It focuses on vanity metrics rather than engagement or retention. A lot is written on that topic, so I won't repeat (see Tim Ferris: http://www.fourhourworkweek.com/blog/2009/05/19/vanity-metri...). This article is excellent optimization for link baiting.
And it doesn't matter because most of these sites don't compete against each other. Each of these media sites has a different value prop, target demo and potential ARPU. For example, the value of a unique visitor on LinkedIn is way higher than on Tumblr, but it's a moot point because they aren't going after the same users (sure there will be some overlap, but that's in the same way that I visit HN and FB).
Second, these stats are utterly wrong.
Tumblr had 133M global uniques in that time, according to quantcast [1]. Pinterest had 45.8M US uniques around then [2]. And this completely ignores their quickly growing mobile traffic. Quantcast directly measures traffic by pixeling sites, so I trust their data over Compete in this case.