Retconning ‘Russiagate’

On June 26, the CIA released a Tradecraft Review of the 2016 Intelligence Community Assessment (ICA) on Russian election interference. That review—an after-action audit—focused on whether the ICA met analytic tradecraft standards. It focused particularly on one sentence: the ICA’s assessment that Russian President Vladimir Putin “aspired to help Donald Trump win.” 

The review questioned whether the “high confidence” level for that conclusion was appropriate, ultimately finding that a “moderate” level would have been more appropriate. It also highlighted CIA objections to FBI leadership’s insistence on including the so-called Steele dossier in the ICA. The dossier had not met basic tradecraft standards and was viewed by senior CIA analysts as unverified and inappropriate for inclusion. Ultimately, it was appended as a two-page annex with a disclaimer stating it was not used to reach the analytic conclusions. However, the review noted that by referencing the annex in a supporting bullet for the “aspired” judgment, the ICA implicitly elevated unsubstantiated claims, thereby undermining the credibility of that key assessment.

Notably, however, the new review did not dispute the conclusion itself: That the Russians had favored the election of Donald Trump was not in question, nor was the judgment that they aspired to help. The Tradecraft Review was a process critique, not a substantive refutation. 

Still, the release was immediately spun as exactly that. CIA Director John Ratcliffe gave an exclusive to the New York Post, which packaged a tradecraft audit that reassessed a confidence level as proof that the 2016 ICA was a “politically corrupted” hit job designed to “screw Trump.” It framed the audit as evidence that intelligence leaders had intentionally manipulated the ICA to damage Trump politically and to mislead the public about Russia’s true intentions. “All the world can now see the truth: Brennan, Clapper and Comey manipulated intelligence and silenced career professionals — all to get Trump,” Ratcliffe proclaimed as he promoted the article on X from the @CIADirector account. 

Substack writer Matt Taibbi, who last year published a theory that the CIA had “cooked the intelligence” to hide that Russia had actually favored Clinton—based on the opinions of Kash Patel and anonymous sources close to Devin Nunes’s House Intelligence Committee team—seized on the CIA review as vindication of his take. This maximalist retcon gained traction quickly on X—despite running headlong into an enormous body of evidence that came out after the January 2017 release of the ICA.

In science fiction and fantasy universes, writers occasionally find themselves constrained by earlier plot points that stand in the way of where they’d like a story to go. To resolve this, they revise the canon: Past events are rewritten, reinterpreted, or conveniently forgotten. This act of rewriting the past to fit the desirable present is called a retcon in fandom speak—short for retroactive continuity. 

The term has a new and disconcerting relevance in American politics—most recently where the events of Russia’s interference in the 2016 election are being actively reframed.

The term “Russiagate” originally referred to suspicions of collusion between Trump’s 2016 presidential campaign and Russian actors—a tangled set of claims, many overstated in constant speculative media coverage. Over time, however, the term’s meaning flipped, and it is now used primarily as a signifier on the right for “the discredited Russia hoax.” But it’s unclear what the boundaries of the “hoax” actually are, among those who use it. In its strong form, “Russiagate” perhaps still references collusion. But in its maximalist form, it increasingly rejects the idea that Russia did anything in the 2016 election at all. On platforms like X, the conversation has come to include radical propositions: that interference never happened in the first place, or that Vladimir Putin actually preferred Hillary Clinton over Donald Trump.

The narrative rewrite is not subtle. It would be one thing if it were limited to extremely online conspiracists, but the flames are now being fanned by government officials. In this most recent effort, authority figures like the CIA director are pointing to the “evidence” of the Tradecraft Review, as if the abundance of investigations that came after it simply never happened. 

In the eight years since the CIA’s January 2017 ICA release, multiple subsequent investigations—by Congress, by Special Counsel Robert Mueller, by journalists, and by academic researchers—have documented how Russian information operations functioned during the Clinton-Trump contest. The Russian Internet Research Agency (IRA) and its military intelligence agency, colloquially referred to as the GRU, demonstrably interfered in the 2016 election. These operations were not subtle. They unambiguously boosted Trump’s candidacy. They unambiguously denigrated Clinton. So it’s worth a quick review of the data that emerged, to anchor the current conversation in reality. 

In 2018, I led one of several independent investigations into a dataset turned over by social media platforms to the Senate Select Committee on Intelligence (SSCI). A second project in 2019 focused on a separate Facebook dataset attributed to GRU activity. These projects were convened under the committee’s TAG (Technical Advisory Group) model: Bipartisan SSCI leadership, at the time Sens. Mark Warner (D-Va.) and Richard Burr (R-N.C.), invited some researchers to field teams that would produce analyses, briefings, and public testimony to inform the committee’s own investigation. 

Participation was an act of service. The process was designed to be blind; the researchers asked to participate didn’t know who else was involved.

The reason for the blinding process was to ensure that the reports produced were independent and insulated from groupthink or political bias. Some of the content turned over included disinformation campaign material that was politically controversial—primary source posts, memes, and ads documenting activities that occurred in the lead-up to the 2016 election. If one analysis team made an incendiary claim about the content related to the election, having independent alternate perspectives might prove important. 

As it turned out, there was very little daylight between the findings of the two teams that completed reports. Each found that the IRA’s operation pushed a strikingly one-sided political line: Elevate Donald Trump, hobble Hillary Clinton, and dampen turnout among likely Democratic constituencies. 

The report produced by my team, “The Tactics & Tropes of the Internet Research Agency,” documents a brief flirtation with boosting Rand Paul during the early days of the Republican primary—and then a hard pivot to Trump when it became clear that Paul was a nonstarter. From July 2015 onward, the trolls ridiculed Ted Cruz, Marco Rubio, and Jeb Bush on Facebook pages and Instagram accounts where they pretended to be fellow conservatives. Once Trump became the nominee, they continued cheerleading on the right—and mounted a sustained, textbook barrage of voter-suppression narratives on their Black community- and left-targeted pages. (The report contains a detailed qualitative analysis describing activity in the week leading up to the election.)

The Graphika and Oxford Internet Institute report, “The IRA, Social Media and Political Polarization in the United States, 2012-2018,” independently corroborates this via its own analyses: Explicit pro-Trump messaging scaled sharply as his campaign gathered steam, while African American, LGBT, and other left-leaning communities were urged to boycott or “tune out” the election. The bottom line: “While the IRA strategy was a long-term one, it is clear that activity between 2015 and 2016 was designed to benefit President Trump’s campaign,” using a twin strategy of right-wing mobilization and left-wing demobilization.

These two reports, in turn, informed SSCI’s own report, “Volume 2: Russia’s Use of Social Media,” the second part of SSCI’s five-volume study of Moscow’s 2016 interference. The committee drew not only on the TAG working-group analyses that the two teams supplied but on additional interviews with researchers, closed briefings from the FBI and intelligence community, open-source media exposés, academic work on computational propaganda, and Special Counsel Mueller’s indictments of troll factory leadership. Weaving those streams together, the SSCI reached corroborating findings:

• Directed intent: At the Kremlin’s behest, the IRA sought to harm Clinton’s chances of success and support Trump. Pre-election posts about Trump were “mostly positive,” for relevant audiences, while references to Clinton were “uniformly negative” from both right- and left-aligned troll personas.

• Voter-suppression focus: Left-leaning audiences—especially African Americans—were bombarded with turnout-depression ads and memes; no comparable effort targeted the ideological right.

• Real-world activation and persistence: Troll accounts orchestrated pro-Trump rallies (“Florida Goes Trump”), staged dueling protests to inflame social fault lines, and increased their posting rates after the election, underscoring that 2016 was one battle in a longer information-warfare campaign.

Commentary from trolls working on the campaigns is featured in the SSCI report: “On November 9, 2016, a sleepless night was ahead of us. And when around 8 a.m. the most important result of our work arrived, we uncorked a tiny bottle of champagne … took one gulp each and looked into each other’s eyes …. We uttered almost in unison: ‘We made America great.’” Indictment documents detailing tasking noted: “IRA employees were directed to focus on U.S. politics and to ‘use any opportunity to criticize Hillary and the rest (except Sanders and Trump—we support them)’.”

Taken together, the TAG reports and SSCI’s synthesis across a broader evidence base paint an unequivocal picture: The Russian operation’s messaging around election time in 2016 was not a symmetric “chaos” play—it was an asymmetric push that boosted Trump while sabotaging Clinton and suppressing key Democratic blocs, executed across every major platform and reinforced by real-world provocations. 

This does not mean that it worked. It does not mean that it swung the election. But it was also not remotely ambiguous. And, at the time of the report’s release, it was not particularly controversial: Then-Sen. Marco Rubio, as it happened, was on the committee at the time; he not only acknowledged the effort but also noted that the Russians had tried to hack his campaign. He would become acting chair of SSCI several months later, presiding over the committee for releases of later volumes of the bipartisan Russian report.

The GRU hack-and-leak operation—which the Senate detailed in Sub-section VIII-A of “Volume 2: Russia’s Use of Social Media”—was not ambiguous either. The committee describes how GRU Unit 74455 stole Democratic committee and campaign emails, laundered them through the DCLeaks and Guccifer 2.0 personas, and then pushed the material across Facebook, Twitter, and sympathetic journalists. Facebook’s own forensics confirmed that Russian military intelligence threat actor APT-28 personas were “used to seed stolen information to journalists,” while the July 2018 Mueller indictment spelled out Unit 74455’s role in “the release of stolen documents … and the publication of anti-Clinton content on social media accounts operated by the GRU.” The first dump of emails happened later in the afternoon the day the “Grab ’Em by the Pussy” Access Hollywood tape was released—changing the national conversation. 

A small team I led at Stanford Internet Observatory conducted a 2019 audit of a dataset that Facebook had attributed to the GRU and turned over to SSCI. We found that the GRU threat actors behind the hack were operating Facebook pages and sock-puppet news sites to launder those dumps, and others, into Western media. The dataset included material related to other GRU-attributed hacks that Facebook had connected to the same account actors on its platform—for example, the World Anti Doping Association hack—bolstering the attribution. There were stark stylistic differences in how the GRU ran its propaganda efforts versus the IRA—long-form front media as opposed to memes and trolling—leading us to a conclusion similar to SSCI’s conclusion: that Moscow was running parallel but complementary lines of effort. Military-intelligence hack-and-leak operations were running alongside troll-farm persuasion, with each ultimately geared to the same 2016 objective of hurting Clinton and boosting Trump. Over the longer time horizon, the operations were trying to divide American society. 

All of this data was uncovered and analyzed after the CIA assessment released in January 2017.

The effort to retcon interference into part of the “Russiagate hoax” has come for both the IRA and the GRU operations. Most grotesquely, right-wing influencers on X have attempted to revive the long-debunked Seth Rich conspiracy theory, claiming the GRU hack emails had in fact been leaked by Rich, who was murdered in retaliation. This is now something we can’t say, the argument goes, because Fox had to pay a large defamation settlement for advancing the theory years ago. Some observers are demanding the FBI reopen the case.

It’s worth stressing how carefully the underlying evidence of interference has always been framed. No report produced for SSCI, for example, claimed that interference swung the election. None made sweeping claims about impact or the persuasive power of Russian propaganda in changing votes; even when pressed, the authors very clearly stated that such questions were incredibly complex and that we hadn’t received the types of data necessary to address them. 

Nonetheless, the researchers who examined the data were targeted and labeled “Russiagate hoaxers” or “Deep State operatives” accused of trying to delegitimize Trump’s victory. Rather than engage with the substance of the findings, critics fixated on funding sources, political donations, past statements, prior work, credentials, daisy-chains of association—anything that could be used to discredit the people, rather than the evidence. In one of Taibbi’s many efforts to dismiss interference, he attempted to undermine Meta’s own calculation that 126 million people had seen the IRA content on its platform by claiming that my (Russiagate hoaxer) SSCI report was the source of their statistic—a take that was both backward and surreal given my adversarial relationship with the company at the time. But the smears don’t have to be coherent; they just have to be prolific. 

Maximalist contrarian narratives about “cooked intelligence” or Russia actually preferring Hillary might sell Substack subscriptions, but readers should be demanding to see the evidence. If Russia supposedly favored Clinton —where’s the primary source material supporting that? We know Moscow has long played both sides on divisive social issues—as it did during the period surrounding the 2016 election. It didn’t try to convert Black Lives Matter supporters or Etsy feminists into Trump voters; Russian operatives understood American political dynamics well enough to know that was a bridge too far. But we have evidence of how they engaged with Black and left-leaning voters—and it certainly wasn’t by producing pro-Clinton content. In fact, for years there was an abundance of publicly available data, including troves of posts released by Twitter from known state-run disinformation campaigns, and outside analysis from organizations like the Stanford Internet Observatory and DFRLab. (This is one reason those groups became targets themselves.). This material isn’t hard to find. And it isn’t hard to discern what Russian, Iranian, or Chinese operatives saw as strategically useful in any given propaganda operation, based on whom they amplified and what narratives they pushed. This is not 5D chess.

What might help settle the question—or at least keep the debate tethered to reality—is more sunlight on the underlying data. When Facebook, Twitter, and Google turned evidence over to the SSCI in 2018, only the 3,000 or so paid ads were released, with minimal redaction. That made sense from a privacy standpoint—many of the hundreds of thousands of organic posts contained stolen photos from real Americans—but it also skewed the public conversation. Headlines fixated on “$100,000 in Facebook ads,” a rounding error compared with normal campaign spending that missed the larger story: Hundreds of thousands of posts that built durable identity-based affinity groups on Facebook and Instagram got millions of engagements and then steered those communities toward cynicism or confrontation. (The average annual budget for troll factory operations targeting the U.S. was estimated to be closer to $10 million per year and appeared to be increasing, according to an indictment of one of the accountants.)

Even the self-styled supervillain behind much of the mess couldn’t resist bragging about it. Evgeny Prigozhin—the hot dog chef-turned-warlord who bankrolled the troll factory—had grown increasingly bold  in recent years, declaring: “We interfered, we are interfering, and we will interfere.” He’s no longer among the living—he fell not from a window but from 30,000 feet when his plane semi-expectedly exploded following his march on Moscow. But it’s hard to square his public gloating—plus the court indictments, platform takedowns, and bipartisan Senate findings—with today’s revisionist insistence that the whole thing was a phantom. If the chef-in-chief of the operation and the rank-and-file trolls involved are proudly confessing, and mountains of primary-source material is out there, pretending the operation was not what the evidence showed starts to feel less like skepticism and more like gaslighting. The only real question is what impact it actually had.

Gauging that remains a challenge. Scholars have tried—one paper got creative with Twitter data, assessing that the Twitter campaign appeared to have no meaningful impact on voting behavior. The trolls used the platforms quite differently; unlike the sustained messaging targeting the Facebook and Instagram affinity groups, the IRA’s Twitter network was opportunistic—chasing whatever hashtag was trending at the moment. YouTube, different still, served as a host site for longer propaganda videos—though, as the Graphika/OII report details, the company turned over only minimal data, making it hard to understand how resonant the content was. Propaganda doesn’t inject new beliefs into its targets like a hypodermic needle, but memes and repetitive messaging can and do have impact in shaping public opinion, and viral rumors can come to feel like reality.

As platform transparency programs have been dismantled and researcher access curtailed, recovering meaningful levels of visibility into influence operations will only get harder—which is precisely why demanding hard evidence about what a campaign did or said matters more than ever. X, once the platform that most publicly shared data and disclosed influence operations on its platform, no longer announces state actor influence operation takedowns. Researchers no longer collaborate with platforms in evaluating these campaigns. Government institutions tasked with mitigating foreign influence were reframed as “censorship” entities, and largely defunded or shuttered. And maximalist retellings—like the “cooked intelligence” pro-Clinton fiction—now thrive in the serialized cinematic universes sold by contrarian-media fabulists.

And yet the truth remains: Russia interfered in the 2016 election. It did so to help Trump and hurt Clinton. It relied on both trolls and hackers, memes and leaks, real-world events and digital manipulation. This conclusion was not remotely controversial at the time. And it isn’t controversial now—not even based on the actual findings of the Tradecraft Review of the 2016 ICA.

It’s therefore disappointing to see the CIA director and his deputy on X amplifying allegations to the contrary. Rewriting history doesn’t clarify anything. It only prepares the ground for the next operation.

The stakes are high: Preserving democratic resilience demands a steadfast commitment to transparent, rigorous intelligence analysis and to independent verification. Without these, we risk surrendering historical truth to politically expedient fiction.

– Renée DiResta is an Associate Research Professor at the McCourt School of Public Policy at Georgetown. She is a contributing editor at Lawfare.