Is Personalization Always Dangerous?
Two days after recommending ‘personalization’ to a client’s site that wasn’t making use of user viewing history, I watched Eli Pariser’s TED talk about the filter bubble. He explains how personalization online—Google, Facebook, etc.— is dangerously limiting our world view. I panicked, thinking I had subconsciously sipped the personalization potion. After all, much user interview analysis seemed to suggest that personalized suggestions—those that were taking into account viewing history—would enhance the already positive experience these users were having. For this client, I had assumed suggestions of relevant learning topics would be beneficial to their users. Am I wrong? I wonder, where do we draw the line between personalization that is harmless or helpful to our UX and that which skews our world view?
If I search Egypt, as Pariser reveals, I get a different result than you do, based on 57 signals that Google uses to personalize your search results. The biggest issue, he suggests, is that we don’t see what is being filtered out. It’s not as though certain results are ‘grayed out’ or off to the side. We just flat out see a selective list. Now, I want some information filtered the same for everyone – like the news, for example. It is disturbing to think I could be viewing news online and only see travel stories while my neighbor sees stories about riots when we both search for ‘Egypt.’ But, is there information that is actually beneficial to personalize? If I go on Netflix, YouTube, Epicurious, Yelp or Amazon do I want to see the same results as my neighbors? What’s in my best interest? Or what’s in the collective best interest? Is there a metric that we can use that helps us figure out when personalization is harmless and when we should stay away?