More accurately: "desktop-based US website visitors who have the comscore panel software installed"
(Yes, comscore does have their website tagging program (http://direct.comscore.com/ ) , but last time I checked their measurement philosophy was still heavily panel-weighted)
That explains it. When I worked at reddit, Comscore massively lowballed their reports of our traffic numbers, and said that the only way to get them corrected would be to install their tracking bug into the site's HTML.
And it wouldn't be sufficient to do it temporarily, just long enough to demonstrate the truth -- as soon we removed their bug, they would go right back to reporting the proven-wrong numbers.
