Final report on Alaska Airlines Flight 1282 in-flight exit door plug separation

"What We Found

We determined that the probable cause of this accident was the in-flight separation of the left MED plug due to Boeing’s failure to provide adequate training, guidance, and oversight necessary to ensure that manufacturing personnel could consistently and correctly comply with its parts removal process, which was intended to document and ensure that the securing bolts and hardware that were removed to facilitate rework during the manufacturing process were properly reinstalled. Contributing to the accident was the FAA’s ineffective compliance enforcement surveillance and audit planning activities, which failed to adequately identify and ensure that Boeing addressed the repetitive and systemic nonconformance issues associated with its parts removal process."

Somehow Boeing is happening to the whole IT industry at the moment where AI is forced upon programmers instead of "properly developing software" ...

It's a byproduct of unchecked capitalism. This behavior will continue as long as there are no real consequences for those in charge.

There's always consequences for people in charge! It's just that all of the consequences are related to not-enough-profit, which explains everything you need to know.

My experience is somewhat limited professionally in software (just under a decade), but with very few exceptions I've seen little in terms of genuine professionalism as it pertains to anything that could be called "engineering."

Most design/implementation decisions were basically (or literally) equivalent to "we use Kubernetes cause we've already got a lot of existing Terrraform for it", or "we have React developers." I know real professionalism and maybe even "engineering" practice exists somewhere (I mean it has to, for something rigorously proven, right?), but I've not personally experienced it; I've seen this everywhere, as a consultant and employee, both in the public and private sector.

The number of times I've been on meetings or similar where there's tradeoffs backed by quantifiable data was a handful, at best, so the AI trend makes perfect sense to me.

I really don't imagine with something like Boeing where there's a far higher burden of proof there's discussions around, like, some equivalent subjective thingy like "code smells" or "anti-patterns."

> I mean it has to, for something rigorously proven, right?

I have been thinking about this recently. What are the most rigorous "software actual engineering" fields, or projects?

Autopilot systems in airliners came to mind. Not just autopilot, but FADEC, and other flight control systems. Medical devices? ... Or, are all those teams just winging it as well?

Indeed--the "invisible stuff" ABS/ECU/electronic throttle etc in cars, pacemakers, software in airliners. I imagine things that run on RTOSes fall in this category.

There's a whole world of software stuff that just isn't discussed in public forums/places where you'd usually find information on the internet, unfortunately.

Interesting question. Adherence to formal standards might be a good proxy for this, perhaps. Dunno. Just the first thing that comes to mind.

I am in Medical Devices software. Our main compliance requirements are 21CFR and IEC-62304. I know Automotive has MISRA (more coding guidelines than development process, but still) and a handful of other standards. Aviation follows DO-178x, and so on...

> Adherence to formal standards might be a good proxy for this

A lifetime ago, right out of high school, I was a sort of sysadmin at a decent sized civil and structural engineering firm. I learned a lot about the profession of engineering there. Adhering to standards does indeed seem like a really good proxy.