How do i quit netflix?

Do recommendation engines provide any social good, or do they just help people to reenforce their existing ideas?

• Take Netflix for example - part of the reason I quit was that they had made it very difficult for me to find new items that they did not already 'know' I wanted. Too bad they were wrong so much! For the last 6 months the only search page I used was 'HD.' I wanted things that were new, not more of the same. Beyond that tho, I am more and more wondering if this line of thinking is fundamentally destructive to both individuals and society. Using automated blinders so that users never happen to come across any content they don't already agree with strikes me as a method for generating ignorance and delusion, dogmatism, and undermining quality discourse in favor of more 'hits.' Do these engines provide any greater social good then convenience and targeted marketing? And if not, isn't it fair to say that their net effect on human culture is ultimately harmful?

• Answer:

  As a researcher in recommender systems, these are exactly the questions that come to mind as recommendations start becoming more and more ubiquitous. Here are some of my thoughts. A recommendation system can be viewed as a tourist guide. A good one adapts to your preferences, listens to your needs and gives you experiences that you could not have had on your own. A bad one takes you to a place where he/she has a vested interest, or worse, cannot even decide that properly. The widespread usage and deployment of recommendation systems is in e-commerce and the entertainment domain (e.g Amazon, Netflix), where recommendations are tailored to your tastes so that you may consume/buy more. And thus experiences such as yours, where one feels that the recommendation spectrum "saturates", and there isn't so much variety after all. Indeed, a completely personalized news recommendation may in fact be detrimental, since it would lead to a "echo-chamber" of the kind that many have been warning about.  But it is not clear if that is the complete picture. Recent research on the Facebook news feed [http://www.facebook.com/notes/facebook-data-team/rethinking-information-diversity-in-networks/10150503499618859%5D, for instance, shows that the Facebook news feed recommendation actually helps in diffusion of varied information, thanks to the weak ties in your network. The "automated blinders" effect  may be, in part, due to the nature of initial research on recommendation which focussed on accuracy--getting the predicted items to match the items a user would anyways like. But increasingly new recommendation algorithms and systems are coming up that focus on other goals such as diversity, optimizing for a network of people, and overall social benefit . For instance, Suggestbot (http://www.grouplens.org/node/105) is a system that recommends articles to edit on Wikipedia--with goals to ensure that articles needing work are matched to the right people.   There's a fine line between recommendation and personalization, and new systems are realizing that. I wouldn't go so far to say that recommendations are harmful to our culture. The internet, and our lives in general,  are burgeoning with information. It only makes sense to take help from computers in dealing with this information deluge. What goals these systems are tuned for, is what makes them better or worse.