Eli Pariser: Beware Online 'Filter Bubbles'

In this special year-end collaboration, TED and The Huffington Post are excited to count down 18 great ideas of 2011, featuring the full TEDTalk with original blog posts that we think will shape 2012. Watch, engage and share these groundbreaking ideas as they are unveiled one-by-one, including never-seen-before TEDTalk premieres. Standby, the countdown is underway!
Watch author Eli Pariser discuss secret censorship on the Internet, and read is accompanying blog post below following up on his talk.

People love sharing lists -- the list is one of the formats that fare well in a Filter Bubble world. So here's a list of five of the most interesting ideas I've come across since I gave my presentation at TED and published The Filter Bubble: What The Internet is Hiding from You.

1. Who owns the right to infer things about you? According to Marissa Meyer at Google, some credit card companies can now use your purchasing decisions to predict whether you're going to get a divorce with 95% accuracy -- two years out. This raises some interesting ethical questions: do companies have an obligation to reveal to us the inferences they make about us? Should you be able to gain access to the fact that your credit card company is betting against your relationship? What about in the health sphere -- if Acxiom infers that you're at high risk of suicide, based on your purchases, does it have an obligation to let you or someone else know? I haven't found any satisfying answers to these questions -- but we ought to start thinking about them more seriously.

2. Transparency's moving in the wrong direction. Imagine a company where every communication is transparently available to every employee. While corruption at the top is harder, overall, the effect would be to empower the bigwigs -- because the kind of private coordination that people use to organize and aggregate power would be impossible. Speak ill of the boss, and you get laid off -- that's how power works.

In an ideal world, I'd argue, transparency would vary with power -- the more powerful you are, the brighter the spotlight on your activities. But what we're seeing now is the opposite: the details of most folks' lives have never been more available to more corporate, governmental or even private citizens. But thanks to the Citizens United Supreme Court decision, the wealthy and powerful are able to cloak their political activities, and there are a variety of services available to scrub private information from the web for a price. We have transparency for the 99%, but not the 1%.

3. Robot journalism. Mostly, I've been focused on the impact of code-based editors on how we consume news. But it's worth noting that drone-like mini-robots are beginning to do some real news gathering as well. Check out this footage from a tiny helicopter piloted by folks at The Daily, or this stunning video from a protest in Poland. It won't be long before every news bureau -- and more than a few amateurs -- are using these things to push past military lines, look in celebrities' windows and generally change all of our assumptions about how video news is gathered -- for better or worse.

4. The difference between curiosity and value. Recently, The Huffington Post tweeted about an article with the headline to the effect of "Guess Which Celebrity Got Into a Horrible Accident Today?" I'll cop to clicking -- HuffPost did an excellent job piquing my interest. But I couldn't tell you which celebrity it was, because I've forgotten -- there was very little lasting value in that article.

These kinds of curiosity-driven clicks are one of the primary signals that sites use to personalize content. But unless they're paired with something that measures the amount of value we take away from a media experience, they're only so useful. And they lead toward a world with curiosity-baiting headlines and no payoff.

What if, in addition to click signals, personalizing websites also sent folks a list of the 50 articles they'd recently visited and asked them to mark the three that gave them some lasting value? A personalized feed that took into consideration not just what we click on but what we take away from it could help us build information diets that are both delicious and substantive.

5. Seven Things Algorithms Do That Humans Don't. As we move toward an algorithmically-edited world, there are still a bunch of things that human editors do better. This Harvard Business Review piece has a bit more detail, but here's the short list: Anticipating what people will be interested in, taking risks in recommendations, giving folks a sense of the whole picture, pairing stories together in a way that adds value, highlighting stories of social importance, valuing content that blows folks' minds and building the kind of trust that leads audiences to topics beyond their core zone of interest.

Oh, and one more thing: As I've been discussing The Filter Bubble, the aspect of the problem I've become most focused on is the Information Junk Food problem. In many ways, the important question isn't just whether you see a diverse set of political viewpoints, but whether most people see anything from the political or civic realm at all. I'm working on a new media project aimed at getting ideas that matter in front of millions of people -- if that sounds like fun to you, maybe you should come work with us.