The Washington PostDemocracy Dies in Darkness

We’re giving Russia’s trolling team too much credit

There’s not really much question that Russia’s efforts to interfere in the 2016 election had a significant effect. Hackers believed to be working for the country’s intelligence agencies gained access to the computer network of the Democratic National Committee and the email account of Hillary Clinton’s campaign chairman, John Podesta. Material stolen from those sources was released through WikiLeaks first in July and then about a month before the election. Day after day during October 2016, new files stolen from Podesta were released, frequently giving new ammunition to Clinton’s opponents, including then-candidate Donald Trump.

"I love WikiLeaks!” Trump declared in October, and for good reason: In short order, those leaks dominated media attention, distracting from allegations of improper sexual conduct by Trump that were slowly emerging in the wake of the release of the “Access Hollywood” tape. It was really only later that month, with former FBI director James B. Comey’s announcement that the FBI had obtained new emails related to Clinton’s personal server, that attention paid to WikiLeaks — to that material apparently stolen by Russian actors — was overtaken.

But those hacks were not the only thing that Russia was allegedly doing. A group called the Internet Research Agency (IRA) created fake social media accounts, bought ads on Facebook and other places, and posted information meant to distort, disrupt and influence the political conversation in the United States. These were the infamous Russian trolls, interested in affecting the election by changing people’s minds about the U.S. political system or about Trump and Clinton.

This effort has garnered a lot of attention for the same reason that those document releases in October 2016 did: The slow accretion of new details over the course of the past year or so has made the effort seem increasingly important. With each new revelation — The trolls bought ads! They had a budget of millions of dollars a month! — the understood scope of the effort expands outward and, therefore, appears to be ever bigger.

Particularly considering the scale of a presidential election, it wasn’t big. There’s no evidence that it really even had much of an effect. For all of the attention it has garnered and the novelty the effort admittedly involved, it doesn’t seem to have been worth the effort.

Consider the effort to dissuade black voters from supporting Clinton. Two reports presented to the Senate this week outlined what that looked like: A push to engage black social media users and then advise them to vote for a third-party candidate or stay home. One Facebook page, called Blacktivist, had 4.5 million likes. Of all of the IRA’s Facebook pages, none got more reactions than Blacktivist, according to a report compiled by a team including staffers from the analysis firm New Knowledge.

It’s important to remember the genesis of this effort. According to an indictment obtained by special counsel Robert S. Mueller III against the IRA and individuals associated with it, the push to disrupt U.S. politics began in 2014 — a point at which one of the most divisive issues was the Black Lives Matter movement and the shooting of a black teenager in Ferguson, Mo.

That’s one reason that Missouri was disproportionately targeted by the IRA’s Facebook ads, as a report from a team from the Computational Propaganda Research Project makes clear.

Another target of the ads was Maryland, where protests erupted after the death of Freddie Gray in police custody in 2015.

These, you will notice, are not places where the 2016 election results were in question. Trump won Missouri by 19 points and lost Maryland by 26. Why target those states so heavily for ads? Because the goal was, in part, to foment disruptions.

Those were paid ads. What about the organic social media efforts of pages such as Blacktivist?

The Computational Propaganda Research Project report includes this chart, showing the number of likes on several IRA-backed pages by month. Black Matters and Blacktivist are shown; each emphasized issues of race in the African American community.

Notice, though, that the big spike in engagement happened after the election. Much of the analysis of these issues looks at the full scope of the IRA’s effort, running from 2014 into 2017.

So let’s say that those two pages got half a million likes in the closing month of the campaign, an assumption that isn’t really justified based on the above graph. Last year, the company reported that 800 million people like something on Facebook every day around the world. That’s individuals, not actual likes; presumably many like more than one thing. That’s about 25 billion people clicking “like” at least once in a month. Just for a sense of scale.

The New Knowledge report looked specifically at the volume of posts over the past few months of the campaign. There were more than 100,000 Twitter posts a week in early October — out of 500 million sent globally each day. There were only a few hundred posts on Facebook and Instagram.

More important, that report documented what the content posted to the pages targeting black users looked like as the campaign wound down to Election Day on Nov. 8.

On Nov. 5, 2016, “The accounts that appeared to be targeting and engaging with Black audiences focused on an entirely different conversation, almost entirely absent any mention of ‘Trump’ or ‘Clinton,’ ” it reads, later adding that the “sparse political content (~5 posts) included commentary about the Jay-Z-Beyoncé concert for Hillary."

On Nov. 6, “Once again, the original content on the left focused nearly exclusively on themes of Black culture, police brutality, and Black erasure — the election was barely mentioned.” There was one post on Instagram “presenting a range of voter suppression/depression narratives,” the report read. It's not clear how much attention that one post generated.

On Nov. 7, a pair of YouTube commentators calling themselves “Williams & Kalvin” posted a YouTube video called “The truth about elections” which, per the report, explained why black Americans shouldn’t vote. Someone took an image of that video in 2017, before it was removed; it had been viewed 257 times.

Another post highlighted in the report was from an Instagram account similarly arguing that black people should stay home on Election Day. “The post,” the report states, “got 453 engagements."

This was the day before the election, on one of the few posts targeting black voters that was specific enough about suppression to warrant mention. And fewer than 500 people engaged with it.

Not that many posts. Not that much engagement. Not particularly focused on the election.

Even stepping back from the content targeting black Americans (which, again, was much more frequent after the election ended), the numbers at play are very small. By contrast, The Washington Post had nearly 100 million unique visitors in October 2016. Other media outlets had tens of millions of visitors. The campaigns spent hundreds of millions of dollars on ads and outreach; the Trump campaign alone was running tens of thousands of different ads each day on Facebook. Those numbers quickly dwarf what the IRA was trying to do.

Black voter turnout did drop in the 2016 election relative to 2012, but there’s a non-Russian factor worth mentioning: Barack Obama was on the ballot in 2012 and not in 2016. Turnout among black voters surged in 2008 when Obama was first on the ballot and dropped eight years later. Yes, Trump won the presidency narrowly thanks to small margins of victory in Michigan, Pennsylvania and Wisconsin, but there’s no indication that this was a function of black voters persuaded by Russian trolls to stay home.

More broadly, there’s no evidence that the Russian social media efforts were particularly sophisticated in their targeting or messaging. (Most of the ads targeting Wisconsin, for example, ran during the primaries.) There’s little evidence that they achieved the sort of scale that would have any significant effect on voters.

Russia does seem to have altered the trajectory of the 2016 election. But this appears to have happened not by leveraging social media in clever and focused ways but by leveraging the traditional media to cover the WikiLeaks dumps.