The OP brings up a very old story in fault tolerance, reliable systems, and systems management: In practice a very significant fraction of all outages are due fairly directly to those famous words of the HAL 9000 "human error" or people in the server farm 'bridge' or the network operations center (NOC) doing the wrong things.