Tribes solves the inaccuracies and other significant flaws in other recommendation systems such as inference engines and collaborative filtering. Tribes solves the problems inherent in existing recommendation systems for “products of taste.” It does this by recognizing that the only relevant information is a single datum: a personal preference expressed in terms of future intentions. Products of personal taste include wine, books, movies, music, cheese, and restaurants and more. Wine is a good example of why current systems fail. It’s nearly impossible for a retail consumer to reliable choose a good bottle that they will like enough for a subsequent purchase. Many retail purchases are so disliked that they get poured down the kitchen sink.
via http://recommendationinsights.com/?p=48