Hunch.com: a review after 60 minutes of tinkering
Earlier today, I received an email from Caterina Fake, team inviting me to check out her latest web venture. Fake, as you probably know, was a founding member of the photo-sharing community Flickr. Flickr was sold to Yahoo! in 2005 and last summer Fake left Yahoo!, presumably to start her Next Big Thing.
Well that next big thing is here (in a limited, private beta). Meet Hunch.
What’s Hunch? As Fake explained in a blog post:
Hunch is a decision-making site, customized for you. Which means Hunch gets to know you, then asks you 10 questions about a topic (usually fewer!), and provides a result — a Hunch, if you will. It gives you results it wouldn’t give other people.
Hunch is a personalized recommendation engine. But rather than, say, track your purchase history the way Amazon does, Hunch uses a combination of your answers and community feedback. The wisdom of crowds gets coupled with your own quirks to help you make decisions.
Let’s say you’re wondering “What’s the best laptop for me?”
A screen shot of Hunch.com
Hunch.com takes you through a wizard that will, by the end, give you an answer to your question that’s based on your answers. In my case, 3 of my 4 laptop recommendations were Macs because I expressed a preference for the Mac platform, and was willing to spend > $1000.
If the answer Hunch gave you was off — say Hunch suggested a paid text editor, but you wanted a free one — you can offer feedback and help correct the system. Users can also suggest alternative answers.
If you want to really up the relevance of the answers you receive, you can tell Hunch all about yourself by answering a series of questions.
And that’s where things get a little weird and slightly creepy. Each correction you add and every question you answer not only makes the system smarter overall, but — much like your Amazon.com purchase history — reveals a little bit more about you.
The advantage is that the more Hunch knows about you, the better your results should be. But the trade-off is that The Machine, and the Hunch community may end up knowing more about you than you are comfortable revealing. It aggregates yo’ sh*t man. Over time, a profile begins to emerge. Throw in community components like followers and that’s just a whole lot of data about who you are and what you might be about.
Quite frankly, it’s a marketers wet dream and I’m quite curious to find out whether — or, more likely, when and how — Hunch will capitalize on this data. That said, I don’t think Hunch is any more of a privacy or data threat than Friendfeed, Facebook, MySpace, Amazon, or my beloved Twitter. But I would like to see Hunch give users the ability to opt-out of being followed altogether.
I would also like to see weighted questions. Perhaps after you answer each question, but before your final answer is revealed, you have the opportunity to weight how important each factor is to the decision. I’m sure that will require some massive code-fu. But it would also improve recommendations.
Conclusion
I’m not sure why people would use Hunch.com to make decisions instead of doing what we do in real life: ask friends, go with our gut, use Google. I like the idea of a trainable recommendation engine that learns what I want and also enhances the community’s knowledge. I’m just not entirely sure it works as a standalone product, and I’m not sure how much community you can form around it.
Maybe that’s not the goal of Hunch, though. After all, Netflix posted a $1 million prize to anyone who can help improve its Cinematch product. Amazon uses recommendations to boost sales. Recommendations are potentially a big business. It wouldn’t surprise me to learn that Hunch is an experiment in building this kind of trainable technology rather than a product unto itself.