I hope you blog more of these practices then. AWS doesn't put this stuff in writing, which is very convenient for them when something goes wrong, but makes it nigh on impossible to build a reliable system on EC2.
I don't think it's an easy problem to solve, but to suggest that the regions won't go down together strikes me as "the Titanic is unsinkable" hubris. I hope the AWS team doesn't share your attitude :-)