Have you read anything by e.g. Sidney Dekker or Nancy Leveson on "human error"? It sounds like you're picturing fully automated systems in which case you're ri

Have you read anything by e.g. Sidney Dekker or Nancy Leveson on "human error"? It sounds like you're picturing fully automated systems in which case you're right; flexibility is at conflict with security (as far as we know now, anyway.) But once you throw humans into the loop, allowing them to be autonomous and not rule-bound, command-obeying robots can improve security a lot.