Interesting analysis, thanks. Maybe it could be expanded by also looking at the average and standard deviation of distinct authors that are being voted for? I would expect that abuse would show up with a lower average number of distinct authors being voted for and/or a higher standard deviation?