Yeah even at a company with as relatively stable a product as twitter, there are so many processes that go on internally that certainly need someone who knows h

Yeah even at a company with as relatively stable a product as twitter, there are so many processes that go on internally that certainly need someone who knows how to operate them. Another commenter mentioned dead-man's switches, but you don't even need explicit switches, especially if an entire team leaves. There will be parts of twitters infra that no one even knows how to deploy, /even if/ it was fully automated. At this point I would be extremely surprised if the twitter infrastructure was anything other than a car with no brakes. It will be interesting to see what parts of it fail first.