Are you more likely to spoon or to be spooned? Are alien abductions fake or real? How good are you at catching fly balls? Hunch.com wants to know, not for its own benefit, but for yours. Launched today, Hunch helps users find information and make decisions by comparing what it knows about them to what it knows about others like them. Hence all the questions: The more Hunch knows about your preferences, style, likes, dislikes, etc., the better recommendations it can make regarding which magazine you should read, which 1980s cult film you might enjoy or what kind of dog you should own.
When you ask Hunch a question, an algorithm built by a crew of MIT computer scientists returns fire with a volley of other questions. Based on your answers, it extracts information about your preferences, lifestyle and more. Each question attempts to narrow the field of possible outcomes, so the more questions you answer, in theory, the sharper your recommendation. With answers to as few as ten questions, Hunch claims it can make a solid recommendation, though you can ask Hunch to go ahead and make a decision at any time. Think of it as Ask.com, but more relevant and annoying.
Hunch also allows users to create profiles so the engine has a head start before it’s asked for recommendations. Inviting users to answer up to 1,500 questions about themselves–Do you live in a rural area, suburb, or big city? Have you ever used a fake I.D.?–Hunch creates a profile based on the responses. When you ask it a question, it already knows if you prefer granite countertops to stainless steel or if you spend more time in business attire or more casual fare. That should translate into better recommendations with each question users answer. For Hunch, it means a demographic database that is sellable.
Now in its infancy, Hunch can fail to surprise. For instance, when I asked it what I should wear today, it asked if I am male or female (male), whether I would be leaving the house (yes), am I going to work (no), and whether I feel snazzy or flirtatious (both). The recommendation: Jeans and a t-shirt. Not exactly earth shattering, but the backbone of Hunch is the notion that the more questions users answer, the smarter and more precise the recommendations will become. The fact that I was already wearing jeans and a t-shirt when I asked Hunch for the recommendation suggests it may already know more about me than I’d like to admit.