[Edit: I worked at Amazon Web Services until recently. Now work at Google. Concern for customers drives everything that happens at Amazon, sometimes to a ridiculous degree.]
This was addressed well by someone from Netflix but I can't find the link. Basically he said that Reddit weren't so much the victim of AWS rather than the victim of not understanding the product enough and not spending enough to buy the right instances and thus ensure their constant performance. The guy from Netflix justified this by the way that the different sized instances are split across their respective hosts and, surprise surprise, if you buy the bigger instances you get more of each machine host and less threat of someone taking you out.
If that were true, you'd think that would be the first thing they were told when they started complaining about crashes a year ago. Certainly, you'd think by the time they had talked to the CIO, somebody would have said, "you guys need to buy bigger instances". We know money isn't a problem for reddit anymore, and all current and former employees have said the problems with EBS just aren't something they can fix by throwing money at.
I would suggest the guy from netflix simply doesn't know enough about reddit's particulars to offer an informed opinion. Either that, or he knows better than everyone at reddit, every support person at amazon, and even the CIO of amazzon how best to solve reddit's problems.
[Edit: I worked at Amazon Web Services until recently. Now work at Google. Concern for customers drives everything that happens at Amazon, sometimes to a ridiculous degree.]