Yeah, I think it would work fine as well. There really doesn't seem to be anything else that could need improving. Of course, even the ratings could still be moderated so that clear (over- and) underratings wouldn't count.
That sounds very interesting, seems perfect on paper my only concern would be people not be able to be objective. Which is something that can never be fixed. The only possible upgrade/change would be to minimize the impact of human flaw, which I see your system doing. . . so kudos.
I don't think I'd want to take it to the thread there but I personally planned out a way of removing this "quality-based voting thing". I discussed it with DrFuji already and he kinda agreed with the idea. It won't be implemented like that but will be taken in consideration if rules are changed.
What I pretty much planned was that everyone could vote for as many hacks being part of the competition as they like. But instead of giving the hacks one vote, they would actually rate them instead. Ratings given between all the voters wouldn't be summed but an arithmetic mean would be formed out of all of them.
So even if popular hacks would catch the most attention, that attention would only balance the rating given for the hack instead of raising it to overcome all the other unpopular, and possibly better productions.