Crowd-sourced decision-making helper Hunch.com is out of private beta today. The site helps people make decisions based on how they answer questions from user-created quizzes (See our hands-on from April). Anyone can create these, however the strength of the site is that they can be linked to credible, outside sources to back up any advice or claims. There's also an algorithm that learns more about you as you use the site. The more it knows, the less of the quiz questions you have to answer to get valid recommendations.
Hunch says you only need to answer 10 sample questions for it to start delivering customized recommendations. However, for some topics, such as personal health or survival situations, it's worth going through the motions of answering each individual question. The efficacy of that depends on how well the quiz was put together though. The way Hunch handles that is to put newly-created quizzes into a holding tank called "the Workshop" where other users can bang on it, and provide feedback or fixes. If enough people like it, it graduates to being a part of the site. Otherwise it gets put into the "scrap heap" where you can see the titles and creators, but not the quizzes themselves.
Sites like Hunch get a great deal of traffic from search engines, which is where it may have a leg up on the competition. Since each quiz is made up of multiple questions, there is a higher potential of it showing up in searches that go outside of the main question. Hunch can then drive people to single topics, or topic pages where the site can sell targeted ad units. As it stands, the site makes most of its money off of referrals to commerce sites like Amazon, and sponsored links on result pages.
I'm still not sold on Hunch. For some things, like troubleshooting, or medical advice there's real potential for a system that can weed out irrelevant suggestions based on how you answer questions. However, others like "whether or not you should quit your job," or "what should I eat for lunch?" should be something people figure out on their own, or with the help of friends who don't need to use an algorithm.