Facebook’s Fake News Fix

Facebook’s Fake News Fix/Sue Halpern is a regular contributor to The New York Review and a Scholar-in-Residence at Middlebury. Her latest book, the novel Summer Hours at the Robbers Library, will be published in February.
 (January 2018)

The New York Review of Books, 23/Ene/2018;

Facebook thumbs-down hand on a iPhone. Facebook is a social media company owned by Mark Zuckerberg. (Photo by Ted Soqui/Corbis via Getty Images)

When Facebook founder and CEO Mark Zuckerberg shared with his (then) 101,545,240 followers that his New Year’s resolution for 2018 was “to fix” Facebook, one might have asked, “Fix it for whom?” It is a question with a number of possible answers: for its shareholders, who saw growth level off in the last year and young people turning to other social media platforms; for its users, who saw their personal information appropriated by political operatives and sometimes used against them in insidious ways; or for the public more generally, which is living with the consequences of that appropriation and with the proliferation of propaganda, camouflaged as legitimate news, not only in the United States, but in countries such as Burma and South Sudan, where Facebook-generated “fake news” has been used to instigate ethnic violence.

Let’s be cynical for a moment. Coming just weeks after a contentious hearing of the Senate Select Committee on Intelligence—where Senator Dianne Feinstein held the cudgel of regulation over the heads of representatives from Facebook, Google, and Twitter, telling them that if their companies did not “do something” about the manipulation of their platforms by foreign agents, “we will”—Zuckerberg’s New Year’s resolution could also be interpreted as his personal response to Feinstein’s threat. After all, his company spent more than $8.4 million last year deploying a brigade of thirty-six lobbyists to walk the halls of Congress defending Facebook from government interference. A New Year’s resolution is a lot cheaper, and it gives its author 364 days to make good.

And it took only a week for Facebook to offer up its fix. It did not come from Zuckerberg directly, but from Adam Mosseri, the company’s Head of News Feed.

The Facebook news feed is not, as its name suggests, a compendium of topical stories from various news media. Rather, it is a collection of posts from one’s Facebook “friends,” from sites one has actively “liked,” from third-party sites that Facebook believes you’re interested in based on your previous online behavior, and from advertisers who have put you into a category that, theoretically, makes you predisposed toward buying what they are selling. None of this is random. Everything that appears in the news feed is there because Facebook’s proprietary algorithm has put it there. That algorithm sorts through and analyzes masses of digital signals, deciding which of your friends’ posts you’d most like to see and which silly cat videos and other material from greater Facebook will grab your attention. It then serves these up along with a slew of paid advertisements, delivering to two billion people a never-ending supply of individualized, targeted content.­­

Mosseri began his message reiterating the company’s aspirational mantra. “Facebook was built to bring people closer together and build relationships,” he wrote. “One of the ways we do this is by connecting people to meaningful posts from their friends and family in News Feed.” And then he described—in vague terms—how the company would be tweaking its algorithm so that “people have more opportunities to interact with the people they care about.” To accomplish this, he said, the algorithm will “predict which posts you might want to interact with your friends about and show these posts higher in [the] feed.” So what will fuel this prediction engine? According to Mosseri,“posts that inspire back-and-forth discussion in the comments and posts that you might want to share and react to.” He also said that the news feed would carry less public content, “including videos and other posts from publishers.”

In the days following, much was made of the impact this algorithm adjustment will have on third-party sites, especially news organizations, which have come to rely on Facebook for traffic (and, therefore, ad revenue). This is a legitimate worry, especially for legacy journalism, which struggled to adapt to the mobile digital world and found a workaround through Facebook. Then, a week later, Mark Zuckerberg announced a further fix, this one an attempt to limit the amount of “fake news”: Facebook will survey its users, asking them to rank the “trustworthiness” of various news sources. Those with higher rankings will get priority in the news feed, while those with lower scores, presumably, will be shut out. As a consequence, Zuckerberg said, the amount of news on Facebook will decline by 20 percent.

There is no way to know, yet, if outsourcing discernment—if that’s what polling a random collection of two billion people is—will cut down on the amount of propaganda, lying, and deception on Facebook, or if such a survey will simply replicate existing ideological divisions. But it is also unclear where the more than 50 percent of Facebook users who get their news from the site will get it now, if anywhere, since there will be so much less of it. And maybe that is the point. This diminution of news might be a way for Facebook to walk away from the public sphere—or, at least, appear to walk away—at a time when it has been taken to task for its overweening influence there. As Slate’s chairman and editor-in-chief, Jacob Weisberg, told The New York Times, downgrading news content also benefits the company in those countries where independent journalism is outlawed.

“Facebook is just desperate to get into China, and it will never do that unless it censors news—and this is actually a neat solution to that,” Weisberg said. “If you only have news on the platform shared by users, users who live under repressive regimes don’t have access to real news and can’t share it, because it’s legally prohibited.” But even in the United States, where publishers were already tailoring content so it might rise to the top of the news feed, Facebook’s pivot to the personal may encourage more stories about the Instant Pot, for example, and fewer analyses of Middle East policy, since one is more likely than the other to be shared, promoted, and drive traffic. Call it censorship by a thousand clicks.

A few months ago, Facebook experimented with a similar algorithm modification in six countries, removing all professionally-produced journalism from the news feed and sequestering it in its own marginal tab called “Explore.” That tweak, like the one announced last week, prioritized posts from friends and family over other kinds of information. The result was disquieting. It made it more difficult for users to find trusted news sources on the platform, kept official announcements at bay, and, rather than eliminating fake news, in some instances exacerbated it.

This should not have been difficult to predict. It stands to reason that giving priority to posts from friends that generate lots of back-and-forth will do nothing to curtail the proliferation of fake news. If I post something provocative and unsubstantiated that I heard from someone else, and many of my Facebook friends comment on it and pass it along to their friends and those users also comment on it, the new algorithm should be primed to make that information go viral. And if what I say to you, and you say to me, and our friends say to each other, makes it to the top of our news feeds by dint of this new “engagement” algorithm, then Facebook’s echo-chamber effect, where like-minded people talk only among themselves, will be more resonant, not less. Connections may be solidified and alliances cemented, but so will divisions and polarities. (On the other hand, if the new algorithm also prioritizes “happy” content—graduation announcements, birthday wishes, job promotions—over other kinds of exchanges, then the news feed will become the Internet equivalent of the Hallmark card aisle in the supermarket, and Facebook will go the way of MySpace and AOL.)

In the meantime, all this personal content, including our journalism preferences, may have another effect, that of delivering more of ourselves—our interests, desires, movements, passions, and emotions—to Facebook and the data brokers that, in turn, sell it to governments, credit agencies, political operatives, and marketers. Is there any doubt that connections are the currency of Facebook—or that they are in large part responsible for the company’s $500 billion valuation? And about those marketers: missing from Facebook’s “fix” is any mention of advertising. This is a fundamental omission, given that Facebook, despite its founder’s insistence that his company exists to connect the people of the world, does not just happen to run a side-hustle selling things to fund his humanitarian project, but, rather, is the second largest advertising platform on the planet. All of the connecting that is done there is done in the service of all its advertising, since Facebook users are both the customer (who buys things) and the product (who provides the raw data that enables advertisers to target users with such precision). That is the real genius of Facebook, and why the company is so valuable.

As we were reminded during the last election, Facebook is used to sell ideas as well as products. As Brad Parscale, the head of digital media for Donald Trump’s election campaign, pointed out, Facebook’s marketing tools are just as useful for selling a candidate as they are for selling a blender. One of those tools was Facebook’s so-called “dark posts,” which enabled Parscale and his team to send certain African-American voters incendiary messages about Hillary Clinton that only they could see, in an effort to drive them away from the polls. Other tools enabled Cambridge Analytica to parse data on millions of Americans, much of it unlawfully obtained from unsuspecting Facebook users, to create ads designed to trigger voters’ fears and anxieties. Facebook employees were even embedded with the Trump team, helping them deploy these tools. Reporters from the journalism nonprofit ProPublica, though, needed no such help when they bought $30 worth of Facebook’s promoted posts and sent them to users who self-identified as “Jew hater,” or who had expressed interest in “How to burn jews,” and “why jews ruin the world,” all of which were algorithmically-derived categories available for purchase on the site. In an earlier experiment, ProPublica was also able to use Facebook to place a housing ad that explicitly prevented it from being shown to African Americans, Hispanics, and Asian Americans. It took just minutes for Facebook’s artificial “intelligence” to approve these ad buys.

In his New Year’s Facebook post, Mark Zuckerberg wrote that “Facebook has a lot of work to do—whether it’s protecting our community from abuse and hate, defending against interference by nation states, or making sure that time spent on Facebook is time well spent.” By offering remedies that can accomplish none of these, his resolution, like most New Year’s resolutions, was broken almost as soon as it was made.