CHISINAU -- A fraudulent Facebook campaign in Moldova used the tactics of Russia’s notorious “troll farm” -- the Internet Research Agency -- to promote a political party that, ironically, is at odds with the Kremlin.
Victor Spinu, co-founder of the Chisinau-based social-media monitor that helped Facebook investigate the network, told RFE/RL that the Moldovan ruse promoted Prime Minister Pavel Filip’s ruling Democratic Party (PDM) and was “very close to the activity of Russian troll farms.”
Filip and his Democratic Party are at odds with the Kremlin over such issues as Chisinau’s aspirations to join the European Union.
Spinu said his civic group, Trolless, found that those who ran Moldova’s fake Facebook operation “learned lessons” and copied tricks used by the Russian trolls who allegedly tried to influence the 2016 U.S. presidential election.
Spinu said social-media monitors in other former Soviet republics at odds with the Kremlin have expressed concerns to him about similar “copycat” tactics in their countries.
“Monitors in Ukraine and Georgia have told us that they also are seeing the same kind of troll activity in their countries – the same kind of coordinated tactics we’ve seen used by the Russian trolls” and copied in Moldova, Spinu said.
'False Amplification'
Like previous Russian troll campaigns, Spinu said, the Moldovan trolls set up fake Facebook accounts to pose as legitimate voters and civic groups.
He said the network generated and distributed fake news, disinformation, and memes ahead of Moldova’s February 24 parliamentary elections.
He said those fake accounts also worked together with legitimate accounts in Moldova to flood web forums with posts aimed at manipulating online public debate.
Facebook has said that’s a Russian troll tactic called “false amplification.”
RFE/RL has documented how the same tactics were used by the Internet Research Agency, the St. Petersburg-based troll farm thought to be financed by President Vladimir Putin ally Yevgeny Prigozhin.
An April 2017 case study co-authored by Facebook security officer Alex Stamos also described similar tricks being used by Russian trolls to try to influence the 2016 U.S. election.
That study defined “false amplification” as a “coordinated activity by inauthentic accounts with the intent of manipulating political discussion,” either by discouraging some groups from joining an online debate or by “amplifying sensationalistic voices over others.”
“From there, organic proliferation of the messaging and data through authentic peer groups and networks was inevitable,” the Facebook study said.
Moldovan Copycats, 'Coordinated' Campaign
On February 13, just 11 days before Moldova’s parliamentary elections, Facebook shut down 200 Moldovan accounts and pages for what it called “coordinated inauthentic behavior.”
Facebook also indicated, along with Trolless and other sources, that the online campaign favored Filip’s Democratic Party.
Although Filip in 2018 publicly stepped back from his staunchly pro-EU rhetoric, the Kremlin continues to favor Filip’s pro-Russia rivals – President Igor Dodon and the Socialist Party of Moldova that Dodon formerly headed.
Facebook's cybersecurity policy chief, Nathaniel Gleicher, said the fraudulent online behavior “originated” in Moldova and was carried out by people who “attempted to conceal their identities” but were known to include “employees of Moldova’s government.”
Facebook also released samples of what the Moldovan troll network posted and shared online. All of those examples either favored Filip’s Democratic Party or cast an unfavorable light on opposition groups and candidates.
Moldova’s troll network "used a combination of fake accounts and some authentic accounts to mislead others about who they were and what they were doing," Gleicher said. They "typically posted about local news and political issues.… They also shared manipulated photos, divisive narratives, and satire."
Altogether, Facebook removed 168 Facebook accounts, 28 pages, and eight Instagram accounts in Moldova that were suspected of spreading fake news, political propaganda, and misinformation ahead of the elections, he said.
In one case, Gleicher said Facebook removed a fake page for impersonating a legitimate Moldovan fact-checking organization that calls out others for “spreading fake news."
Two samples of troll posts released by Facebook confirm the impersonated fact-checking group was StopFals – a Chisinau-based misinformation watchdog led by journalist Lilia Zaharia from the Association of Independent Press of Moldova (API).
Zaharia told RFE/RL that all of the material published by the fake StopFals Facebook page either favored Filip’s ruling PDM or denigrated its opponents.
“They stole our articles and took our images and they shared this on social media,” Zaharia said. “On a Facebook page that mimicked our work, they shared only some of our material, mixing it together with their own fake stories and manipulated photos.”
A Democratic Party official also confirmed that his authentic party page – Good Step -- was among those removed in Facebook’s crackdown against “coordinated” troll activity.
Ion Harghel, vice president of the Democratic Party’s youth branch, denies being part of an online troll network.
He accused Facebook of helping opposition parties by removing his Good Step page less than two weeks before the parliamentary elections.
Spinu says data collected since 2016 by Trolless also confirms that only pro-Democratic Party accounts were involved in the kind of “coordinated” troll activity described by Facebook.
“We reported 700 fake accounts to Facebook in early February,” Spinu told RFE/RL. “Some of them were inactive since they were set up as far back as 2016, but there was a surge of activity with others as the parliamentary elections approached.”
“There were many fake accounts that favored the Democratic Party,” Spinu said. “There also was a smaller number of fake accounts that favored opposition parties,” including Dodon and the pro-Russia Socialist Party.
“We didn’t report any authentic accounts to Facebook because people have the right to exercise their freedom of speech,” Spinu said, noting that Harghel’s Good Step page was not among the 700 accounts and pages that Trolless reported to Facebook.
“We did monitor the posts of hyperactive trolls involving both fake and authentic accounts,” Spinu said, adding that the trolls were very active on the Good Step page.
“The coordinated activity we saw only favored the ruling Democratic Party,” Spinu said. “We didn’t see any coordinated activity that supported opposition parties. So we can conclude from our observations that the 200 pages deleted by Facebook were only accounts favoring the Democratic Party.”
Government Denial
Filip’s administration distanced itself from Facebook’s charge that “employees of Moldova’s government” were part of the troll network.
A government statement issued on February 14 suggested that any fraudulent social-media activity by government employees on behalf of the Democratic Party was carried out by rogue workers.
The statement declared there are more than 200,000 government employees on the state payroll and that Moldova’s government does not check social-media activity on the private accounts of its employees.
“They have different political options and opinions,” the statement said. “The state is bound to keep separate the issue of fighting fake news and guaranteeing freedom of expression to its citizens.”
In the aftermath of the parliamentary elections, monitors from the Organization for Security and Cooperation in Europe (OSCE) concluded that the vote was “competitive” and that “fundamental rights were respected.”
But the international monitors concluded that the campaign was marred by “strong indications of vote buying” and the misuse of state resources.
A preliminary ballot count shows the Socialists captured the most votes to win 35 of the 101 seats in parliament.
Those preliminary results showed Filip’s Democratic Party taking 30 seats, the pro-EU opposition ACUM group taking 26 seats, and the conservative Shor party winning seven seats.
The remaining three seats were expected to be taken by independent candidates.
