I enjoyed this read from Mordy Oberstein. In his post he looks at how Google may be training its machine learning properties to define what good and bad content looks like, before profiling various health sites to understand some more about good and bad indicators in a vertical.
I didn't attend the Pint-Sized Marketing meetup, but clearly I've looked over most of the slides (because I'm a nerd). I particularly enjoyed Matthew Kay's slides on their in-depth content briefing process at Aira. Brilliant!