, , , , , , , ,

I was planning to get to this blogging thing a little earlier today, but got sidetracked by a website I read about in the newest edition of Wired magazine. The article that mentions said website is called “Should You Read This Article?” and it’s about Caterina Fake, co-founder of Flickr and now chief product officer of Hunch, the website in question. (The article is not yet online, so I can’t provide a link. Ah, the benefits of subscribing!)

Along with being appreciative of Wired’s feature-length profile of a female techie, which seems to be a rare occurrence in its pages, I was curious about Hunch, which asks a shitload of questions in order to determine people’s tastes and then takes the results and makes recommendations ….

Oops. Just got sidetracked looking for how Hunch makes recommendations. I ended up answering more questions.

In my brief analysis of the site, Hunch is very good at asking questions, makes it easy to answer those questions (allowing you to skip those you don’t like or don’t want to answer), and comes up with interesting correlations such as, “Mud is a gateway drug: 46% of moms who used a fake ID got really muddy as a kid. Just say no!” The site also shows the correlations between users’ answers and the famous Twitterers they follow. Here’s the page for Stephen Colbert.

What’s puzzling to me is the recommendations. When I type a topic into the recommendation box at the top of the screen, a list of seemingly random items related to that topic is returned. I’m not sure if the list is truly based on the responses I gave to the questions I answered (and I answered a lot of them) or not. For example, I entered “music” in the recommendation box and while a handful of the 393 recommendations returned were albums I already owned, most of them were from bands I’d never heard of. And nowhere among the 200 or so that I searched did Dave Matthews Band or Nine Inch Nails appear. You’d think they’d have popped up somewhere.

When I searched for recommendations on books, a number of familiar titles appeared. In the drop down box provided for this search, one of the selections was, “Should I or shouldn’t I write a book?” The recommendation that came back was

Yes, you should write a novel

Yes, you should write a novel

You may just have what it takes.

This was the result before I answered a series of questions about writing. After answering the questions that appeared alongside the result, I was given a 99 percent probability of success. As I’ve already written a couple of books, I’d say this is accurate.

My hunch is that Hunch will take more exploration. If you’re interested in checking it out, you can sign in to the site using either your Facebook or Twitter log-in. I chose Twitter because there’s less personal data associated with it. Be forewarned. If you start answering questions for Hunch, you may have difficulty stopping.