i202 Fall 2010 School of Information, UC Berkeley

Taking us back to the old ways of gathering information, web search companies are now tapping the power of immediate social circles instead of social networks to suggest answers to any question we might have. With virtually endless information available to us on the internet, companies have been trying to hone in on how best to present us with information that interests us, and how to make money from it.

Hunch.com, a New York based startup, is attempting to 'personalize the internet' by categorizing our tastes and preferences after a few questions and some behavior monitoring. It's machine learning algorithm will then be able to compare us with others and suggest things we might like to do/see/buy etc…

Do you like to tweet? Then you probably a fan of the Museum of Modern Art. If you are not on twitter, Hunch believes it is more likely that you'll be at the theater.

Extracting useful information from endless reviews on a product or service can be difficult. Do we believe that reviewers are legitimate, or could they be planted by those with financial interests? Can we be sure that we will enjoy the new bestseller because of praise from completely anonymous reviewers? By honing in on our immediate social network, companies are looking to improve the relevance of search results and reviews. We are naturally more willing to trust the suggestions of those we know. Loopt.com claims that its users are '20 times more likely to click on a place their friends had liked or visited than a place simply ranked higher in results.'

But can our preferences fit into categories? How much information might we miss by filtering out things our friends don't like? There is a trade-off between our comfort in knowing the source of recommendations, and harnessing the sheer power of internet and its information, albeit overwhelming at times.

NYT Article here:

http://www.nytimes.com/2010/09/13/technology/13search.html?_r=2&hpw