It's laughable that people complain about Menendian's very reasonable Q1 sample size yet I'm supposed to care about an invite only 22 man meta single event. For what it's worth I don't even know why we are having this discussion; I did nothing but compliment the deck however, citing an extremely small sample size to try and validate a single card that has dropped off remarkably since 2012 is loose at best.
According to TCDecks there are over 160 hits for "Auriok Salvagers" since 2012. Granted, I started playing Vintage after that, but if that's a 'remarkable dropoff', I'm sorry I missed it. The mirror was a skill-intensive durdlefest of awesomeness based on the many matches I played against Justin Kohler and others in the NE.
Regarding your other points, I for one am not dismissing the online data - I think it's useful and have contributed to Steve's dataset myself. The reason I have not been playing much Dragon or Bomberman recently is due to the infeasibility of these decks on MTGO. I play a decent amount of MTGO and paper Vintage and I tend to spend time on decks that I can play on both media. Both combos are certainly viable and can place in major events, such as Vintage Champs, the prelim to Vintage Champs (JR Goldberg took Dragon to a top 8 out of over 100), or the smaller Asian Championship. Discounting these decks because of the online metagame, or the online metagame because of the absence of these decks, are equally foolish in my opinion.