Originally Posted by
slast
However, I reject the idea that "this is a "trap", because I am asking you to say what will happen after the computer has failed because nobody could think of the scenario, yet I have just thought of the scenario." I am using it as an example of something that neither I nor anyone else thought of, before it actually happened. I maintain that totally unpredicted and unpredictable events will continue to happen to both piloted and any pilotLESS aircraft, and want to know how the pilotless ones will deal with them.
I doubt any programmer is going to be saying 'in this scenario, do this. In this other scenario, do that. In this scenario do something else,' it's likely to be a set of heuristics which they'd then check against different scenarios to verify they'd be handled acceptably. Maybe there'd be special case code for each specific scenario the aircraft manufacturer documented, where they'd determined the best response, but you can't build in code for every possible combination of failures.
But I agree with your final comment. The systems we produce are vastly more automated than the ones they replaced, with self-monitoring and maintenance capabilities. When I get a phone call at 4am, it's usually because of some scenario we never considered, which leads to pathological behaviour until the system shuts down (e.g. some piece of hardware fails, so the servers reset it, but it won't reset, and that triggers some watchdog timer, from which it can't recover without the failed hardware, and then it all goes to hell). It's much more robust, and needs much less human intervention, than the previous generation, but it still breaks. Fortunately, hundreds of people don't die when that happens.