We've been releasing a bunch of sexy Hunch data via the academic API, which has revealed all kinds of great, and often hilarious, correlations such as the fact that people who are frequently thrown out of bars for rowdy behavior have worn cufflinks during the past several weeks. (You won't be able to get to a bunch of these links unless you're logged into Hunch.)

All the sexy data is a byproduct of the larger goals of Hunch and what we're really trying to do is use this data to help people make better decisions. As my friend Chris said yesterday, doing research online is something that super adept internet users, like us geeks, already know how to do, and is how we have been using the web to do our research since it began. We've trained ourselves to think like a computer and enter in just the right keywords to find what we need. We talked about how incredulous our geek (snob) peers are that so many people use Answers sites like Yahoo Answers (which I worked on) and Mahalo. We talked about how circa 1999 a lot of people publishing weblogs snorted at the idea of using "blogging software" like Blogger or Movable Type -- when we had Perl scripts or BBEdit and ftp why would we need anything else?

Hunch is essentially a tool for experts to help non experts -- and when we say experts, we don't necessarily mean people with Ph.D.s, but more often people who have taken the time to do research. For example, a colleague was researching whether or not he should spend the time and money to register a trademark, how much it cost, what the benefits would be. Once he was done, he could build a Hunch topic and share that research with others. Others who had similar experience with it could add their expertise as well. Hunch is re-indexing information that is already out in the world, and on the web, and putting it in the form of decision trees.

Someday, after the system learns enough about people, it should work like a really awesome Magic 8 ball on steroids and just give you the answer you need. It's not awesome yet, but will, we hope, be awesome in the future. On the Transcapitalist site yesterday, I found a post about a Hunch credit card search that's an early indicator of possible future awesomeness:

We find some of these posted on the site, like this one from the Canon PowerShot SD770IS 10MP Digital Camera Pros/Cons, written by Hunch user Kim Rossi: "Took this to see how accurate it would be. My top result was the camera I actually bought six months ago! Well done, Hunch." Hooray! A lot of the time Hunch still fails, but the more that it gets used, and the more different people we have adding tiny bits of data, the better it will get. We need to stay humble, stick to our knitting and keep improving it. Kaizen!

Another anecdotal success story from a new user, awgraham:

This is where we're trying to go.

LINK | 11:27 AM | TB

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .