Hunch

I signed up for hunch, and have been playing with it a lot over the past couple weeks. It’s a fun system, very addictive at the beginning, while you are answering the questions about yourself (called “teach hunch about you” or THAY questions).

How it works is pretty interesting. Topics are the big questions, “What laptop should I get?” etc. Under these topics are questions and results. Questions are the filters that lead to results. Things like “Should the monitor be bigger than 17″?” for example. Results are the solutions. There are two types of results, yes/no, and specific things. In the laptop example, a result might be a MacBook Air. Some questions, like “Should I learn to ski?” would have just two results, yes and no.

Results are “trained” which means for each result someone goes through all the questions in the context of that result and says which specific answers apply to it. For example, the topic “What font should I use?” would have a question, “Do you want serif, or sans-serif?”. The result Palatino would be trained in each question. For example, Palatino, would be trained as yes for that question. The training is done mostly by humans, but if the algorithm sees something it thinks correlates, it will train on its own. The ranking of results is based on a combination of the training, and what people like you have rated things. (people-like-you being based on the THAY questions)

It works surprisingly well really. Even without the training, and questions to guide it, it can do a pretty good job basing it’s answers only on people-like-you data. The team is also very responsive to suggestions and things in the forums, and is iterating a lot. It’s fun to see employees so obviously excited about the product they are making.

If I have one criticism, it’s that there isn’t enough to do, and no great ways to tell people what needs to be done. There’s also no good way to compare work and create any feedback loops. I think these things help build a real community. I’m thinking of this a little like Wikipedia, and maybe that isn’t a correct comparison. Maybe they don’t need a community in the same way really. It’s certainly very easy to use, and they have built the tools for people to correct their own results. Maybe that’s enough.