing locations even back in 2008.
[1]rapidly shrinking reserves of
> Boeing has categorically said that MCAS
It doesn't make sense due to low prices for everything in their country.