Of course, since the internet is a bundle of server-to-server connections relying on each other, the same outage could happen to other platforms.

Certainly, there are lessons here (mostly about DNS, I suppose). Perhaps 5__ errors should be re-engineered for various different responses depending on the severity of connection loss, or network protocol designers should develop backup systems that work in the event of a series of servers losing touch?