Personally I have trouble believing in some "political" conspiracy to regulate ratings.
A great wallpaper takes skill to create, don't get me wrong. But you can't really compare the ratings on windowblinds to wallpapers. Wallpapers can easily be seen for what they are. Indeed, are probably the only thing you don't need to download to rate knowledgably. A good windowblind, though, will give you much more than any screenshot, in animations, mouse-overs, and compatability with one's particular style, and so on. This helps explain why seasoned Windowblinds creators get higher ratings--their work is better in a real sense through their experience and detailed knowledge.
Ratings are useful and interesting, I feel. They serve a purpose in ensuring the best is shown to casual visitors. And they can, if given a chance to even out, give some feedback to creators.
The biggest problem with ratings is that not enough people do it. You saw a "drastic" change, but only four ratings were involved. Just guessing that the wall had about 100 downloads when it was pulled suggests that about 1 in 25 downloaders rated it. Few skins will get enough DLs for a 1 in 25 sample to be fairly representative. The catch is, usually you want someone to try out your skin before rating. And if they do that they have to come back to rate. Which they will do only if they really like or dislike the skin--which can mean widely varying scores that will sway the overall. Which is exacerbated again by too few ratings.
Also, there may be a lot of selective memory in these ratings complaints. No one posts a thread: "Quitting because my rating went from 2/10 to 6/10 in twenty minutes!"