Even though this is a 0-1 matrix, another thing to try is PCA: compute the correlation matrix for the players (perhaps subtract the mean first) where is the number of voters.
I guess the first few eigenvectors would show the same patterns you found.

]]>