I understand that the first D-Mail is what caused the initial shift in AF. It wasn't necessarily the contents of the message, but the fact that SERN was able to identify that Okabe had a "time machine".
What I don't quite understand is why that particular D-Mail caused a shift in attractor fields, but the others after that didn't. Does the change need to be significant before the attractor field changes? I know that there are numbers associated with each AF (0-.99%, 1-1.99%, etc). Does that have anything to do with it?