Does Crowdsourcing Generally Result In Better Outcomes?
In The Wisdom of Crowds, James Surowiecki provides some background, plus some examples and research that suggests that crowdsourcing does result in better outcomes than individual decision-making and predicitons.
In fact, the question: "Does crowdsourcing generally result in better outcomes?", isn't the correct one to be asking. With respect to pretty much anything, the issue isn't whether it works or not, but WHEN it works, and under what conditions it will work. Almost everything works sometimes. Nothing works in every context, and we'll see that when applied to performance reviews, it's a terrible idea.
So, the answer is that it "may" work in some situations, provided you have the right outcomes or problems to solve, but it doesn't work in many other situations. We should also keep in mind that crowdsourcing, by definitiony, requires a "crowd", or a LARGE group of people, in order to work. One reason is because one of the ideas behind why crowdsourcing works (when it does) is that the crowds bring diversity to problem solving. Small groups are simply not diverse enough. How big is big enough? Who knows.
Of course, if you require a large group, you run into logistics problems, because you have to have some practical way to capture the input of the crowd, PLUS a way to pull all the input together to yield some sort of conclusion. Collecting and using data is a double-edged sword since it has to be "processed" and analysed to gain actionable insights -- knowledge that suggests clear changes in action.
So, the final answer is: It depends on the task, group qualities, group size, logistical issues, and a lot more, so in fact, you can't make any generalizations about crowdsourcing.