When he was executive director of the progressive advocacy organization Move On, Eli Pariser had the chance to meet lots of fellow liberals. But he had fewer conservative friends, and he worried he was missing out on their perspectives on political and social issues. So he made an effort to add conservatives as friends on Facebook, hoping to get a mix of left- and right-wing perspectives in his Facebook News Feed.
It didn't work. Over time, Pariser noticed he was getting fewer stories from conservative friends in his Facebook feed. Facebook's EdgeRank algorithm saw that he was clicking on links from liberal friends more often that those from conservative friends, so, gradually, it personalized his News Feed to prioritize the topics he'd showed the most interest in. Despite his best intentions, his Facebook community became a "filter bubble." Pariser's The Filter Bubble: What the Internet Is Hiding from You (Penguin Press) looks at the personal and political implications of internet personalization. We talked about what individuals, companies, and governments might do to deliver us from overfiltering.
I've been seeing ads for your book all across the web, and they started after I did a couple of biographical searches on you. Is this sheer chance, or are you using personalized ads to target ads about the book? And if so, do you want to address the possible irony there?
It's a little self-reflexive joke, and if you click on the ad, it explains to you a little bit about re-targeting, which is the phenomenon that the ads use. It essentially means that you're putting a cookie on someone's computer if they're interested in something, and then you're showing them ads about that thing. I thought it would be a fun way of making visible and poking a little fun at this mechanism that is usually used without any explanation or 'We're-targeting-you-for-this-reason.' It's mainly been my more technologically literate friends who have connected the dots. There are all these different ways in which companies are tracking people that they are very unaware of. And part of what I'd like to do with the books is help people understand how that happens and why it happens and what the consequences are.
So this is an explicit use of a technology that you find disturbing, to point people to why they found it disturbing. My experience was that it's an internet ad, so I didn't click on it, and I assume that no one else does. There seems to be an assumption in the book that we're headed toward this age of personalization because it's so much more effective than an untargeted age. Do we have good evidence to support that? What's the difference between targeted and untargeted ads? And further than that, what's the difference between targeted and untargeted content in terms of getting people to participate, follow along, so on and so forth?
On the ads front, the numbers I have are mostly from people who have an interest in selling this technology, so take it with a grain of salt. But you can recapture really significant percentages of customers that have shown some interest in a product if you bombard them with ads about that product after they've left the page. Having had this happen to me a number of times, it definitely turned what was an idle click into something more, where you start second guessing yourself: "Maybe I was supposed to buy those sneakers." The numbers seem to indicate that throughout the book.