Februs:
Not every person who voted bothered to vote on all the entries and some only voted for one or two.
That's always going to be the case, to some degree. I would be disappointed, though, if too many people voted on some stories without even having read most of the others. I'll be the first to admit that, self-promoter that I am, I rated my own story highly, not simply because it's my own story, but because I happen to like my own writing. But I read every single story in the contest and conscientiously tried to rate every other story relative to how well I thought it was written.
This is a suggestion very late in the process, but is it possible to calculate a story's score as an average of the times it was actually read? For example, if only 125 of 146 voters actually read a story, and those 125 voters gave it 500 total votes, could the score be 500/125, or 4. In other words, a story's score is the average score given to it my the people who actually read it. That way, an author would lose in the ranking simply because fewer people read his/her story? I would understand, of course, if this sort of scoring system is too difficult to implement at the last minute.