For google, all regional domains are counted separately, google.com, .de, .ro, etc., while for youtube, as far as I know there's only .com that's viewed all over the world.

Also, the YouTube.com measure that exceeds Google.com is only 'page views' -- the Alexa 'reach'/'rank' relationship between the two sites is much more similar to the Compete.com numbers that TC thinks is "getting it right".

Perhaps Alexa counts offsite embeds of YT videos elsewhere as 'page views', and Compete does not? Or, Alexa might reasonably count clicks within a single long-displayed page (paging through comments, starting/stopping/replaying video) as multiple 'page views' even without a reload.

Compete's numbers on pages/visit (16 youtube.com :12 google.com) and visit duration (16 mins : 6 mins) suggest YT 'views', if counted any differently, could way outpace its 'reach'.

And while journalists and techies might visit Google dozens of times a day, but YT far less often, I can believe there are plenty of people for whom the relationship is reversed. The TC view -- "this must be wrong!" -- could be inside-the-bubble thinking.

Alexa's stats have issues -- but TC's analysis is pretty flimsy, and could be missing something interesting in its haste to mock the Alexa totals.