I think that whenever a version is released that substantially addresses the point of a particular vote-item, that item should be wiped clean. Yes, it'll be a pain and it might not recollect all the votes it had before, but consider:
If Toady already addressed it once, he knows there's heavy demand for it. Wiping it and having it shoot right back up to the top tells him more than just leaving all those old votes -- it tells him there's still demand for it. Leaving it at the top doesn't tell him anything.
It's not a popularity contest, so we shouldn't worry about being "fair." The point is to provide information to Toady; I think wiping votes under circumstances that radically change their underlying assumptions (like, say, having massively more underground diversity added) tells Toady more in the long run than maintaining the totals as eternal monuments or whatever.
I find myself in complete agreement, especially the parts I highlighted.
Maybe it is just being used as an example a lot, but I am getting the vibe that people want the UD entry removed now. I assume that is on the basis of its origins in zagibu's Top 10 list (which was completed), and its references in the dev notes. I really don't think that is in the spirit of what this system is for, but I'll do it if enough people ask.
I just want to point out the danger of such precedent: removing a popular suggestion that has not yet been implemented.
If you don't like how affects the apparent pecking order, just pretend it is not there. UD losing 70 votes (some of which are probably register-postx8-vote), does not mean that your favorite topic is going to gain any.
And if you don't like vague suggestions that have no absolute end point, we need to remove the following:
Farming Improvements, Improve Mechanics, More Detailed Religion, Stats Improvements, Improved Resource Management, Full Gfx Support, Improved Site Finder, Improve the Morale System, Stockpile Improvements, Improved Trading Interface, and Interface Improvements.