Hunch.com, the hot startup from Flickr co-founder Caterina Fake and angel investor Chris Dixon, has redesigned to focus on recommending products such as magazines, TV shows and books to users, rather than answering questions about specific topics. Although Q&A features still exist on the site, it now emphasizes its ability to build a “taste profile” of each user based on their answers to a series of questions, which it then uses to create a recommendation page. Hunch’s new focus could put it squarely in the sights of Facebook, which is almost certainly looking at the same kind of “taste-graph”-based recommendations.
In its original incarnation, Hunch was described by Fake as a “decision making” engine, aimed at helping users answer questions such as “Should I buy a Mac or a PC?” or “Should I ask my boss for a raise?” Hunch would ask a series of related questions and then come up with an answer. It also asked some unrelated and seemingly random questions, and compared those answers with other users’ responses to help predict what a user should do. Now, Hunch is spotlighting its ability to take your answers to random questions and build a page of what it thinks are recommendations you will like.
One of the interesting things about Hunch is that it doesn’t just ask you the kinds of questions you might expect, such as your age, gender, whether you use a Mac or a PC, how you voted, etc. It also throws in some more unusual queries as well — including “What kind of french fries would you prefer to munch?” (with pictures of McDonalds, Burger King, bistro fries, etc.), as well as “Do you live in the suburbs?” and “Is it wrong to keep dolphins in captivity and teach them to do tricks? and “Do you prefer concrete or fluid tasks?”
So why the unusual questions? As a recent Wired profile of the company and its founders noted, Hunch’s algorithm has come up with some interesting behavioral correlations, including:
People who swat flies have a thing for USA Today. People who believe in alien abductions are more likely than nonbelievers to drink Pepsi. People who eat fresh fruit every day are more likely to desire Canon’s pricey EOS 7D camera. And respondents who cut their sandwiches diagonally rather than vertically are more likely to prefer men’s Ray-Ban sunglasses.
Hunch’s big challenge now is that there are lots of other much larger companies going after the recommendation-engine market. Amazon is obviously already a player in that space, with its product-based recommendations based on your behavior and those of other similar users — and the company just recently added the ability to connect your Amazon account to your Facebook account, which it then uses to make recommendations based on things you and your social network have “liked.”
This is almost certainly a market that Facebook is also likely to tackle, since its open-graph protocol now powers millions of “like” buttons and social-activity plugins around the web (Hunch co-founder Dixon was recently involved in starting an open-source alternative to Facebook’s feature called Open Like). Facebook is already using user profile data to target advertising — and it could well be planning use its new Questions feature to add intelligence to recommendations as well, just as Hunch is trying to. The startup may have the better algorithm, but Facebook has more than half a billion users providing it with taste-graph data.
Related content from GigaOM Pro (sub req’d): Why Google Should Fear the Social Web