This article's analysis is dross. This is purely a selection bias issue. ANY significantly smaller video site with similarly-sized content will have a higher engagement metric.
This is no different than "deep" cable television channels, or really, ANY form of niche media down to individual book authors.
FWIW, A.C. Nielsen knows this all too well and says as much in their "cultural impact" reports (Full Disclosure: I once worked on A.C. Nielsen software). A fact which is conveniently missing from this article.