We select a small set of recommendations that lie in the upper and lower tail of the empirical distribution of divergences between a recommendation, and the consensus over the window (-30, -1)-days prior to that recommendation. We classify these extremely divergent recommendations as bold, and...