News reports on the popularity of fake news in the past two months have spiked. Ever since Donald Trump won the election, mainstream media has seized on the narrative that fake news had a large influence on American voters during the last election.
Canadian BuzzFeed News media editor Craig Silverman, who has covered fake news for years, wrote an article entitled "This Analysis Shows How Fake Election News Stories Outperformed Real News On Facebook" a week after the election. It was this piece, and others by Silverman, that the mainstream press made hay and ran wild with, conflating his findings as proof fake news was rampant online, misinforming the public.
"[I]t's definitely just one slice of the picture and it was therefore unsettling for me to see people take this one data set and declare that it means fake news won Trump the election. No, it doesn't. But in the more than two years I've been looking at this kind of content I've never seen it hit this big in such a sustained way," explained Silverman by email.
The New York Times published a frontpage article, "As Fake News Spreads Lies, More Readers Shrug at the Truth", and the rest of the mainstream followed suit, devoting ad nauseum coverage to the supposed rise in fake news. Then the conspiracy theory of Pizzagate culminated in an armed gunman holding up a pizzeria so he could investigate a non-existant paedophile sextrafficking ring being run in the restaurant's non-existent basement. The mainstream media's confirmation bias was falsely confirmed by this one story, and another wave of news stories of moral panic over fake news were published.
However, the mainstream media overestimated and ginned up the power fake news sites have on the overall electorate, and some of them conflated Silverman's methodology and findings as their evidence. A closer look at Silverman's research suggests that the apparent rise and power of fake news has been greatly exaggerated by the mainstream media.
In Silverman's article claiming top fake news outperformed real news on Facebook, he looked at the amount of Facebook engagement fake election news received compared to real election news. For his analysis he took the top 20 real and top 20 fake articles from the three time periods of February to April, May to July, and August until election day (Nov. 8). His analysis concluded that in the last three months of the election, the Facebook engagement of top fake news stories spiked to surpass that of top real news stories.
Reviewing Silverman's raw data, it is mostly accurate, but there are some caveats.
The first exception is Silverman's categorization of some articles. A Breitbart article and Twitchy article were deemed fake news in his top Facebook-engaged fictional stories published in the last three months of the election. Although these two articles were hyper-partisan, I wouldn't categorize them as fake because they don't make up their facts from thin air. And to single out a right-wing publication like Breitbart for one of its more torqued pieces, categorizing it as fake news, meanwhile treating the left-wing hyper-partisan publications like Vox, Huffington, etc. as completely legitimate and real news is unfair. If real news sites, which Breitbart most certainly is one, were all nitpicked for fake news stories then there would most certainly be false stories from all real news outlets. Just take a look at CNN's many whoppers, like when it told viewers it's illegal for the public to read emails from Wikileaks.
The second exception to Silverman's analysis was the decision to only look at 19 real news sites while not putting a limit on the amount of sites looked at for top fake news stories. This severe limiting of the real news pool meant some top performing real stories in Facebook engagement were left out from Silverman's analysis. Even from his list of 19 publications, several top articles in Facebook engagement were left out of Silverman's analysis in the time period of August to election day. For instance, a Business Insider election article and Huffington Post election article had 747,000 and 537,000 Facebook engagements respectively, and both were not included in the real news category, despite these two publications being included in Silverman's list of 19 real news sites. The sharp decline of real news in Silverman's chart is exaggerated from the reality.
Although limiting of real news sites to 19 and the omission of several of the top performing articles within Silverman's list skewed Facebook engagement of real news stories to drastically decline, the top 20 fake news articles (in which a few top stories were also overlooked by Silverman as well) would still have likely surpassed and "outperformed" real news stories for Facebook engagement in the last three months of the election. However, the notion that Facebook engagement numbers of top fake news stories is proof of mass consumption and deception of the American people at large is foolhardy.
The actual popularity of fake news sites pales in comparison to real news sites. When looking at search engine optimization (SEO), the ranking of fake news sites is underwhelming. Ending the Fed, the most popular fake news site by far, with about 40 per cent of all the Facebook engagement from the top 20 fake election articles from August to election day, merely ranks as the 7,519th most popular site in the world. Other fake news sites, other than hyper-partisan publication Yes I'm RIght at 2,566th, are way back with rankings like 10,119th, 41,546th, 203,620th, 589,896th, and 5,555,984th.
Real news publications on the other hand, as one would expect, have way more real engagement with actual traffic to their sites. CNN is 89th, The New York Times is 103rd, Washinton Post is 173rd, and Huffington Post is 182nd. The only two real news publications on Silverman's list outside the top one-thousand are Vox at 1,311th and New York Daily News at 1,032nd.
Silverman was dismayed that some in the media have conflated his analysis as proof fake news outperforms real news for actual consumption.
"No they definitely don't outperform on a daily basis, and in terms of overall traffic the mainstream sites dwarf them. I mean, it would be insane if it was any way other than that! These fake sites are often one-man bands. They are sometimes run as a hobby or side hustle. The biggest ones may have a roster of low paid freelancers. So they cannot compete on a daily or overall traffic basis with big mainstream news sites. It's really about the hits and the engagement they can get on social for them," said Silverman.
"The way I'd put it is this: It seems crazy that I'd have to keep adding more than 19 of the biggest English-language news websites in the world in order for their content to get more total engagement than the fakes. These mainstream sites, with big resources, huge FB pages and dedicated social teams, should not be in the same realm as fake news when it comes to engagement on Facebook," explained Silverman.
So why did the top fake news sites get similar or better Facebook engagement than real news sites in the last three months of the election, despite being way less popular than real news in actuality?
First, Facebook engagements include likes, emoticons, comments, shares, and link clicks, all of which can be questioned for the authenticity of whether or not they actually captured a user's interest. Likes and emoticons mean very little in whether or not a user is actually affected by the content. Similarly, comments are often left by users who haven't read the article, and sometimes comments are racked up by a few users having conversations or arguments. There is also the likelihood that these fake news sites have paid bots to increase engagement on their Facebook posts. Silverman noted that fake news was also spread by users seeding it in Facebook groups, which members would then tactically spread across the internet. Finally, if Facebook engagement is cheap to come by, and fake news sites can entice interaction by having the most creative and visceral headlines out there (truth is no impediment), then it's unsurprising that fake news stories with well-executed promotion and attention-grabbing headlines could begin trending on Facebook's trending news module (more on how that was able to happen below), which would then further skyrocket the fake stories' engagement.
Second, in late August Facebook fired its trending news team after it was reported by Gizmodo that the human curators suppressed legitimate conservative news from the trending module. With the removal of human editors, Facebook relied on an algorithm to choose its trending news stories. From that point up until the election, fake news stories began breaking into Facebook's trending news list. Most of the top 20 fake news articles from August to election day on Silverman's list were published after the trending news team were fired, suggesting that fake news engagement may have proliferated because there were no gatekeepers removing fake news from Facebook's influential trending news module (44 per cent of Americans get their news from Facebook) during this time period.
Silverman disagrees there is a correlation between the rise in fake news's Facebook engagement after the removal of human curators of the trending news module (created in 2014).
"I don't think the two are connected. Those editors worked only on Trending and that product is still fairly new and not a huge driver of traffic. It's really all about the News Feed when it comes to getting traffic and exposure for your content on Facebook, and that is a totally different product. (Though my understanding is that the Trending algorithm takes signals from the News Feed algorithm.)"
Whether or not the removal of the curators was a main reason fake news stories engagement spiked in the last two months is impossible to verify since Facebook is secretive on its internal numbers and overall performance of its site.
Nevertheless, the amount of people tricked by fake news stories is verifiable. In another BuzzFeed piece, Silverman analyzed an Ipsos poll, which asked 3,015 American adults about real and fake news stories. Silverman found that an average of 75 per cent of respondents who were familiar with fake news stories believed in the lies. However, only about 17 per cent of respondents on average had heard of the fake news stories, so only an average of 12.5 per cent Americans polled were actually duped by a fake news story. A significant portion to be sure, but not nearly enough to justify the mainstream media's overblown moral panic over the issue. One can find at least 12.5 per cent of the American population believe much more ludicrous things than fake news.
Now that Facebook is designating ABC News, Politifact, FactCheck, and Snopes as fact-checkers, fake news engagement should wane from the inevitable crackdown on fake news. What Facebook users should now watch for is if these left-wing organizations begin once again suppressing conservative news stories, as well as other digital start-up competitors.
The real fake news dilemma comes not from a sudden rise in Facebook engagement of these low-budget hoax sites, but rather where fake news has always been the most powerful and deadly, from within the mainstream media itself.
The Washington Post alone has published a slew of egregiously false news stories within the last two months. In one story, the paper had to publish this embarrassing retraction: "An earlier version of this story incorrectly said that Russian hackers had penetrated the U.S. electric grid. Authorities say there is no indication of that so far. The computer at Burlington Electric that was hacked was not attached to the grid."
Another Washington Post article entitled "Russian Propaganda Effort Helped Spread 'fake news' during election, experts say" irresponsibly relied heavily on a hyper-partisan PropOrNot's phony blacklist of fake news sites (including Breitbart and other real news sites PropOrNot deemed political enemies), despite the group's amateur website and the members hiding behind anonymity. In a delicious bit of irony, the Post's story on fake news sites turned out to be fake news itself. The paper was forced to include an editor's note in which the paper distanced itself from the dubious PropOrNot and its phony blacklist.
Less amusing fake news stories are ones that have been used to manufacture public consent. Who could forget the false cries of "weapons of mass destruction" used as an excuse to invade Iraq? How about Ben Rhodes, Obama's former deputy national security adviser for strategic communications, bragging about how he duped the press into reporting the false narratives the administration fed them? Or more recently, the extremely blown out of proportion stories about Russia "hacking the election", stories with no real evidence other than intelligence agencies' and anonymous sources' blusterous claims.
If anything should be learned from the mainstream press's hyperventilation over these fake news stories it is that no news outlet is omniscient and the reader should remain ever vigilant and skeptical of all news sources.
Written by Graeme C. Gordon
