Scott Alexander has some interesting thoughts over at SlateStarCodex on what makes somebody a prescient and trustworthy contrarian rather than a crackpot. But in order to come to an agreement on contrarianism, there needs to be agreement about what constitutes the scientific establishment.
Scott proposed five (well, six if you take the bifurcated group no. 4) groups, starting with those on the front-lines doing actual research in their fields. I was so tickled by this idea that I decided to turn it into a graphic.