Hunch: A Real Decision Engine

Caterina Fake’s enthusiasm for her latest project is contagious. Hunch is now open to the public, but it’s still three days pre-launch when we talk and, at this point, about the only uncertainty is how best to describe it in all the media interviews she’s doing. Phrases like “Q&A community” and “knowledge tool” are mentioned, but “decision engine” is what the Hunch team settled on some time ago.

And, for a couple reasons, the timing appears to be perfect: First, Microsoft is throwing around $80 million to promote Bing.com as its new decision engine. “We owe them one,” Fake says. “Bing has done Hunch a big favor. They’ve popularized the idea of going online to make decisions.” Second, and more importantly, Fake says the web is ready for an idea like Hunch — a collective knowledge system that gets smarter as more people use it. “Five years ago, maybe even three years ago, we couldn’t build a product like Hunch,” she says. “Hunch had its way paved by Wikipedia and Yahoo Answers. It’s become more acceptable” to have crowdsourced, collective knowledge web sites.

What Hunch Is

Hunch is a new concept, but it’s probably best described in terms of the familiar. Fake says it’s a hybrid of several different ideas, similar in some ways to both Wikipedia and Yahoo Answers, but different in others. The analogy she uses is of a high school student asking questions of a Guidance Counselor. “Hunch gets you to a decision better when you don’t already have a decision in mind,” she says. “It can show you the right questions to ask.”

Greg Sterling’s companion article goes into detail on how Hunch works. In short, it helps users make decisions by guiding them through a series of questions on their chosen topic. Above, the topic is about choosing which island in Hawaii to visit. After answering a series of questions (all created and modified by Hunch users), Hunch provides a recommendation based on your answers, on what it knows about you from other activity on the site, and from the likes and dislikes of similar Hunch users.

Much like Wikipedia, Hunch gets smarter as more people use it. And people who use Hunch more often should have a better experience than casual users; the more Hunch knows about you, the better its recommendations will be.

What Hunch Isn’t

While there’s an obvious similarity to question-and-answer sites, Hunch is a very different experience. “The most common pre-disposition people have is that Hunch is like Yahoo Answers or Mahalo Answers,” Fake says. “But it’s not. We have some hurdles because those sites are successful. People already know how they work. But we’re different.”

Indeed, if you go to Hunch with the idea of asking a question and waiting for other users to answer, you’ll probably be disappointed in a hurry. On Hunch, starting a topic (i.e., “Which island should I visit in Hawaii?”) is more like creating a new page on Wikipedia. It took me a couple hours on Hunch, not to mention a slow, focused reading of the Help page before I caught on to what Hunch is and isn’t.

“There is a learning curve,” Fake realizes. “That’s probably our biggest challenge.”

How Smart Is Hunch?

At the moment, it’s probably best to think of Hunch as a school-aged child. The site has about 40,000 users who created accounts during the invite-only beta preview. The site’s Fact Sheet says there are only about 2,400 decision topics in the system now. Hunch has a lot of growing still to do.

Fake has been here before, of course. She co-founded Flickr, the wildly popular photo community site, and after Yahoo bought Flickr in March 2005, she worked on Yahoo Answers, which launched in December of that year and is now the No. 1 answers site on the web. As Hunch launches today, she has a feeling it can have the same kind of impact, but it won’t happen overnight.

“It might take five years for Hunch to reach maturity. Right now, it’s like Wikipedia circa 2002,” Fake says. “To me, what makes social software great is that it improves over time. Hopefully, Hunch’s knowledge accretes.”