I'd say less than that. Remove all https traffic, js frameworks, identical images, video, newspapers and major blog entries, and I'd guess the remaining text traffic must be 1-2% of the global traffic.
One thing I'd guess too is that the NSA must have the mother of all password databases. Given all the login forms that went through unencrypted and still do. If I was the NSA that's the first thing I would have captured.