It would probably work but it's only a few hundred megs I'm talking about. I logged onto my Linode and created a 588 meg file with this:

  dd if=/dev/zero of=testing bs=196k count=3k oflag=dsync 

Then I downloaded it and it went fine. So the problem is on Google's side.

The fact that I tried downloading the first archive so many times their service told me finally that I'd downloaded it too many times then the second archive failed around the same rough time frame. Their service is counting the failed downloads as successful, that's bad in itself.