We can't allow ourselves to forget Russia's attacks on our election

Carter wasn’t the first former U.S. official to make this type of statement: Several months ago, former Director of National Intelligence James Clapper stated that Russia swung the election for Trump.

Clapper [made] the argument that with the Russian Troll farming reaching 127 Million Americans in an election where only 130 Million people voted, it’s pretty likely that they had an impact.  He then goes further and says that in the end with less than 80,000 votes in three key states making the final difference, and with the Russians generating ads that linked Hillary to #BlackLivesMatter which were specifically targeted to White conservative voters which were intended to get them really, really pissed off — that was likely enough to make the difference.

Clapper also points out that the joint intelligence assessment about Russia’s attack on our election from 2017 doesn’t make the case that “Russia Failed” to have an impact, as Trump repeatedly likes to say.  He points out that that assessment didn’t even try to answer that question because it was outside the scope of what the intelligence agencies do.

Neither Clapper or Carter made a definitive claim; both argued that if only someone, somewhere would bother to study the issue in depth, they could prove the case that, if not for the difference made by Russia’s efforts, Hillary Clinton would have won the 2016 election.

However, it’s possible that we do have enough data to make that determination now, without a specific study intended to measure the impact of the Russian meddling efforts. More important, why hasn’t anyone yet done a comprehensive study to either prove or disprove this assertion? The integrity of our future elections could depend on knowing and understanding the intimate details of what has happened previously.

Despite the claims of Trump and his supporters that the initial Intelligence community assessment of the Russian effort determined that it “didn’t affect the election,” the truth is that that assessment didn’t make any determination at all on the impact of Russian meddling on the final election results.

We did not make an assessment of the impact that Russian activities had on the outcome of the 2016 election.  The US Intelligence Community is charged with monitoring and assessing the intentions, capabilities, and actions of foreign actors; it does not analyze US political processes or US public opinion.

Similarly, the Mueller report does not make an assessment of the results of the Russia attack on our elections, but it does state that its efforts to affect those elections were systematic and sweeping.

The Russian government interfered in the 2016 presidential election in sweeping and systematic fashion. Evidence of Russian government operations began to surface in mid-2016. In June, the Democratic National Committee and its cyber response team publicly announced that Russian hackers had compromised its computer network. Releases of hacked materials hacks that public reporting soon attributed to the Russian government began that same month.Additional releases followed in July through the organization WikiLeaks, with further releases in October and November.

[…]

As set forth in detail in this report, the Special Counsel’s investigation established that Russia interfered in the 2016 presidential election principally through two operations. First, a Russian entity carried out a social media campaign that favored presidential candidate Donald J.Trump and disparaged presidential candidate Hillary Clinton. Second, a Russian intelligence service conducted computer-intrusion operations against entities, employees, and volunteers working on the Clinton Campaign and then released stolen documents. The investigation also identified numerous links between the Russian government and the Trump Campaign. Although the investigation established that the Russian government perceived it would benefit from a Trump presidency and worked to secure that outcome, and that the Campaign expected it would benefit electorally from information stolen and released through Russian efforts, the investigation did not establish that members of the Trump Campaign conspired or coordinated with the Russian government in its election interference activities.

These assessments don’t claim that Russian interference changed the results of the election and don’t indicate that Russian hacking efforts change any votes—but they also don’t say that Russia had no impact at all, as Trump and his supporters have claimed; and they do indicate that Russia had a clear and deliberate goal of aiding the campaign of Donald Trump, despite the claims of former DHS Secretary Kirstjen Nielsen, who said “she hadn’t seen this information.”

“I haven’t seen any evidence that the attempt to interfere in our election infrastructure was to favor a particular political party,” Nielsen responded when asked about Russia’s intentions Thursday during the Aspen Security Forum.

“What we’ve seen on the foreign influence side is they were attempting to intervene and cause chaos on both sides,” she continued.

But Mueller’s indictments of members of the Russian troll farm known as the Internet Research Agency and his later report make it clear that this isn’t the case.

As the report details, in the early stages of its U.S.-focused political operations, the IRA mostly impersonated U.S. citizens but into 2015 it shifted its strategy to create larger pages and groups that pretended to represent U.S.-based interests and causes, including “anti-immigration groups, Tea Party activists, Black Lives Matter [activists]” among others.

The IRA offered internal guidance to its specialists to “use any opportunity to criticize Hillary [Clinton] and the rest (except Sanders and Trump – we support them” in early 2016.

While much of the IRA activity that we’ve reported on directly sowed political discord on divisive domestic issues, the group also had a clearly stated agenda to aid the Trump campaign. When the mission strayed, one IRA operative was criticized for a “lower number of posts dedicated to criticizing Hillary Clinton” and called the goal of intensify criticism of Clinton “imperative.”

All of this indicates that, far from being ineffective and inconsequential, the efforts of the troll farm clearly had some type of impact; what we don’t know in detail is exactly how small or large that impact actually was. We know that its goal was to influence the election, but exactly how did it go about it? Is there any evidence that its specific messages managed to get through?

Many of the defensive claims from the Trump camp have been that, although the election systems of many states were attacked, and many were penetrated, there isn’t any evidence that votes were changed or that voter registration databases were impacted. But it’s not like there was a lack of trying by the Russians; on election night, they were quite busy.

Reports at the time were that it’s highly likely that Russian hackers changed votes in the seven states where they successfully penetrated election tabulation systems: Alaska, Arizona, California, Florida, Illinois, Texas, and Wisconsin.

The U.S. intelligence community developed substantial evidence that state websites or voter registration systems in seven states were compromised by Russian-backed covert operatives prior to the 2016 election — but never told the states involved, according to multiple U.S. officials.

Top-secret intelligence requested by President Barack Obama in his last weeks in office identified seven states where analysts — synthesizing months of work — had reason to believe Russian operatives had compromised state websites or databases.

Three senior intelligence officials told NBC News that the intelligence community believed the states as of January 2017 were Alaska, Arizona, California, Florida, Illinois, Texas and Wisconsin.

This is in addition to election results in four states that swung wildly for Trump in comparison to exit poll estimates.

Using exit poll totals compiled by election researcher Theodore de Macedo Soares, seen in the table below on this page as well as available at this link, compared to an ongoing tally of raw votes totals posted at this link by Dave Wasserman of Cook Political Report, here is the discrepancy that has caused the suspicions of what Trump himself would call a “rigged election.”

Note that actual vote totals are compiled as of November 16.

FLORIDA — 29 Electoral Votes

(numbers equal percentage points)

Exit Polls: Clinton 47.7, Trump 46.4 — Clinton wins by 1.3

Actual: Clinton 47.8, Trump 49.0 — Trump wins by 1.2

Trump gain between exit polls and actual results: 2.5

NORTH CAROLINA — 15 Electoral Votes

Exit Polls: Clinton 48.6, Trump 46.5 — Clinton wins by 2.1

Actual: Clinton 46.1, Trump 49.9 — Trump wins by 3.8

Trump gain: 5.9

PENNSYLVANIA — 20 Electoral Votes

Exit Polls: Clinton 50.5, Trump 46.1 — Clinton wins by 4.4

Actual: Clinton 47.6, Trump 48.8 — Trump wins by 1.2

Trump gain: 5.6

WISCONSIN — 10 Electoral Votes

Exit Polls: Clinton 48.2, Trump 44.3 — Clinton wins by 3.9

Actual: Clinton 47.6, Trump 48.8 — Trump wins by 1.2

Trump gain: 5.1

Pennsylvania’s vote tally systems are vulnerable to remote accessthere were reports there of “votes flipping.” Georgia’s election system was also specifically targeted by hackers. And as former Senator Bill Nelson stated, Florida’s election systems were also penetrated. The Mueller report also confirms that election registration systems and voting machine companies were hacked.

The Russian military intelligence unit known by its initials GRU targeted U.S. state election offices as well as U.S. makers of voting machines, according to Mueller’s report.

Victims of the Russian hacking operation “included U.S. state and local entities, such as state boards of elections (SBOEs), secretaries of state, and county governments, as well as individuals who worked for those entities,” the report said. “The GRU also targeted private technology firms responsible for manufacturing and administering election-related software and hardware, such as voter registration software and electronic polling stations.”

The Russian intelligence officers at GRU exploited known vulnerabilities on websites of state and local election offices by injecting malicious SQL code on such websites that then ran commands on underlying databases to extract information.

Using those techniques in June 2016, “the GRU compromised the computer network of the Illinois State Board of Elections by exploiting a vulnerability in the SBOE’s website,” the report said. “The GRU then gained access to a database containing information on millions of registered Illinois voters, and extracted data related to thousands of U.S. voters before the malicious activity was identified.”

Because so many of our elections systems are “offline” and not specifically connected to the internet, it’s been assumed that they are “unhackable”—but that doesn’t mean that a virus introduced into the election server couldn’t be transmitted to a voting machine through the “sneakernet” of removable media, such as files placed on a CD-R or thumb drive. That happens to be how the U.S. was able to get the Stuxnet virus introduced into Iran’s uranium centrifuge systems a decade ago, even though those systems also aren’t connected to the internet.

Unfortunately, digital or paper audits of voting records to confirm the reason for 2-5 point swings in election results compared to exit polls simply aren’t available, and all we can do in this area is speculate. There is no definitive proof that Russian hackers changed votes, but there isn’t really any definitive proof that they absolutely didn’t, either. It’s far more likely that they may have impacted some of the registration machines, which caused long lines at the polls on Election Day, thereby suppressing votes. This tactic could have been effective if it was used against urban election centers, which would tend to be largely Democratic.

Over a year ago, last February, I did a long diary about this same subject, in which I attempted to follow the various changes in the polls between Trump and Clinton during the last six months of the campaign, and one thing that was fairly clear was that one of the largest impacts on the election was the letter by James Comey to Congress announcing that the investigation into Clinton’s email server had been “re-opened” just a week before the election. Prior to that, Clinton had held a substantial lead ever since the Democratic National Convention in late July.

Impact of the James Comey letter release

Prior to that, Trump had made his strongest surge in early October as the debates took place; this was also the same time that the Access Hollywood tapes were released, and the same day that WikiLeaks released the Podesta emails.

Debates.png Impact of the Debates, Access Hollywood and Podesta Emails

FiveThirtyEight documented the impact of Comey’s letter on Clinton’s chances as follows, particularly when it came to the three key states of Michigan, Pennsylvania, and Wisconsin, which she ultimately lost by a razor-thin margin.

Nonetheless, Clinton lost Michigan, Pennsylvania and Wisconsin by less than 1 percentage point, and those states were enough to cost her the election. She lost Florida by just slightly more than 1 point. If the Comey letter had a net impact of only a point or so, we’d have been in recount territory in several of these states — but Clinton would probably have come out ahead. I call this the “Little Comey” case — sure, the Comey letter mattered, but only because the election was so close.

Even a small Comey effect could have cost Clinton the 270 electoral votes she needed to win

ADJUSTED VOTE MARGIN
CLINTON VOTE MARGIN SMALL COMEY EFFECT* BIG COMEY EFFECT* CLINTON’S ELECTORAL VOTES HAD SHE WON
Michigan -0.2 +0.8 +3.8 248
Pennsylvania -0.7 +0.3 +3.3 268
Wisconsin -0.8 +0.2 +3.2 278
Florida -1.2 -0.2 +2.8 307
Nebraska’s 2nd C.D. -2.1 -1.1 +1.9 308
Arizona -3.5 -2.5 +0.5 319
North Carolina -3.7 -2.7 +0.3 334
Georgia -5.1 -4.1 -1.1 350
Ohio -8.1 -7.1 -4.1 368
Texas -9.0 -8.0 -5.0 406
Iowa -9.4 -8.4 -5.4 412

*Adjusting for a small Comey effect adds 1 percentage point to Clinton’s vote margin. A big effect adds 4. Hypothetical scenario starts with the 232 electoral votes Clinton actually won, ignoring faithless electors.

Clearly there were many different specific events that impacted the polls and electorate during the campaign, from Comey’s letter to Clinton’s failure to visit Michigan and Wisconsin more often, beyond just what Russia did with its hacking and the disinformation campaign conducted by the Russian troll farm. As Nate Silver wrote on FiveThirtyEight when Mueller issued his indictments regarding the Russian troll farm, it’s exceedingly hard to tell what the impact of the Russian efforts was, or what it wasn’t.

Overall, then, my view on the effects of Russian interference is fairly agnostic. I tend to focus more on factors — such as Clinton’s email scandal or the Comey letter (and the media’s handling of those stories) — that had easier-to-prove effects. The hacked emails from the Clinton campaign and the DNC (which may or may not have had anything to do with the Russians) potentially also were more influential than the Russian efforts detailed in Friday’s indictments. Clinton’s Electoral College strategy didn’t have as much of an effect as some people assume — but it was pretty stupid all the same and is certainly worth mentioning.

But if it’s hard to prove anything about Russian interference, it’s equally hard to disprove anything: The interference campaign could easily have had chronic, insidious effects that could be mistaken for background noise but which in the aggregate were enough to swing the election by 0.8 percentage points toward Trump — not a high hurdle to clear because 0.8 points isn’t much at all.

Consequently, it’s difficult to pin the final results on any single event or element; but at the same time, it’s difficult to sustain the claim that Russia’s efforts made no difference at all, particularly since some of the bogus Facebook groups that were created by false Russian accounts did result in actual rallies and protests that took place in America, which were seen by more than 300,000 people.

Posing as American activists, Russian government-linked trolls created 129 Facebook events between 2015 and 2017.

On multiple occasions, the events prompted real Americans to take to the streets.

In a written statement Facebook gave to the Senate Intelligence Committee released on Thursday, the social media network said that the events created by one Kremlin-linked troll group were seen by more than 300,000 Facebook users. About 62,500 users marked that they would attend the event, and an additional 25,800 expressed an interest in attending.

Facebook told Congress it does “not have data about the realization of these events,” but CNN has previously found evidence that the Russian group successfully convinced Americans to attend the demonstrations.

Russians were able, in just one case, to get over 60,000 Americans to attend a real-life political rally. Is it that hard to imagine that they may have been able to impact at least that many, if not more, in terms of the final vote on Nov. 8, with the .08 swing that was needed to give Trump his win in the three key states that gave him the Electoral College?

This rally was far from a one-off. Mueller reported that there were dozens of these types of rallies that had been instigated by Russians, and that Facebook, Twitter, and Instagram posts generated by the Russians were actually reposted and retweeted repeatedly by members of the Trump campaign.

Special counsel Robert Mueller in his highly-anticipated report said his team identified “dozens” of U.S. political rallies organized on social media by the Internet Research Agency (IRA), a Russian troll farm that was later indicted for attempting to interfere with the 2016 presidential election.

[…]

The troll farm used its Facebook and Twitter accounts to organize and promote U.S. political rallies, often sending direct messages to its followers on social media asking them to participate in the events, Mueller wrote.

“IRA employees frequently used … Twitter, Facebook and Instagram to contact and recruit U.S. persons who followed the group,” Mueller wrote. The IRA targeted and recruited racial justice advocates as well as the moderators of conservative groups.

The Hill in 2017 reported that thousands of Americans attended a march organized by the troll farm in New York City. The IRA used a group called “BlackMattersUS” to coordinate the event.

The Mueller report also lays out the ways that Trump campaign officials and surrogates amplified the IRA’s messages on Twitter and Facebook as they sought to interfere in public discourse and amplify divisive political rhetoric.

Trump campaign officials, including senior adviser Kellyanne Conway and Donald Trump Jr., cited and retweeted content from the troll farm about topics including voter fraud and Clinton’s handling of classified information, according to Mueller. 
Mueller’s team found that Trump campaign affiliates promoted “dozens” of tweets, posts and other political content created by the IRA.

There were dozens of times that posts and groups created by Russian trolls actually managed to have a direct influence on Americans in relation to the election and public policy issues.

We know that Russians were attempting to hurt Hillary Clinton and help Donald Trump, and that they did have some impact on Americans. The remaining questions are, how many Americans were impacted, and did the messages that the Russians were sending resonate with large portions of the American public? A report generated by the Senate Intelligence Committee with the cybersecurity firm New Knowledge documents the methods that were used by the IRA in its posts and groups. These were broken down into three types.

  • Right-wing posts that were intended to inspire anger and outrage in order to push voters to the polls.
  • Left-wing posts that were intended to inflame internal Democratic tensions and either cause Democrats not to vote or encourage them to vote for Jill Stein in protest of Hillary Clinton’s battle with Bernie Sanders.
  • African American-focused posts that were intended to heighten frustrations with the police, American society as a whole, and the U.S. government, intended to cause black Americans not to vote.

The report indicates that the number of posts and bot-generated reposts across Facebook, Twitter, and Instagram were much more numerous and had a much farther reach than was previously known.

The report by New Knowledge is based on a review of 10.4 million tweets, 1,100 YouTube videos, 116,000 Instagram posts, and 61,500 unique Facebook posts published from 2015 through 2017. This is not a complete data set of Russian influence operations, but it’s still the largest such analysis to take place outside of the companies themselves. And it shows that the Russians weren’t just running a bland content farm, churning out propaganda in broken English. The operation was deeply sophisticated, and at times, downright funny. As the report’s authors note: “The IRA was fluent in American trolling culture.”

[…]

Nevertheless, the report lays out ample obvious examples of how Facebook and Twitter were both used to discourage turnout. In some cases, the trolls tried to mislead people into texting their votes. In others, they encouraged Americans to vote for third-party candidates like Jill Stein or give up on voting all together, with messages that read “F*CK THE ELECTIONS.”

[…]

The focus on police brutality and content targeting African Americans wasn’t limited to YouTube. Among more than a dozen web domains the IRA registered, the vast majority, including DoNotShoot.us and Blacktivist.info, were aimed at black communities. Of the 33 most popular Facebook pages linked to the IRA, nearly half focused on black audiences. This effort was particularly successful on Instagram, where the account @blackstagram_ amassed more than 300,000 followers and elicited more than 28 million reactions. Much of this content seemed designed to stoke distrust among African Americans in democratic institutions and depress black turnout for Democratic candidate Hillary Clinton.

Many of the posts by the troll accounts weren’t specifically election-related and seemed to be basically serving up “red meat” to their targeted audience to help build their brand and influence over the course of a couple of years, starting as early as 2014. However, as the election drew near, they grew more political, with specific suggestions for voting or not voting.

“Choose peace and vote for Jill Stein,” said one ad from Blacktivist, a fake Black Lives Matter account, in an ad that ran from Nov. 3 through the election and received over 18,000 impressions. “Trust me, it’s not a wasted vote.”

The troll account Williams and Kalvin ran an ad on Nov. 7 and Nov. 8, 2016, that targeted users interested in Martin Luther King Jr., Malcolm X, the civil rights movement and the presidential election. “Hillary Clinton is a traitor! Hillary Clinton is a liar! Hillary Clinton is insane!” the message started. “I know that many black people support this old dirty bitch. I don’t know why they do this, still it’s their personal choice and we are a free country yet. But, listen to my word of truth and don’t let them fool you.”

“We don’t have any other choice this time but boycott the election,” said another Election Day ad run by Williams and Kalvin. “This time we choose between two racists. No one represents Black people. Don’t go to vote.”

In addition to the 127 million Americans who were reached through Facebook, another 187 million were reached through Instagram, through literally hundreds of fake accounts and bots that would perform automatic retweets in order to cause their posts to trend. At least 40% of these Russia-generated accounts on Instagram had more than 10,000 followers, most of whom were real people.  That’s on top of 3,800 Twitter accounts that attracted more than 1.4 million followers with 73 million engagements and dozens of fake news websites, which were used to create false content or push advantageous memes. The Wired article on the New Knowledge report ends on this sobering note.

What these millions of digital artifacts do show, when taken together, is just how much planning and coordination went into the IRA’s scheme. Between the Twitter handles, the Facebook pages, the Instagram posts, the YouTube personalities, the fake local news sites, and in at least one case, a phony geopolitical think tank, the trolls created their own mini-internet to prop up Trump and spread distrust in his opponent and the election system itself. What’s more, their efforts remain ongoing, years after the election. Using the trove of data, the researchers were able to uncover even more IRA-linked Facebook pages, including one that was updated as recently as May of this year.

All of this demonstrates, according to the report authors, that “over the past five years, disinformation has evolved from a nuisance into high-stakes information war.” And yet, rather than fighting back effectively, Americans are battling each other over what to do about it. “We have conversations about whether or not bots have the right to free speech, we respect the privacy of fake people, and we hold congressional hearings to debate whether YouTube personalities have been unfairly downranked,” the report reads. “It is precisely our commitment to democratic principles that puts us at an asymmetric disadvantage against an adversary who enthusiastically engages in censorship, manipulation, and suppression internally.”

It should also be noted that the activities of the trolls and bots didn’t end with the election. They continued on into 2017, pushing memes that attacked James Comey and later Robert Mueller’s investigation in a manner that exactly mirrored Trump’s own use of social media to claim repeatedly that there was “No Collusion, No Obstruction.” They didn’t stop until most of their accounts were taken down by Facebook, Twitter, Instagram, and Youtube, but that doesn’t mean that the social media services have eliminated all of the accounts, or that new troll accounts haven’t been established since.

In my previous article I pointed out that it really didn’t take much more than influencing 6,000 voters to make the final difference in the election, mostly by discouraging Democratic voters who still supported Bernie Sanders and blamed the DNC for allegedly sandbagging his campaign, as was claimed by WikiLeaks, pushing them either not to vote at all or to switch their support to Jill Stein.

For actual data from the Guide to the 21016 Cooperative Congressional Election Survey, 16% of those who voted for Sanders in the PA primary voted for Trump in the general.  In WI and MI, it was 9% and 8% respectively.  To put this into raw numbers, Sanders-to-Trump voters ultimately gave Trump the margin he needed to win in each of those states:

  • In Wisconsin, roughly 51K Sanders voters backed Trump in a state he won by just 22K votes.
  • In Michigan, roughly 47K Sanders voters backed Trump in a state he won by just 10K votes.
  • In Pennsylvania, roughly 116K Sanders voters backed Trump in a state he won by just 44K votes.

Keep in mind that a Sanders-to-Trump vote is doubly painful because it likely represents a net +2 for Trump (absolute +1 vote for Trump and likely -1 vote for Clinton).  Also, the above analysis in these three states is before you even get to Sanders voters who protest voted for Stein/other or didn’t vote at all.  And these folks were not Rs.  They were generally ideologically progressive and voted for Dems in the past.

Again, if Russian trolls were able to generate real-world public rallies that with attendance as high as 60,000 people, and had an online presence that could reach 120 million people on Facebook and over 180 million people on Instagram, with the specific goal of energizing the GOP vote for Trump and discouraging the Democratic vote for Clinton, it’s not exactly beyond the realm of possibility that, with a fairly average click-through rate of just 0.5%—which would have produced 900,000 potential clicks for Instagram and 600,000 for Facebook—that they could have easily created the 77,000-vote shift across Wisconsin, Michigan, and Pennsylvania that we ultimately saw. Many things pushed Clinton to the edge of the cliff all on her own, leaving her with little room for error. The combination of Comey and the Russians were then enough to push her over the edge and usher Trump into the White House. But that wouldn’t have been possible if she hadn’t made her own mistakes. This didn’t happen in isolation, and it wasn’t the only thing that had an impact on the election; but it’s hard to imagine that anything the Russians did were to Clinton’s benefit.

In fact, based on the millions of people we do know were reached and engaged by the trolls, and the tens of thousands we do know physically responded to their efforts, it seems hard to believe that the entire impact of the Russian effort was just 77,000 votes and only 0.8%. We don’t know how much they impacted things, but based on how many people they actually reached, it was probably much more than what we’ve been told. We won’t know for sure until someone goes through and specifically correlates troll activities by state to shifts in the polls.

Continue reading...