Hunch, The Taste Graph
Hunch.com is a recommendation engine based on user provided input. After signing up for Hunch users are asked to answer multiple questions to train Hunch about themselves. Surprisingly, the questions that provide Hunch with its data may be the most interesting part of the site. Hunch makes these questions fun and I easily found myself looking up from the monitor 30 minutes later not having realized how much time I spent teaching the app all about my likes and dislikes. Caterina Fake, the co-founder of Hunch, calls her service a “taste graph” in a nod to the ever expanding social networks such as Facebook and Twitter. Fake formerly co-founded the expansive photo sharing site Flickr so she knows a thing or two about the social web. The genius of Hunch is when it’s decision time. Hunch culls down the fire hose of information and provides specific recommendations based on the information you provide. Are you looking for a new digital camera? Hunch can give you a list of options tailored to your exact needs. The questions that lead to a recommendation often seem completely irrelevant to the topic at hand and that’s part of the fun. Co-founder Chris Dixon states in the Hunch blog “Speaking of Apple, one of the best predictors of whether people agree they should switch to a Mac: whether they like to dance. PC users really are less fun.” I tested the service on some products that I have already purchased and it very rarely misread my opinion. Whenever a user purchases a product based on a Hunch recommendation the site receives a portion of the profits from the sale. Like any service based on user input Hunch gets better the more info it receives both on specific users and as a whole. The service benefits from recommending the best fit for its users so they will continue to participate in the recommendation engine, provide more data and make the service more accurate.
Your taste in art can tell Hunch what kind of shoes you might like.
Hunch also provides some interesting tools to display the power of its algorithms. A “Twitter Predictor” game answers questions about users simply based on information gained by entering in your Twitter user name. After answering 50 questions the predictor was 94% accurate based on my username @esoterictechie. Services that collect large amounts of data from its users will always face privacy concerns and Hunch is no different. According to its privacy policy “your answers remain confidential and may not be viewed by other Hunch users, unless you specifically choose to share your answers.” Hunch says all the right things, but as Facebook is finding out now, saying the right thing and being seen as doing the right thing are not always the same. The fact is that the information Hunch is collecting and the seeming accuracy of its algorithms is a valuable commodity. Marketers begin to salivate when faced with such juicy data with which to target their advertising. Hunch will do well to remember that this information flows only as long as its users trust the service to be a good steward of their online identities.
The more you teach Hunch the better the suggestions.
Providing recommendations is a serious business. It’s one that e-commerce giant Amazon.com has buttered its bread with for over a decade. Hunch reaches beyond commercial products to provide these recommendations for every aspect of a users life. What breed of dog should I choose? What podcasts would I enjoy listening to? What color should I paint the bedroom? These are only a few of the choices Hunch provides help with. The internet is vast and ever expanding. Some estimates have 3-4 new websites being published every second. With this sea of information it could be very helpful to have an accurate aggregator that knows your tastes and reduces the noise. Hunch aspires to be that service. Initially, I am impressed with their attempt.
Tags: Applications, Cloud, Hunch, Recommendation, Search, Social Media
