How is it even possible for this (global blackout) to happen when the entire operation is highly distributed, both geographically and techically? I mean, even implementing a "red button to shut everything off intentionally" would be hard.
Config update error sounds like a good explanation but everything (mail, news etc) went down simulatenously and globally. Might be a global load balancer / routing configuration error though.
I'm also curious about the people involved. I mean, when I break something when say a couple hundred of people are online on our website, I panic a little. If I break Google, that'd cause a heart attack.