[yoda222]

> How many of the hull loses were due to pilot error (e.g. AF 447)?

He is counting 11 hull losses, which is the correct number of hull loss ( http://aviation-safety.net/database/types/Airbus-A330/losses ), but he didn't bother to look for anything more than that. Out of these 11 losses, four were due to ground attack by armed groups (Two in Sri Lanka in 2001, two in Libya in 2014. These are the one marked "C1" in the page)

Pretty sure that these should not count when assessing the security of a airplane (at least a civilian airplane)

[jmgao]

And looking through the rest, two of them were "chemical leakage/fire in the cargo bay" and one of them was obvious pilot error ("fell off the runway").

[tialaramex]

"pilot error" isn't good enough. Good design and engineering can mitigate human failings, bad can exacerbate them.

The only accident investigation I've ever read (I read a lot of these) that actually had no broader recommendations for safety, was one where two fishermen used enough heroin to render themselves incapable of operating their boat safely and they drowned. The investors concluded that yup, taking so much heroin that you can't operate the boat is a bad idea, don't do that. Heroin is, of course, already a class A drug and you're not supposed to use it at all while operating a boat. No new recommendations.

Back in software, it's theoretically possible to do PKCS #1 v1.5 decryption safely but you're going to keep seeing Bleichenbacher Oracles until we stop using it in online protocols, because doing it correctly is hard. So we should just stop doing it altogether.