>>And stress, I'm sure. They literally do make life and death decisions. If my program has a bug in it, nobody dies from it.
>
>Until you make programs that do. Maybe software controlling somebody's IV drip and confusing a plus with a minus (just a small bug). Or a program that control's nuclear reaction dampeners in a power plant, where a small bug like that could be a little worrisome. :)

That's why the NRC allows monitoring only. Unless that's changed, it has been a while since I worked on those systems.