The idea of voting on the quality of posts is obviously ingenious. However, why did Jeff and Joel decide for the up/down-vote scheme? Wouldn't a scoring system1 be more helpful?
My thought is, if few people but specialists find a low-view question awesome, it will still end up with as little total (up-minus-down)-votes as a controversial high-view question, while something like List of freely available programming books with a broader audience get's up to 100x as many upvotes.
Yes, obviously more upvotes indicate more popularity, but the one thing that the system does maybe lack is a neutral vote. You can't say if someone who looked at a question but didn't vote just didn't feel like it or whether it's really a statement of "this question is kind of okay but not great".
What would be the disadvantages of taking instead a score scheme ranging from, say -2 ("This question may be on-topic, but it still totally sucks") to +2 ("I never knew I'd want to know this, but now I most certainly do"), and instead of upvotes showing the average score2? I mean, obviously the up/down-vote system works great, but I'm still curious about this.
1) like e.g. http://www.newgrounds.com/ uses, but let's add SE's once-only-vote-locking and maybe leave the high-rep-weighting away
2) and maybe still a total score more or less equivalent to upvotes, but the average would tell more about neutral votes