Off the Beat: Bruce Byfield's Blog

Hunch.com is one of the few tech launches I've seen this year. It isn't free software, but it is very Web 2.0, full of opportunities for you to contribute and to edit details on the site, so I suspect that the site will have plenty of people from the free sotware community dropping by for its novelty value. However, whether Hunch can retain those visitors or thrive as a business is another matter. I'm guessing that it won't, because it fails to deliver any real value.

The purpose of Hunch is to help you make decisions by asking you ten questions or less. You then receive three ranked answers, plus a wild card, a somewhat unlikely choice that you might want to try if you are feeling adventurous.

If you create an account, you can comment on the accuracy of the answers, or add answers to the existing ones. The site also asks a number of random questions, such as whether you sing in the shower or which of four photos you find most attractive and remembers all your answers so that – somewhat ominously, if care about privacy – it can learn more about you and tailor the results so that they are closer to your tastes.

The site stresses repeatedly that the more people who contribute, the more accurate its results will be. That is undoubtedly true. But, for me, the problem is that the current answers seem not particularly accurate in predicting my tastes. Consequently, I have little interest in the effort I would need to make to improve the results.

The best way to test Hunch, I decided, was to ask questions I already had strong opinions about then compare the results with my opinions. The comparison did not make Hunch shine.
The first question I asked was what author I should read. Right away, I ran into difficulties, because the quiz that followed asked whether I preferred fiction or non-fiction. Given the right mood, I will read either, so immediately I faced a false dichotomy. But, playing along, I answered that I liked fiction, specifically science fiction and fantasy, and preferred famous writers to those with a small following, and those active since 2000. All these answers were more or less arbitrary, since I am extremely eclectic, but I thought they would do for now.

The first suggestion was British comic fantasist Terry Pratchett, whom I like well enough but is not among my favorite writers unless I'm looking for intelligent light fiction. The second was Christopher Moore, whose work I don't care for. The third was H. P. Lovecraft, who died in 1937, and the wild card was Ken Follett, who does fall into science fiction and fantasy by any definition that most people would accept. I counted this trial a complete failure.

For my next trial, I answered questions to learn which incarnation of Doctor Who I would most enjoy watching. The first question was another false dichotomy, since I care little if I'm watching black and white or color so long as the acting is decent. But I soldiered along through some questions clearly intended to differentiate between particular actors, and got a first choice of Christopher Eccleston, which was exactly right. The results also suggested I would like Peter Davison and David Tennant, but reversed the order of my true preferences. The wild card was William Hartnell, whom I dislike. I scored these results a partial success, feeling that I was being a little generous.

Other questions produced equally mixed results. Asking about which GNU/Linux distributions I should use, I was steered towards Debian and Fedora, my two main ones, but they were given in the wrong order of preference. Asking what art I would like produced no answers even remotely close to my tastes. When I went back and added Northwest Coast Art, defining what answers should result in it being among the suggestions, it wasn't recommended. However, Art Nouveau was, making one of eight results accurate. Asking about which pet birds to buy and where I should live in my home town of Vancouver, I received no suggestions that even vaguely corresponded to my actual preferences. These results gave me no confidence whatsoever that Hunch could be trusted to give advice in areas in which I knew nothing.

Even worse, many of the questions seemed to make questionable distinctions, or to leave out important information. For example, why would advice about where to live in a city omit statistics about homelessness and crime?

True, I could stick around and improve the results. But why should I? Hunch is hardly an essential online utility. I am also deeply cynical about crowdsourcing, viewing it as rarely a genuine effort to empower consumers and more often an attempt to make users do a company's work.\

I'm sure that Hunch employees or fans would say I have a bad attitude. But, considering that Hunch eventually plans to have deals with the vendors of different products, why should I help it to target me more exactly and sell me more things?

But the strongest reason I have for doubting that Hunch has a future is the speed with which ennui set in. -- sheer, mind-numbing I've-spent-too-much-time-on-the-computer boredom. After asking five or six questions, I felt very much as I do after taking several quizzes on Facebook. In fact, several of the decision trees on Hunch already seem to have been added to Facebook as quizzes.

Like many people, I sometimes take these quizzes, but I hardly take them seriously, even when I post flattering results. Do I really have a 100% understanding of women or resemble the Egyptian god Ra, as recent? My ego is as strong as the next man's, but I doubt both results – and that skepticism spills over on Hunch because of its similarity.

I don't know. Before I worked freelance, I didn't have a good record of choosing startup companies to work for, so maybe I'm wrong about Hunch. But my guess is that very few people are going to use Hunch for serious decision-making. It might survive for a while for its amusement value, but when you can get dozens of quizzes elsewhere, why in the long run should you return to Hunch more than a few times? Unless I missed something, the site doesn't provide any real service, and isn't likely to – not without far more improvements than I'm prepared to offer for free.

Comments