Filter Bubbles and DuckDuckGo
Today, DuckDuckGo launched a new ad/awareness campaign around the concept of Filter Bubbles, promoting DuckDuckGo as a way of escaping the filter bubbles surrounding your searches on other search engines such as Google or Bing. My initial reaction was “I’m not sure what to think.” Bubbles seem like a credible potential problem, but DDG seemed to be jumping on a bandwagon on this one without any consideration to the complexity and subtlety of the filter bubble issue.
“Filter bubbles” are the name given to Eli Pariser for a potential effect that we’ve been aware of in the recommender systems community for several years (under terms such as “balkanization” and various diversity-related issues). The basic idea is this: as web services become more and more personalized, using recommender systems to tailor your experience to your interests, likes, and dislikes, you become isolated in an echo chamber of your own thoughts. If your news service decides what to show you based primarily on what you like reading, you may well get a one-sided or otherwise narrow view of current events. You may not come in contact with other views as often, and perhaps forget they exist. If your music service only plays music it knows you like, you may not find anything new.