I tried to understand what all of you said, but it was too complicated for me.
I think that the fixed votes methods (pick1, pick3, pick10, whatever) result in a more or less exponential curve, were few top suggestions attract most of the votes and the rest has to be content with a lot less. This is because people are forced to choose a subset of their preferences, and then they choose those that are also popular among others, because these have the highest chances to top the list. What I'm not sure about is whether more votes increase or decrease the steepness of the curve (probably decrease).
The fractional method and the up/down or /up/neutral/down voting should produce a much flatter curve, because here, people can vote for any number of things. Of course, the fractional method could turn into a pickX farce, because people realize that the more things they spread their vote on, the less power each single item receives. The up/down method though is different, because each up is 1 vote, so it doesn't make sense to vote for only your favourite thing. It's more like a dual/triple partition of the list into things you like, things you don't care about (if neutral vote is included), and things you dislike. It might be pretty hard on some suggestions.
The direct compare method on the other hand is hard to predict. With our 200 suggestions till now it would lead to max. 20000 comparisons PER USER. with 1000 Users voting, this would lead to 20 millions entries in the db. This means it won't be possible to dynamically calculate the rankings on every page load. So it will definately be necessary to calculate a ranking already at the comparison level (like add 1 for every "win" in a pair).
I have the feeling that the direct compare method would be the most accurate in representing the demands of the community, because it kind of let's everyone construct their private ranking, then calculates the average of those lists. Also, it honorates hard work, because those who do more compares also have a stronger influence on the total outcome.
What I also thought of was a fractional ranking, where the total weight of a person's vote is determined by their forum post count...maybe in some limits, but maybe also direct (so that Granite26's vote would be equal to that of over 3000 newbies...wouldn't you like that, mate?).