So doing an actual proper A/B split test is next to impossible due to a number of different factors but time and time again one of the quickest ways I have been able to boost SEO traffic to a site without doing a single thing to the rankings was to implement things like the authorship markup or other kinds of Microdata that triggers the rich snippets.

I totally take your point on board about finding the results looking like ads to you but from everything I have seen to date you would firmly be in the minority. Don't under estimate how "not informed" most people are. There is still a very very sizeable portion of users who can't distinguish things the other way around and think that the ads are organic results.

But if you are going for increasing traffic across the board and you don't specifically target some niche like say the HN crowd who find it weird or spammy, I would say Authorship and Rich Snippets are absolutely the way to go.

My only caveat is that I wouldn't be putting it across the whole site. Like JitBit have done with the stuff on their product pages. That might legitimately confuse some people in certain instances and it's really not what it was designed for in the first place. Blogs and news content though is perfect and as I said, works consistently well in all of my experiences and those of everyone else I know in the industry.