This system is not exact. Consider YouTube videos have the same conundrum, number of views far out weighs the number of likes or dislikes. As so many people don't search for posts before they post, we end up with A LOT of duplicates, this creates a lot of noise.
The primary or best worded, well documented feature requests get the most attention. Even if they don't have a lot of votes, if the feature requests make sense, or are very forward thinking or just downright cool, they're all candidates.
For posts that have a lot of traffic or lots of off-shoot posts, then we look for "Most Votes" and "Most Views" and try to find all the posts that are relevant and add them all together for an "average"
Thanks for the feedback, it means that some folks are genuinely interested in how this thing works.
For a best in case way this system could work, look at
https://www.ideascale.com