I know a lot of the people at Kink.com (they're a solid part of the local SF tech scene, regardless of what people think of the content) and they have some interesting technology across the board.

However I wouldn't actually use their websites as good examples of best-practice in this vertical. Having disrupted the traditional DVD-based content houses, Kink.com is now getting disrupted themselves a little by the 'tube sites' and they really are not on top of this new era for that industry.

The main issue for video sites - of all content genres - is discovery. The author's premis about 'drawing in the viewer' doesn't really allude to this greater issue. YouTube doesn't really do a great job of discovery - relying only similarity clustering and string search.

I'm not going to link to sites from here, but there are a number of adult content tube sites that have some really quite excellent taxonomy, categorization and deeper search functions - surfacing the content that each viewer wants is a unique experience for each person.

Many of their video player experiences and social features are excellent too - ones that YouTube should really take a look at (in fact I know of YouTubers who gone to work for these sites).

For me, the tube world is yet another example of where the adult website industry is once again leading the R&D in the space.