We're going to blog about it which should shed more light on methodology -- TC just linked to the infographic. But yeah this was across the entire Disqus network, ~600 MM monthly UVs and ~60 MM commenters, and I believe "pseudonyms" include proprietary site logins + social media handles besides FB.
The major insight was that pseudonyms yield significantly more quantity and quality on the network. May seem obvious to you but wasn't so for many, including a number of us who've been with Disqus for years. The conventional wisdom was that anonymous comments would rank highest in quantity and FB comments would score highest in quality.