There's A Reason That Misleading Claims Of Bias In Search And Social Media Enjoy Such Traction
from the but-it's-not-a-good-reason dept
President Trump’s tweets charging that Google search results are biased, against him and against conservatives, are the loudest and latest version of a growing attack on search engines and social media platforms. It is potent, and it’s almost certainly wrong. But it comes at an unfortunate time, just as a more thoughtful and substantive challenge to the impact of Silicon Valley tech companies has finally begun to emerge. If someone were truly concerned about free speech, news, and how platforms subtly reshape public participation, they would be engaging these deeper questions. But these simplistic and ill-informed claims of deliberate political bias are the wrong questions, and they risk undermining and crowding out the right ones. Trump’s charges against Google, Twitter, and Facebook reveal a basic misunderstanding of how search and social media work, and they continue to confuse “fake news” with bad news, all in the service of scoring political points. However, even if these companies are not responsible for silencing conservative speech, they may be partly responsible for allowing this charge to gain purchase, by being so secretive for so long about how their algorithms and moderation policies work.
So what do search engines actually do when users access them for information or news? Search engines deliver relevant results, nothing more. That judgment of relevance is based on hundreds of factors: including popularity, topic relevance, and timeliness. Results are fluid and personalized. There’s plenty of room in this complex process for overemphasis and oversight, and these are important questions to examine. But serious researchers who actually already study this are careful to take into account the effects of personalization, changes over time, and the powerful feedback effects of users. This is a far cry from looking at your own search results and being troubled by what you see. (Even the author of the report Trump was likely reacting to acknowledges that it was unscientific and disagrees with the suggestion that regulation of search should follow.)
To understand, for instance, the results for “Trump” in Google News, or “Trump news” in Google — different things, by the way — we would need to consider some much more likely explanations than deliberate political manipulation: major outlets like CNN may publish a lot more content a lot more often; more users may click on, read, and forward links from these sources; outspoken right-wing sites like Gateway Pundit may have much less trust outside of their devoted base than they imagine; CNN may be much more congruent with centrist political leanings than Trump and conservative critics admit; well-established news sources may already circulate more widely and successfully on social media platforms like Facebook and Twitter, boosting their rankings on search engines; users may simply be more convinced by these news sources, “voting” for them with their clicks and links in ways that Google picks up on.
In truth, there are important questions to be asked about search engines, social media platforms, and the circulation of news online. There are profound concerns about the economic sustainability of journalism itself when it has to compete on social media platforms. There a profound concerns about the subtle effects of how algorithms work. But the noise that right-wing critics are stirring up is not subtle, it is not helpful, it is not well informed — and more than that, it is clearly about scoring political points. Those claiming political bias seem wholly uninterested in acknowledging the inquiries already underway.
Charges of left-leaning bias are not new, of course. They come from a very old playbook conservatives have used against newspapers and broadcasters for decades. Unfortunately, Silicon Valley is partly to blame for why it is working so well today. Search engines and social media platforms have been too secretive about how their algorithms work, and too secretive about how content moderation works. In the absence of substantive explanations, users have been left to wonder why search results look the way they do, or why some posts get removed and others don’t. This uncertainty breeds suspicion, and that suspicion goes looking for other explanations. This leaves room for trolls, conspiracy mongers, and demagogues to suggest that the platforms are silencing them for their political speech — conveniently overlooking the fact that they been suspended for making hateful threats, or can’t reach the first page of search results because readers trust other sources. And Silicon Valley has bruised their users’ trust for so long, that even their genuine explanations sound suspect.
Some of the press coverage, when it’s not careful, can inadvertently make the very same easy assumptions that these critics do. Search results, trending lists, and content moderation are not the same thing, they are not managed by the same people, and they are not handled in the same way. Too often, a critic will thread together ill-informed charges against search, one outdated incident regarding trending, and continued uncertainty about moderation practices, and lace them together into a blanket charge of bias. But they are simply different things.
It is unnerving to feel like an apologist for these tech companies. There are real and concerning questions about how search and social media work. I ask some of these questions in my own research, and my field has been thinking about them for years. The ways these companies have addressed, or often failed to address, the public ramifications of search algorithms and moderation policies has been deeply problematic. But these questions of bias distract us from the deeper problems.
It is also disconcerting, just as the public is finally grasping the subtle ways in which search and social media platforms matter, that we are ready to fall back on so simplistic a charge as deliberate political bias. I feel a bit like critics of mainstream news media, who for years have tried to highlight the way contemporary US news organizations are subtly centrist, structurally cautious, founded by commercial imperatives, and under attentive to marginalize voices — who now have to bracket those critiques and come to the defense of CNN when the President dismisses them as “fake news.” Those of us who ask hard questions about search and social media should do so, but we must also steadfastly refused to lump these real concerns in with facile, politically motivated charges of bias that miss the deeper point.
Tarleton Gillespie is the author of Custodians of the Internet: Platforms, Content Moderation, and the Hidden Decisions That Shape Social Media. He is a principal researcher at Microsoft Research and an affiliated associate professor at Cornell University.