Maarten Schenk, cofounder of hoax-debunking site Lead Stories, didn’t set out to be a fact-checker. He wanted to be a trend forecaster. “I wanted, basically, to predict the front page of Reddit,” Schenk, a gregarious blond Belgian, told me over Skype in June. He had created an analytics tool, Trendolizer, as a way to figure out what was popular online and Lead Stories as a way to aggregate the viral news items his tool identified. But the blog didn’t catch on. “Whenever something was trending, all the major other sites were already covering it,” he explained. There was one exception: When Schenk called out lies on the internet, people paid attention.

“Hoax sites and prank sites were getting these massive viral stories, tens of thousands of likes on stories about how Putin secretly met with Trump in Switzerland or that Lady Gaga performed a Muslim tribute at the Super Bowl,” Schenk said. He created a feed to isolate misinformation picking up viral steam, from recurring hoaxes about the pope to a ring of websites from Ghana falsely posting about celebrity deaths. Lead Stories changed from a general-interest news aggregator into a shoestring debunking operation.  

Not only did Schenk become a fact-checker; quite by accident, Schenk became a fact-checker of a fact-checker. This summer, he exposed a plagiarist who was stealing his work, as Buzzfeed reported. After Schenk realized that his original reporting was being copied by a man posing as a fact-checker, he set a trap—first identifying the IP address that appeared to be stealing his content, and then creating an alternate homepage only accessible to that IP address. On the decoy homepage, Schenk published a deliberately false George Lucas death hoax, and then waited until his plagiarist copied it. Then he reported the imposter.

By unmasking a fellow debunker as a fraud, Schenk highlighted fact-checking’s stickiest issue: figuring out whom to trust. Not all fact-checks are created equal. Despite exponential growth in the fact-checking industry, there is no match for the oceanic swell of bad information getting pushed out online. What’s more, major online platforms appear fundamentally ill-equipped to handle the task of sorting good information from bad. Last week, for example, Mark Zuckerberg’s assertion that Facebook should not ban Holocaust-denial content from the platform highlighted how hesitant and messy the company’s approach to determining truth has been. Facebook’s attempts to clarify its approach to misinformation following this interview have only further muddied its relationship with fact-checkers.  

There is now a wide range of independent fact-checking operations, both in the United States and internationally. Most of these outlets are digital, although print newsrooms and magazines have long employed specialists to safeguard their accuracy. In 1913, to cite an early example, Ralph Pulitzer established the Bureau of Accuracy and Fair Play for the New York World. Today, the presence of an in-house fact-checking department is a marker of quality and prestige. (The New Yorker’s in-house fact-checking department, launched in 1927, is legendary.) These internal departments, which proofread, clarify, and double-check writers’ work, are often part of a magazine, website, or newspaper’s copy department, and in recent years, budget cuts have made these jobs scarce. “The decline of magazine checking departments began in the mid-to-late 1990s,” media analyst Craig Silverman explained for Poynter in 2012. But as in-house fact-checking came to be seen as a media luxury, independent and dedicated fact-checking operations—with their eyes fixed on the work of others—started to emerge in the early 2000s.

Unlike internal fact-checking departments, which attempt to perfect the accuracy of writing before it is published, these externally focused operations scrutinize statements that have already been made by public figures, attempting to offer a corrective to the record. Some are armed with full newsrooms and ample cash, like PolitiFact. Some, like Schenk’s Lead Stories, resemble hobbyist operations. Others have professionalized along the way. Snopes, for instance, began as a one-man urban-legend reality check on Usenet, and founder David Mikkelson launched its dedicated website in 1994. Gradually, it evolved into an incredibly popular general-interest fact-checking operation, which also assesses political statements and media.

As more fact-checking groups have begun policing media outlets, viral stories, and public officials, the world of fact-checking has ballooned to include more than just stump speeches and campaign ads; the Mexican fact-checking project Verificado 2018, for example, has attempted to debunk false information spread through WhatsApp, the messaging service owned by Facebook. Major platforms like Google and Facebook have pushed the stream of information into hyper-speed. Their information feeds have created the conditions for instantaneous amplification of the news—which is a term that now also often applies to unvetted, deliberately misleading, and sensationalist content. These companies are expert at surfacing stories that grab attention. They are far less authoritative when it comes to surfacing stories that have been verified by experts as accurate. To try to stem the tide of misinformation they have allowed, these platforms have recently begun asking fact-checking organizations for help.

While fact-checking organizations originally sprang up as attempted antidotes to political misinformation and hoaxes, their role has ballooned into ad hoc and woefully incomplete corrections departments for the digital world. Some major fact-checking organizations have entered into asymmetrical relationships with big platforms, which means their efforts at debunking misinformation rely on the same social networks responsible for spreading misinformation. The end result is maddening for anyone trying to figure out where to find trustworthy information. The rise of fact-checking has not resulted in a more orderly or easy-to-understand internet. Right now, fact-checkers fighting lies online resemble volunteer firefighters equipped with pails of water to fight a five-alarm blaze.

The first organizations devoted to fact-checking grew up with the internet. Fact-checking within media organizations existed long before computers, but it wasn’t until the 1990s that third-party organizations devoted to debunking established themselves. As Lucas Graves laid out in his 2016 book Deciding What’s True, the first dedicated fact-checking site helmed by journalists, FactCheck.org, was established in 2003 as a project within the Annenberg Public Policy Center of the University of Pennsylvania. “The site was conceived as a project for the 2004 election but proved unexpectedly popular and became a year-round venture,” Lucas wrote. FactCheck.org continues to comb politicians’ statements, advertisements, and campaign materials in 2018, though it has also expanded to debunking viral stories on Facebook and dubious health and science claims, like former Environmental Protection Agency administer Scott Pruitt’s suggestion that global warming may not be “a bad thing.”

PolitiFact, another early fact-checking site, began in 2007 as a project seeking to debunk untrue statements from public officials in the St. Petersburg Times (now the Tampa Bay Times) but is now part of the Poynter Institute, which owns the Tampa Bay Times. The Washington Post’s Fact Checker also launched in 2007, identifying untrue statements from politicians; although it operated within the Post, it was separate from the newspaper’s internal fact-checking department, instead focusing on assessing the truth of statements made in the political realm. Many of these first fact-check organizations were focused on politics and saw themselves as more aggressive than the mainstream media at debunking political lies. PolitiFact founder Bill Adair saw the emergence of fact-checking as a corrective to milquetoast reporting. “I think political journalists became overly cautious, and fact-checks became overly watered down, and we got a lot of on the one hand, on the other hand journalism,” he said.

The 2016 election spurred intense interest in fact-checking what gets said online, from viral tweets to mainstream media reporting, as well as renewed attention to investigating politicians’ claims. This interest awkwardly coincided with waves of introspection and doubt about the premise of fact-checking itself; “fake news” turned from a label for the torrent of fake stories flooding social media into a useless buzzword President Trump deployed to dismiss critics. Trump’s victory, in spite of his lies and in spite of dogged efforts from the media to debunk them, prompted a familiar complaint: Is fact-checking effective? Previous elections had already produced grievances with its efficacy. Late New York Times media critic David Carr dismissed fact-checking initiatives in 2012. (“Fact checking, as it turns out, is more of a cottage industry than a civic corrective,” he wrote, arguing that organizations like PolitiFact and Fact Checker had not succeeded in meaningful persuasion of readers.) After the 2016 election, the debate about whether fact-checking could sway people picked up again in earnest. Media analysts and even a former fact-checker argued that pointing out Trump’s profligate lies had, in fact, made people like him more. Fact-checking defenders, like writer Daniel Engber, have argued that the failure of facts has been greatly overstated. In 2016, researchers from Ohio State and George Washington universities dismissed an earlier study claiming that fact-checking could backfire. “By and large, citizens heed factual information, even when such information challenges their ideological commitments,” they wrote.

Amid the tussling over what fact-checking accomplishes, the organizations dedicated to checking facts and determining media credibility have continued to grow. According to the Duke Reporters’ Lab, the number of fact-checking groups around the world has tripled since 2014, with new entries like The Nevada Independent and Univision’s Detector de Mentiras, the first Spanish-language fact-checking vertical in the United States.

Despite their rising numbers, these outlets are frequently tiny operations, like Schenk’s. “It’s pretty much a one-man show right now,” Schenk said. Even Snopes, one of the best-known and longest-running fact-checking operations, has a surprisingly lean staff. “We’re only 17 people total,” Snopes co-owner Vinny Green told me. At the Global Fact-Checking Summit, an international conference held in Rome this June, International Fact-Checking Network director Alexios Mantzarlis laid out the shoestring financial nature of many of the organizations. “Mantzarlis noted that of the 42 IFCN verified fact-checkers attending the conference, 26 operated with a budget of $100,000 or less in 2017,” The Washington Post’s Glenn Kessler reported.

The patchwork of fact-checking organizations remains in flux, but there has been a constant through line to all efforts in recent years. In addition to investigating claims that originate from politicians and other public figures, each organization has had to adapt to the rapidly increasing pace and volume of a news ecosystem super-juiced by the internet and its dominant communications companies. “We’re in a position where communication is voluminous beyond understanding,” Green said. “We have to adapt.”

The contemporary fact-checking movement’s growth has run parallel to the rise of platforms like Facebook and Google (as well as YouTube, which is owned by Google) as news publishers. Nothing about these companies is tiny or shoestring, in contrast to most fact-checking organizations. These tech giants have enabled instantaneous information sharing and equally instantaneous misinformation sharing. This has resulted in a seismic change in how the media works and how people communicate.

The relatively explosive growth of the fact-checking world is closely linked to the growth of these platforms, as they guarantee that the organizations will never run out of bad information to debunk. “There have been a few big changes since I wrote Deciding What’s True,” Lucas Graves told me via email. “The biggest changes [for fact-checking organizations] are the partnerships with Google and Facebook—arguably the two most important information companies in the world—and the growing emphasis on finding viral misinformation and bogus news reports in addition to checking political claims.”

Google continues to struggle with promoting bad information, particularly on YouTube, where the recommendation algorithm has a bad habit of pushing conspiracy videos; recently, a conspiracy video was uploaded following the school shooting at Parkland and featured in YouTube’s “Trending” section, showing that the problem persists. (YouTube pulled the video, but not until it had accumulated over 200,000 views.) These issues persist in the company’s flagship search product as well. Last spring, Google search erroneously listed “Nazism” as the ideology of the California Republican Party.

There have been outside attempts to curb Google’s role in spreading misinformation; Encyclopedia Britannica recently created a browser extension called “Britannica Insights” which will fact-check incorrect Google search-result snippets. But Google is working on a number of projects with fact-checkers to try to fix its problems internally. Leading up to the French presidential election in 2017, the Google News Lab supported a virtual newsroom called CrossCheck, where fact-checkers shared information they had deemed correct and debunked false reports. In 2017, Google expanded an experimental program called “Fact Check” for both Google News and Google Search, which pulls information from independent fact-checkers to the top of appropriate search results. Google encourages fact-checking organizations to use specialized tags from collaborative data group Schema.org to signal that they have published fact-checked content. (If you search for “Elvis alive,” for instance, one of the top results will be a Snopes fact-check.) Additionally, Google’s Jigsaw technology incubator partnered with the Duke Reporters’ Lab to create Share the Facts, a widget that scans content and provides a fact-checked summary, which is now used by PolitiFact, The Washington Post, and FactCheck.org.

Justin Kosslyn, a product manager at Jigsaw, emphasized that the rise of third-party fact-checking has made it possible for Google to approach fact-checking more aggressively. “I was a product manager in Google News six or seven years ago, and I tried to launch a fact-check feature within Google News at that time,” he said. “But we couldn’t figure out how to structure the data, how to understand what a fact-checked article is, what isn’t a fact-checked article, what do the algorithms do, when is human policy enforcement needed. So I gave up.” Kosslyn began to see potential for future fact-check collaboration again in 2015. “The space had advanced enough that we realized it was something we could really do.”

Not all fact-checking efforts have been fruitful, to put it mildly. In January 2018, Google had to suspend a fact-checking feature called Reviewed Claims over quality-control issues after right-wing media outlets protested; the feature had suggested that a Daily Caller article had made a false statement when it had not published the falsehood in question. And the obvious potential downside to the Schema.org feature is that anyone might use the metadata tags and mold themselves into a fact-checker. I asked how Google protects against bad actors promoting fake news using the fact-checking tags. Jigsaw head of communications Dan Keyserling told me that users can complain about incorrect fact-checks. “If enough users are giving feedback on something, that will trigger a review,” he said. Considering the sheer volume of conspiracy theorists, hoaxers, and sensationalists flooding Google search and YouTube with incorrect information, the open nature of this approach to fact-checking may result in trouble in the future. “I think the platform partnerships are a work in progress,” Lucas Graves said. “They’ve come a long way in a relatively short time. The claim review standard which Google helped to develop could really transform the field in the next several years.”

Meanwhile, Facebook—where many an aunt and uncle have shared many a horrifying and blatantly incorrect political meme—has entered into a fact-checking partnership with 24 fact-checking outlets in 14 countries, including Snopes. The crackdown on misinformation has resulted in a few minor victories, including a purge of create-your-own-headline websites. “There’s these prank websites where people can just enter their own headline and fool their friends,” Schenk said. “Those used to be relatively big, until about a half a year or a year ago. But since Facebook has been really cracking down on them, nothing from those sites seems to go viral anymore.” There have been missteps, as well: Facebook had to overhaul the way it marked stories disputed by fact-checkers, as it discovered that its first method (literally flagging the story by attaching a red “disputed” symbol) may have increased sharing of the bad information instead of reducing it. “Just because something is marked as ‘false’ or ‘disputed’ doesn’t necessarily mean we will be able to change someone’s opinion about its accuracy,” Facebook product designer Jeff Smith explained, citing research on how some attempts to fact-check can backfire.

This year, Facebook has gone on a full-court PR press to emphasize its belated blitz on bad information, and it recently released a short film called Facing Facts to promote the campaign. While it has not shared exact numbers about how its attempts to crack down on bad information have worked, the company asserts that the partnership has yielded results. “Once a story is rated false by a third-party fact-checker, we’ve been able to demote it in News Feed and reduce its future views by 80 percent,” a Facebook spokesperson told The Ringer.

But Facebook’s relationship to determining truth is still extremely fraught, and its fact-checking partnerships and efforts also appear precarious. Much of this struggle is of the company’s own making. This past April, the Tow Center for Digital Journalism released a report on Facebook’s union with fact-checkers that highlighted some of the partnership’s issues. “Almost all of the [fact-checking] partners said the system needed to process memes, images, and video content far more effectively,” it reads. Last week, Facebook’s John Hegeman, its head of News Feed, inadvertently showcased the central failure of the company’s efforts to squelch lies. When a CNN reporter asked why Facebook has not banned Infowars, a media company notorious for spreading false information, Hegeman emphasized that Facebook does not “take down” false news. “I think part of the fundamental thing here is that we created Facebook to be a place where different people can have a voice. And different publishers have very different points of view,” he said. By casting the content that Infowars produces as a different viewpoint instead of harmful, deliberately false information, Hegeman highlighted how uneasy the company is with its role as an arbiter of truth.

This week, Facebook attempted to clarify its policies on how it treats media outlets and other users who spread conspiracy theories, after Mark Zuckerberg told Recode that Facebook should not be in the business of evaluating facts. “Fact-checking,” he said, “I don’t think that we should be in the business of having people at Facebook who are deciding what is true and what isn’t.” Following blowback from Zuckerberg’s interview, Facebook announced that it would remove misinformation from its platform—if the information caused real-world harm. “Reducing the distribution of misinformation—rather than removing it outright—strikes the right balance between free expression and a safe and authentic community,” a Facebook spokesperson told The Ringer via email. “There are certain forms of misinformation that have contributed to physical harm, and we are making a policy change which will enable us to take that type of content down.” The company will be partnering with third-party organizations, like threat-intelligence agencies, to determine when bad information reaches the threshold of “harmful.” Facebook has already applied this new policy in Sri Lanka; last month, it removed content alleging that Muslims are poisoning food given to Buddhists after local partners determined the false information had the potential to cause real-life harm.

“We’re evaluating the idea of leaning on third-party fact-checkers to surface content for us, but for now we’re starting with orgs like CSOs whose expertise lies in evaluating risk of physical harm,” the Facebook spokesperson told The Ringer. “We’ll still use third-party fact-checkers in countries where we have them (current list here) to review stories and check their facts, but we’re still figuring out how they’ll factor into this new policy process.”

“The partnership with Facebook—it’s hard,” Snopes’ Vinny Green said. “They’re a decade behind on coming up with the necessary solutions for this problem that academics already knew existed. I think they’re doing the best they can with the constraints that they have, and they’re realizing it’s an existential crisis. I think they can and should be doing more.”

Although Snopes is happy to be part of the conversation, the organization is aware that its own brand of human-monitored fact-checking cannot be easily replicated at Facebook’s colossal scale. “We knew from the jump that all we were doing was help Facebook train an algorithm,” Green said. “They didn’t say that explicitly, but anyone who is savvy knows that these companies that operate on the 2 billion monthly active-user scale are not going to hire a million fact-checkers to write a million fact-checks. That’s just not how that works.”

Green emphasized that the next step to rooting out bad information will be trickier, as there are many outlets publishing sensationalized and biased articles that require a defter hand to fact-check. “The more fundamental and nuanced issues are not things that are going to be solved with algorithms,” Green said. “At some point in time, when you get all the hoaxes and obvious fake news out of the mix, you have to make a decision about inflammatory publishers that have positioned themselves as un-fact-checkable but that are still able to reap the benefits of the platforms.” (Facebook did not respond to a request for comment on Green’s statement.)

Since even Facebook’s (and Google’s and other tech giants’) most rudimentary attempts at content moderation have been frequently fraught, the chances that these platforms will emerge as a trustworthy authorities on healthy information ecosystems seem negligible. Even if these platforms find better and more thorough ways to integrate fact-checking experts into their assessments of information, there’s a compelling argument to be made that the root of the problem is the sheer volume of information: the size that these tech behemoths have grown to and the entrenched roles they’ve assumed in distributing information. Sally Hubbard, a former assistant attorney general in the New York State’s Antitrust Bureau, contends that the “fake news” problems at Facebook and Google are actually antitrust problems.

“The tech platforms’ business incentives run counter to the interests of legitimate news publishers, and the platforms pull technological levers that harm publishers’ business models and advantage their own” she wrote. These effective monopolies have worsened the problem, as misinformation purveyors only have to figure out how to game a handful of algorithms—and that these algorithms are not meant to surface quality news. “Weak competition in social media platforms means Facebook can tailor its News Feed to serve its financial interests, prioritizing engagement on the platform over veracity. Lack of competition in online search means Google does not face competitive pressure to drastically change its algorithm to stem the spread of fake news.” Hubbard’s perspective, then, is that no number of fact-checking partnerships can fix the misinformation problems baked into the tech giants’ expand-at-all-costs business models. In this view, the most realistic treatment for fake news is the breakup of its most powerful distributors.

Even if antitrust crusaders radically transformed the tech industry, the world of fact-checking would still be messy and tense. Fact-checkers are part of the media, after all, and most Americans assume that the majority of the news they consume is biased and that a significant chunk is flat-out wrong. To feed an audience of media skeptics, fact-checking projects devoted to assessing the credibility of media organizations have prospered alongside traditional fact-checking outlets. Some of these projects, like Schenk’s Lead Stories, focus on debunking hoaxes perpetuated by opportunistic media grifters. BuzzFeed News established its own misinformation beat, tracking hoaxes in the wake of tragedies like the Toronto van attack and broadly analyzing what it calls “digital deception,” from cryptocurrency cons to conspiracy-theory videos. Other new organizations attempt to assess what more mainstream organizations are getting wrong or whether they should be trusted. Popula, a new general-interest website, recently ran a column devoted to fact-checking a New York Times op-ed. NewsGuard, a new offering from journalists Steven Brill and Gordon Crovitz, plans to release “nutrition labels” for various media organizations, building a newsroom specifically to evaluate the work of other newsrooms.

One new journalism startup, Civil, which is currently launching a variety of digital newsrooms (including Popula), has created in-house credibility indicators in order to encourage audiences to trust its unknown and untested journalism. “We talked to journalists and people who already paid for news somehow. What we ended up finding was interesting and a little depressing: Even people who subscribed and paid money sometimes to multiple news outlets weren’t really as media literate as one might think. They really didn’t necessarily know what a reporter did, what original reporting was,” Civil product designer Julia Himmel said. Civil proposes to solve the problem with indicators, which journalists would insert into their stories, to flag when a story features original reporting or uses quotes from a reputable expert. It’s not the only organization with this idea: The Credibility Coalition, a working group of journalists, entrepreneurs, academics, and activists, recently published a paper suggesting that outlets use potential universal credibility indicators—such as the use of quotations from outside experts and whether logical fallacies like straw-man arguments appear—as a way to establish trust with readers. These stabs at improving media literacy from within newsrooms might be heartening if they weren’t so grimly indicative of the crisis of trust between readers and the media.

Fact-checking organizations face a parallel set of challenges to the publishers and pundits they scrutinize. They are media organizations, too, and thus subject to claims of bias. Conservative media, in particular, can be aggressive toward fact-check groups. The National Review’s Jonah Goldberg once memorably described PolitiFact as “the hackiest and most biased of the fact-checking outfits, which bends over like a Bangkok hooker to defend Democrats.” During the 2016 campaign, Breitbart ran its own fact-check on other fact-checking outlets during the first presidential debate in order to correct what it saw as pervasive bias in the fact-checking world. “In practice, ‘fact-checking’ is weighted against Republicans,” Joel B. Pollak wrote. “Largely because fact-checkers evaluate Republicans more than Democrats.”

While groups like the International Fact-Checking Network are attempting to professionalize and establish universal standards for members, whether a person finds a fact-checking outlet credible can still boil down to matter of individual taste or perception of bias. When I asked Daily Caller cofounder Neil Patel how the media company’s fact-checking vertical, Check Your Fact, contends with the company’s right-wing reputation, he insisted that it does not shy away from exposing the falsehoods of Republican leadership. “We have a work product, and people can review it and critique it,” Patel said. “Did small business optimism just hit a record high? Trump said yes; we pointed out that that’s not true. Is crime in Germany up 10 percent because of an influx of immigrants? Trump said yes; we say it’s actually lower.” Patel rattled off a string of other Trump lies recently identified by the project.

One fact-checker is proactively trying to avoid readers bristling about bias by presenting his corrections in the most appealing way possible. Internet Archive senior creative technologist Dan Schultz is working on a browser extension called “Truth Goggles” in partnership with the Tech & Check Cooperative from the Duke Reporters’ Lab. (It will be an overhauled version of a prototype he created in 2011.) This new version will ask users what they think about an issue before presenting them with facts, customizing the way it presents the fact-check to try to make them more receptive to the information. “When you’re feeling like you’re being attacked or somebody’s trying to change your worldview because they’re right and you’re wrong, you shut down,” Schultz said. It’s a far-fetched-sounding project in the contentious world of fact-checking, but one that also digs into how essential trust is to fact-checking.

Truth Goggles is the most obviously idealistic of the newer fact-checking projects cropping up, with its aim to bridge ideological divides. But most of the fact-checking organizations are, at heart, connected by the idea that it is possible to improve both the quality of information coming from the media and the reactions to that information. This does not, of course, mean that they should be viewed without skepticism. All of these organizations are run by people, and thus girded by the biases and objectives of the humans who run them. Whether an audience trusts fact-checks coming from The Washington Post versus fact-checks coming from The Daily Caller will inevitably be colored by their overall trust in their parent organizations. These organizations all aim to show people the truth, but their individual ideological tilts can vary wildly and require a discerning eye. Earlier this summer, Elon Musk praised the fact-checking efforts of an organization called The Knife, a subscription-based media watchdog organization; The Knife was created by members of the alleged sex cult NXIVM.

It’s also imperative to understand how limited and incomplete the fact-checking efforts from platforms like Google and Facebook are thus far and how far away they are from providing a functioning informational ecosystem. “How we get news and information is broken; how we teach children and adults to be web literate and use the internet as a tool—we have to acknowledge that and make the swiftest change in direction,” Green said. “If we don’t, we’re going to see a massive erosion in trust in institutions. And if we don’t trust institutions, how can we trust our own democracy? And if we can’t trust our democracy, we don’t have one.”

Kate Knibbs