Use Hunch To Power Personal Recommendations

It gets exciting when the computers start getting smarter than the concierge at the Park Hyatt. Hunch is seeking to prove that evolution. Hunch bills itself as a “taste graph” of the Web personlized to your specific interests. In essence, Hunch offers a decision engine powered by your own preferences and behaviors on Hunch’s site as well as other, linked sites. The more you use Hunch, the more it learns about you, the better it can answer questions like “what is the next car I should buy?”

It all started as a machine learning experiment by a bunch of MIT-ers. Hunch, the current iteration of the machine, improves its IQ through community participation in the site, as well as your own interactions with it. Hunch strives to provide users with an “educated” response akin to what a panel of knowledgeable experts might provide or hours of internet searching might yield. From the site:

Contributions can take many forms. When Hunch makes a recommendation, it will also show you why it proposed what it did. If you disagree with some of the reasoning, you can correct it. If you think Hunch missed asking a crucial question, you can submit one. And if you think Hunch is missing a good result, you can add that, too. Hunch collects and organizes all this input so that it becomes smarter for the next user.

You start by answering twenty random questions so that Hunch can establish a “taste” profile for you. You get some off the cuff suggestions, which I found pretty strikingly on the mark. As you work within the site, you, as a user, refine questions and topics, making them smarter, and in turn, Hunch smarter. The more questions you answer and topics you explore, the better Hunch will do for you. You can respond to Hunch’s suggestions with your own critique as to their usefulness to you, which further aids the process. So, there is certainly a human element to the education, but this element is fed into the machine to feed the process.

For what it is worth, I kept answering until I had blown through about 50 questions. Hunch then recommended to me that I read the New Yorker Magazine (my favorite) and watch Monty Python’s Flying Circus (yes, I can probably quote better than 50 percent of the lines).

Hunch has to appeal to the masses, so you can follow others and there are ways to increase your social cred within the community and earn badges. All trappings aside, the idea that my computer can help me quickly solve the mundane and the more meaty questions seems very intriguing. Bing is seeking billing as a “decision engine”, but uses human curation to aid its process. And, while Quora opens the door to problem solving via a wiki-like community, it too rests on a human foundation. Hunch is offering to do the same with powerful algorithms supporting its decision engine. Fast forward a handful of years when the machines get more and more fine-tuned and we are turning to engines like Hunch to suggest service professionals (who have maintained enough of a web presence to hit the engine’s radar).

Man or Machine? Not sure yet, jury’s still out, but if you hear me asking Hal to open the Pod Bay doors, and suggest a mixed drink to serve at my next cocktail party, you might suspect I am following my Hunch.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s