This is perfect for a project I'm working on. One component downloads zip files (10s to 100s of MB) from websites for indexing, and I archive them on S3 because it's just easier/faster to deal with them from there on EC2 instances that do the processing.
I don't need super redundancy, because I can always download the files again, and so RRS would make the S3 bill even cheaper.
Personally, I'd feel better if they'd tack on another 9, but that's not a bad deal for an enormous amount of uses out there.