I've been calculating voter-centricity in polls for several years now, so I can't believe I only just thought of the way to re-apply voter-centricity to the things being voted on: Retabulate the album (or whatever) ranking, inverse-weighting each vote by the voter's centricity. I.e., the closer the voter was to the consensus, the less their vote is worth. Then take the ratio of weighted scores to vote-counts, and you get a measure not of popularity, but of cultishness. You probably want to get rid of the albums that got very few votes, but in the 30-voter ILM Metal poll I only had to eliminate albums that got fewer than 3 votes before the results started looking interesting. In the 577-voter Pazz & Jop poll I cut off the albums with 5 votes or fewer, but even the 6- and 7-vote albums are distributed across the score-range pretty well.  

The only real metric of idiot statistics tricks like this is whether you find out anything new by looking at them. In this case, you can make up your own mind. I have named this new stat "kvltosis" (in a combined metal/statistics joke for which possibly I am the entire target-audience), and added it to my Pazz & Jop analysis. If the poll's consensus bores you, perhaps this can be another antidote. (If the poll's consensus thrills you, on the other hand, just mentally invert this list and you have consensus squared...)