Trust, feelings and instinct are important.
That's why the FAA performed an assessment very early on of who 'touched' the MAX, and their involvement in other models, especially since.
It's why all X documentation to-date is being reviewed.
It's why at least 2 contributors usually here, working on the X, are silent.
They're only important if you refuse to be objective. Everyone can make mistakes at any point, but you can't lie with a mathematical proof. It only takes one mistake in just the right system to turn an otherwise flawless plane into a death trap. That's exactly why relying on trust and testing is foolish. Testing is ALWAYS incomplete. Relying on it is lying to yourself. Prove the design, maintain processes to the spec as proven, and you'll never have a problem. All kinds of manufacturing disciplines already do this. Most of your simpler electronics like toasters undergo basic validations and start shipping with certifications while never going to a testing lab. This is true in China, Europe, and the U.S..
There is a vast difference between a toaster, a tv or some other consumer item compared to say a car. All cars must go through crash testing if only to confirm that as built the results conform to that which the designers saw in the computer.
An airframe is so much more complicated than almost any other item around. The number of systems that there are and how they interact with each other in all phases of flight and flight conditions - let alone in various different failure modes is extremely complicated. I have seen no automated ways to do this fault analysis on such complex systems. It relies on the knowledge and experience of the engineers.
The otiginal 777 may have been able to be tested in 10 months. I do not see how that is still possible as the system complexity has increased.
All cars must undergo crash testing because regulation requires it, not because that testing is objectively useful or going to save lives where an absence of testing would not. If the DOT and car manufacturers had high-caliber mathematicians proving the designs and verifying the proofs (which takes days vs. weeks), then the time to market would improve with no negative impact to quality. This applies to anything and everything that can be made. Sure, the giant stack of papers of passed tests makes Joe Consumer and Manager Mark comfortable, but who is to say the tests are truly comprehensive? Where is the closed proof that the battery of tests is complete?
Nope, even our most advanced commercial airliners do not hold a candle in complexity vs. a Tesla with AI-driven autopilot. A complete order of magnitude apart. In fact in many ways, autopilot of a plane is strictly simpler since you don't have to factor in "idiots on the road".
You not seeing the automation is not evidence it hasn't been or can't be done. That comes down purely to how much you invest in your mathematicians and computational capacity. As long as you can encode the proof, we have the computing power to flesh out the validation of the design against the proof.
You assume increased complexity = increased time. I can hand you plenty of algorithmic examples where the opposite occurs. It also comes down to keeping your battery of tests clean of duplicates or superfluous tests. And in the case of testing the airframes themselves, get enough pilots trained and hire the staff to keep the birds in the air practically 24/7 to keep the data flowing. But again, testing is pointless for multiple reasons: either the design was flawless or it wasn't, and if that flaw cannot be covered by the tests you have, then passing all the tests was just an expensive waste of everyone's time while saving no lives. If the design was in fact flawless, the testing was an expensive waste of time that saved no lives. All it did was provide a feel-good stack of papers and ink to imperfect regulators, managers, investors, pilots, and fliers.
The truth is testing is a lie unless you can mathematically prove the battery is closed and complete. If you are capable of creating THAT proof, you were capable of proving the design had no flaws to begin with, meaning you did twice the work in at least twice the time as needed.
Last edited by TFawkes
on Tue Feb 25, 2020 7:59 pm, edited 1 time in total.