3 ms·

I'm sceptical how the co-pilot somehow did the single most wrong thing to this aircraft possible (unhinging the wing). There must be some interesting report to

I'm sceptical how the co-pilot somehow did the single most wrong thing to this aircraft possible (unhinging the wing). There must be some interesting report to be had on how design can kill if that is what went wrong. It's a fun exercise think about how this kind of failure can be compensated while always having to ask, what if that compensation also fails. The flight computer could easily work out that the wing can only be unlocked AFTER the violent acceleration only during a complete free-fall, but then it's no problem thinking of a hundred ways that computer could fail. Plus you probably want a manual override anyway to the lock etc. If that was actually my job I'd never sleep easy.

codeulike 11y ago
Its an experimental aircraft doing test flights, I don't think they spend a lot of time trying to make it blunderproof. Theres a million things the test pilots could do to endanger the craft. I imagine they mostly just assume that the pilots will not mess up.
- xanderstrike 11y ago
  Exactly. A pilot could also fly it into the ground, or into a building, or into WKII. Do we need physical safety mechanisms to prevent this? Or can we depend on pilot training?
tjradcliffe 11y ago
More importantly, there are a bunch of scenarios where the wing needs to swing that don't involve the pre-conditions you've specified. No one knows what they are, but we know they exist--because they always do--and the last thing needed is for the computer to tell the pilot, "I'm sorry I can't do that Dave" when the pilot tries to do something it has been programmed to not allow. At that point a software developer's imagination is flying the aircraft, which is never a good thing. That we can't imagine those scenarios right now is irrelevant to this point: we can't imagine the correct physics beyond the standard model, either. That doesn't mean we won't discover it as we gather more data. But in the aircraft industry "more data" often involves dead people. We generally find it easy to reason like, "In context X do (or don't do) Y until condition Z." The hard part is figuring out what consquences that will have in context P, Q or R... each of which is likely to have a very low probability, but one of which is almost certain to happen eventually.

codeulike 11y ago

Its an experimental aircraft doing test flights, I don't think they spend a lot of time trying to make it blunderproof. Theres a million things the test pilots could do to endanger the craft. I imagine they mostly just assume that the pilots will not mess up.

xanderstrike 11y ago
Exactly. A pilot could also fly it into the ground, or into a building, or into WKII. Do we need physical safety mechanisms to prevent this? Or can we depend on pilot training?

tjradcliffe 11y ago

More importantly, there are a bunch of scenarios where the wing needs to swing that don't involve the pre-conditions you've specified. No one knows what they are, but we know they exist--because they always do--and the last thing needed is for the computer to tell the pilot, "I'm sorry I can't do that Dave" when the pilot tries to do something it has been programmed to not allow. At that point a software developer's imagination is flying the aircraft, which is never a good thing. That we can't imagine those scenarios right now is irrelevant to this point: we can't imagine the correct physics beyond the standard model, either. That doesn't mean we won't discover it as we gather more data. But in the aircraft industry "more data" often involves dead people. We generally find it easy to reason like, "In context X do (or don't do) Y until condition Z." The hard part is figuring out what consquences that will have in context P, Q or R... each of which is likely to have a very low probability, but one of which is almost certain to happen eventually.