Trying out Hunch
Posted: March 30th, 2009 | Author: Alex | Filed under: Uncategorized | 6 Comments »Caterina Fake, of Flickr fame recently announced Hunch and fairly quickly generated some healthy buzz. Daniel Tunkelang, author of The Noisy Channel, generously invited me to try out the service. I checked it out this weekend and here are some of my thoughts.
Hunch is essentially a crowdsourced decision support system:
In 10 questions or less, Hunch will offer you a great solution to your problem, concern or dilemma, on hundreds of topics. Hunch’s answers are based on the collective knowledge of the entire Hunch community, narrowed down to people like you, or just enough like you that you might be mistaken for each other in a dark room. Hunch is designed so that every time it’s used, it learns something new. That means Hunch’s hunches are always getting better.
Conceptually I enjoy that idea quite a lot.
I worked through a number of different question sequences. Hunch arrived at the conclusion that I should be an engineer, an entrepreneur or a computer programmer. Furthermore, I use grammar at least at a sixth grade level and yes, it is time to switch to a Mac. That’s all pretty encouraging.
The software is of course currently in closed beta and has not been heavily used by large numbers of people. Applications of collective intelligence tend to really play their strengths, once lots of people are involved. I am curious to see how the quality of the content and accuracy of the answers is going to change over time.
A few discussion points:
- What if the actual best answer is not part of the available answer set of a given decision tree? This is mostly a problem if I don’t know much about the subject matter.
- In how far do personal preferences play into decisions here? If someones answered an exhaustive number of individual questions on the homepage, would it make sense to offer some decisions after only one or two specific questions?
- A few times when presented with the possible answers to a question, I did not see a good fit. Maybe I did not like any of them or more than one option seemed correct. Maybe one option seems reasonable in some context, but a different one in a different context. On the one hand, this really is where the advantage if crowdsourcing comes in. On the other though, it leaves me wondering: In how far is this going to be used for complex decisions and how much is the 10-question limit per decision going to matter?
Again though, I do enjoy the concept and will continue experimenting with this.
Indeed, I wonder how much the systems learns about you from the answers.
But my main reaction to the site is that it doesn’t offer me answers to many questions I care about. Maybe I just don’t have enough burning questions to need a tool like this.
And the answers for the few I do have aren’t very satisfying. For example, it told me I should pay someone to help me with my taxes. But it didn’t tell me why.
So, it’s cute and still intrigues me, but still feels to me like more of a magic eight ball than a decision support tool.
Presumably, as more people contribute to the site, we will have a richer variety of topics and decisions to choose from.
I noticed a Why did Hunch pick this? link after a decision was revealed. The presented explanation was not very descriptive though. Essentially, for a few questions: Because you responded like that.
Alex, thanks for trying out Hunch.
To your first question: It’s possible, even probable at this point, that the best solution isn’t there. We hope that users will add them as time passes, and the best for each individual will emerge from the combination of the “hard” inputs (say, how much you will pay, or what your GPA is) the “soft” inputs (I care enough about design to pay more for a Mac) and the array of results. Hunch is very embryonic now, so a lot of things are the equivalent of Wikipedia stubs.
To your second question — yes! Over time Hunch should be able to deliver an “I’m feeling lucky” kind of result for you. That’d be magic.
To your third question — a question back. Do you think there are certain questions that would be better served by *more* than ten questions?
Thanks for your feedback!
Cheers.
Interesting, thanks for your feedback. Regarding (3) – yes, I think I could see cases for that. This may be outside your intended scope, but how let’s take the following question for example: What math prerequisites am I missing for graduate school? Or more general, What knowledge am I missing to understand [any subject/article/situation/etc.]?
I would imagine, if we want a detailed picture, then exploring prerequisite knowledge cannot be easily done in just a few questions. It’s not enough to ask “Have you taken Discrete Mathematics?” Rather individual concepts could be quizzed.
A lot of decisions end up being pretty personal and require knowledge of personal circumstance (which of course changes over time). If I genuinely need information on an area I am not very familiar with, I may not know the best way of asking the question either.
Thanks for the invite Alex. I wrote up a little review of my own after trying it out
http://unseen.ws/2009/04/hunchcom-review/
I share your same thoughts about some of the shortcomings. Hopefully this can turn into something useful because it would be nice to be able to spend a little less time researching some topics, or at least be able to get a second opinion you can trust.
BTW: I do have a few more invites available.