All recommendation engines suck (I'd be happy if you found an exception) and hence any startup wanting to make one needs to actually make good recommendations. Every single one I have tried just wastes my time, and I don't see anything in this populistic description that indicates that any of these companies will succeed.<p>It's like promising fusion or general AI in 10-15 years. They keep disappointing me.<p>I'd like to also note that the Netflix prize to date has been unable to nurture real improvements in how these algorithms perform. The way I see it, some paradigm shift needs to happen before this kind of technology comes of age. Automatic recommendation is right up there alongside automatic forum moderation, general pattern recognition and natural language processing for things that no one ever is able to pull off.