I've been meaning to respond to Christopher Allen's Using 5-Star Rating Systems post but haven't had time until now. Like him, I see problems in the 5-point rating scale but my solution is somewhat different. Unless I misunderstood his solution, his approach aims to improve how distinctive meaning of each rating score is conveyed to the scorer. In doing so, he assumes the scorer will make a fair intelligent judgement given proper understanding of the scoring system.

My solution tries to focus more on primordial aspects of the scorer, emotions and group behavior, to address the problems of 5-star rating system. The key idea is to use peer pressure to normalize individual rating scores. Peer pressure is applied by displaying a miniature score distribution chart directly above (pressure, get it?) the 5-star scoring UI. In the example shown above, peer group's scores were mostly average and so was the scorer's.

Above are examples of a mostly negative rating and a mostly positive rating. In both, peer pressure is strong so the scorer must have strong emotions to break out.

Here we have an interesting example, a split in the peer group which weakens the peer pressure effect.
There are many subtle factors as well as hints in this solution but I'll leave that for another post later or as discussion fodder for others. ;-p
Disclaimer: this solution has not yet been proven in the field.