TheOrange

Again, hate to be that guy, but in my experience last year most of the successful applicants will have published at least one paper. Depending on what subfield you work in, even 5 or 6 (computational work, for example, typically publishes faster).

It's JohnS Hopkins. Making that mistake is the kiss of death for that school. Each program is going to be slightly different in terms of how they handle international applicant. You should defniitely send some emails and make some calls to find out. And of course it'd be better if you have a connection. Instead of looking blind for programs, I would suggest you talk to your research supervisor first and see if he/she can get you any connections.

You should definitely try to retake it if you can. As it was mentioned before, the low writing score will raise some concerns. Whether that's worth waiting a whole year for, that'll depend. Can you explain your way out of it?

Hate to be that guy...but those stats are pretty typical for some of the top schools.

The one thing I've learned from all these application process is to never count yourself out. But the lack of guide material on this fellowship is pretty annoying. Unlike NSF, it's a bit more difficult to pin down on what they want.