So first off, after speaking to a professor who has reviewed for NSF before multiple times and is familiar with the process, apparently NSF did away with panels this go around due to the government shutdown, which means their internal people are reviewing our applications. Can anyone verify this? Assuming that is true, what that means for us I don't know. I would assume we are less likely to be reviewed by experts in our fields though.
I had to dig to find an explanation of the scoring, but here it is:
E = excellent
VG = very good
G = good
F = fair
P = poor
Regarding reviewers being mean, I'd assume they just get tired of reading applications and start writing as concisely as possible while still communicating their problems with an application, which I can easily see leading to sounding mean. Someone could just be an asshole, though. I'd imagine it's similar to reviewer comments from academic journals. Sometimes they're nice and sometimes they're mean.