Editor’s note: This piece by TheAtlantic.com editor Adrienne LaFrance runs in the Fall 2017 issue of our magazine sibling, Nieman Reports. We’re sharing a sneak peek with Lab readers.
It took a terrorist attack for Google to enter the news business.
On September 11, 2001, after hijackers crashed two commercial jets into the World Trade Center as well as a third plane into the Pentagon and another into a field near Shanksville, Pennsylvania, Internet users turned to the search engine for information. Again and again, they typed in terms like “New York Twin Towers,” but found nothing about what had happened that morning. Google’s web crawlers hadn’t indexed “Twin Towers” since the month before, which meant every result that Google returned was, given the context, totally and painfully irrelevant.
Google quickly set up a special page for “News and information about attacks in U.S.,” with links to the websites of about four dozen newspapers and news networks, along with links to relief funds, resources, and phone numbers for airlines and hospitals. A link to this makeshift news page stayed there for weeks, just below the search bar on Google’s minimalist homepage. Within a year, Google had incorporated a news filter into its search algorithm so that timely headlines appeared atop a list of search results for relevant keywords.
A new era of personalized news products began, in earnest, as a reaction to horrific global news.Today, a Google search for news runs through the same algorithmic filtration system as any other Google search: A person’s individual search history, geographic location, and other demographic information affects what Google shows you. Exactly how your search results differ from any other person’s is a mystery, however. Not even the computer scientists who developed the algorithm could precisely reverse engineer it, given the fact that the same result can be achieved through numerous paths, and that ranking factors — deciding which results show up first — are constantly changing, as are the algorithms themselves.
We now get our news in real time, on-demand, tailored to our interests, across multiple platforms, without knowing just how much is actually personalized. It was technology companies like Google and Facebook, not traditional newsrooms, that made it so. But news organizations are increasingly betting that offering personalized content can help them draw audiences to their sites—and keep them coming back.
Personalization extends beyond how and where news organizations meet their readers. Already, smartphone users can subscribe to push notifications for the specific coverage areas that interest them. On Facebook, users can decide — to some extent — which organizations’ stories they would like to appear in their news feeds. At the same time, devices and platforms that use machine-learning to get to know their users will increasingly play a role in shaping ultra-personalized news products. Meanwhile, voice-activated artificially intelligent devices, such as Google Home and Amazon Echo, are poised to redefine the relationship between news consumers and the news.
While news personalization can help people manage information overload by making individuals’ news diets unique, it also threatens to incite filter bubbles and, in turn, bias. This “creates a bit of an echo chamber,” says Judith Donath, author of The Social Machine: Designs for Living Online and a researcher affiliated with Harvard’s Berkman Klein Center for Internet & Society. “You get news that is designed to be palatable to you. It feeds into people’s appetite of expecting the news to be entertaining … [and] the desire to have news that’s reinforcing your beliefs, as opposed to teaching you about what’s happening in the world and helping you predict the future better.”
As data-tracking becomes more sophisticated, voice recognition software advances, and tech companies leverage personalization for profit, personalization will only become more acute. This is potentially alarming given the growth of websites — news-oriented and otherwise —inhabiting the political extremes, which on Facebook are easy to mistake for valid sources of news. When users can customize their news, and customize to these political and social extremes, civic discourse can suffer. “What’s important is how people use the news to have a discussion,” says Donath. “You may have friends or colleagues, and you read the same things in common. You may decide different things about it. Then you debate with those people. If you’re not even seeing the same news story, it leaves you with a much narrower set of people with whom you share that common ground. You’re losing the common ground of news.”