Today Twitter did a great public service by allowing journalists to download and comb through datasets of troll tweets taken from Iran and Russia’s Internet Research Agency. This piece examines one of the smaller Russian datasets, and will be followed with more journalism on the media and the tweets sent by the trolls.
Patribotics is grateful for the support of our readers, which makes our reporting possible. Your support will help us to break more stories on Trump and Russia in 2018. Please consider a donation here
The independent news research organization, DFR Lab, part of the Atlantic Council – the initials stand for Digital Forensic Research Laboratory – were given a preview of this massive dataset, and published a medium piece and Twitter thread. Ben Nimmo, from the organization, proved it lives up to its name by listing the takeaways:
Troll operations were :about the home government first – had multiple goals – targeted specific activist communities – apolitical – opportunistic – evolved – not always high-impact.
First point: the troll operations started off as defending the home government. Timeline on 9 million Russian tweets here.
Russian-language posts outnumbered the English ones, and peaked higher and earlier. Russians were the first targets.
[The Russian ] troll accounts posted 19,000 times the day before
#MH17 was shot down, and 57,000 times the day after. Main hashtags: #КиевСбилБоинг (“Kiev shot down the Boeing”), #ПровокацияКиева (“Provocation by Kiev”), and #КиевСкажиПравду (“Kiev, tell the truth”).
The English-language [Russian] troll accounts defended Russia when it was needed, too. Case in point, the 781 posts they made on the Mueller “witch hunt.” Part authored, part retweets of sites like Infowars and Zerohedge.
Their highly recommended Medium article adds plenty more detail:
Of the [ten million] total, over nine million tweets were attributable to 3,800 accounts affiliated with the Internet Research Agency, also known as Russia’s infamous St. Petersburg troll factory.
The Russian operation had multiple and evolving goals. One main purpose was to interfere in the U.S. presidential election and prevent Hillary Clinton’s victory, but it was also aimed at dividing polarized online communities in the U.S., unifying support for Russia’s international interests, and breaking down trust in U.S. institutions.
The authors go on to cite several examples of Russian trolls on the ostensible left; a “black activist” claiming that Shakespeare’s plays were actually authored by a black woman, the hashtag #GunFreeZones, and so forth. More difficult, however, at least without data to sustain it, is the conclusion of DFR Lab that the trolls had “no effect” on behavior:
The Russian trolls often chose targets of opportunity, especially elections and terrorist attacks, in their attempts to interfere in local politics. This included promoting anti-Islam hashtags after the Brussels terror attacks, a pro-Leave hashtag on the day of Britain’s Brexit referendum, and leaks targeting French President Emmanuel Macron before his election. These opportunistic attacks had little to no discernible impact on the target populations’ political behavior, indicating the limitations of online troll operations….
7. Low Impact
Other than in the United States, the troll operations do not appear to have had significant influence on public debate. There is no evidence to suggest that they triggered large-scale changes in political behavior, purely on the basis of their social media posts…
The positive conclusion of this is that the trolls were less effective than may have been feared. Many achieved little or no impact, and their operations were washed away in the firehose of Twitter.
I am hopeful that in successive articles the authors will explain how they drew this conclusion. ‘Other than in the United States’ implies that the trolls were successful in influencing behavior in the United States. But based on what? If based on the fact that Trump won against the polling, then Brexit was a shock result to pollsters too – and the polls on the day of voting had called it for Remain by four points, when Leave actually won by four points. Betting odds in Britain gave Remain a 75% chance of victory. What criterion then shows that the trolls were unsuccessful in their ‘day of’ hash-tagging?
Perhaps the phrase “influence on public debate” measures how DFR Lab concluded that the trolls did not affect the Referendum. If the organization was scraping mainstream news stories that referenced the same topics pushed by the trolls, and found very little evidence that real news outlets paid attention, that might explain their conclusion that trolls “weren’t effective”.
And yet in point of fact we see that Russian trolls have been supremely effective. Their work is not about trying to seed the mainstream media, but about radicalizing, disaffecting, and disenfranchising. All of these things have an effect on voting behavior. Brexit is indeed a prime example. Almost at no point in the Brexit campaign did Leave sustain a real lead (there was one brief period before the murder of the MP Jo Cox). And yet Leave won, and won comfortably.
We know from Mueller’s indictment of GRU officers of Russia’s military intelligence services that ‘Guccifer 2’, long described by US media as an individual, was in fact a persona run by an entire Russian directorate inside GRU headquarters. These officers made sure that their hack of the DNC was distributed by Wikileaks to the widest possible audience. Did the Russian trolls spread these leaks and links? As the analysis shows, Russian trolls from the IRA worked a multi-year operation. In saying that they were ineffective, DFR Lab raises the interesting question of why the IRA would pay – for years – to run troll operations that “had no effect”. Clearly, they say, some trolls were getting better at the job – “the most effective Russian trolls used exactly the techniques which drive genuine online activism and engagement”. And yet they didn’t change voter behavior? This seems unlikely on the face of it, but we must wait for further articles to discover how and why DFR Lab draws that conclusion.
‘MAGA’ Trolls Comfortably Beat ‘Resistance’ Trolls
Moving on from DFR Lab’s analysis, a look at the data yields some interesting results. Here are a few bios I found in the smallest dataset, the hashtagged or concealed bios. There are plenty of Resistance, BlackLivesMatter and “Progressive” examples:
“California, USA”,”Not a lawyer #resist” “2018-04-03”
“A voice of resistance from a person of color living in America | #BlackToLive!”,””,”836″,”809″,”2013-07-26″,”en”
“Follow the example set by Mrs Obama; peace, love, acceptance & vigilance #Impeach45 #Resist #GunReformNow” 2017-09-05
“Yellow dog Democrat. GOP=New Russian Party #TheResistance #Indivisible trump=illegitimate Gorsuch=Illegitimate Follow me”,”https://t.co/4srF9ALSJp”,”0″,”0″,”2017-07-22
“JemiSHaaaZzz”,”United States”,”Teacher, Reader, Writer, Resister. #Resistance RESPECT LOVE LOYALTY”
Super lefty- liberal feminist. Only here to troll Trump & his moronic minions I follow back, unless you’re weird. #Resist #BLM #LikesToSwear ??Follow me”
“Resistance Girl ✊🏿”,”LaChristie”,”Flint, MI”,”Progressive. Activist. Warrior. Inspiration. #Resistance”,””,”9097″,”7173″,”2017-09-11
“Here to state my point of view, make you laugh and most importanly to stand up for and represent my people #BlackLivesMatter”
“‘Don’t Shoot’ aims to empower and encourage people to take action in struggle for social justice. Unite and fight against #PoliceBrutality!” – “2016-01-28”
,”everybody wants to be a nigga, but nobody wants to be a nigga. #BlackLivesAlwaysMatter”,””,”902″,”894″,”2013-07-25″
“Smart, strong, sensual woman #BlackLivesMatter activist”,””,”1051″,”620″,”2013-08-12″
However, it is worth noting that not all of these are equal in number. There are many more conservative bios than liberal ones in the hash tagged dataset, as will be noted below. “Resist” comes up 13 times, “Liberal” 24 times, “Progressive” 5 times, “Dem” or “Democrats” 17 times, “Hillary” 4 times (and one is “Never Hillary”), and “Bernie” once and, amazingly, “Clinton” not at all. By contrast, “Trump” hits 172 times, “Conservative” 126 times, “Republican” 12 times and “GOP” 12 times.
Russian Trolls Love Targeting the US Military
But the resistance hashtags paled in comparison to those targeting the US military. In the small dataset of hashed users, there were 6 uses of ‘veterans’, 48 uses of ‘vets’, 24 uses of ‘military’, 10 of ‘Army’ 8 of ‘Navy’ and 5 of ‘USAF’.
“Christian, husband and I are USAF #Veterans Mom&Gigi??YSU Alum M.S. Ed, Tarheel born, Buckeye fan, yogi/healthy lifestyle. I believe in #MAGA????#TRUMP”,”https://t.co/4srF9ALSJp”,”232″,”983″,”2017-08-08″
“Real Estate & Mortgage Broker, Mom of 3, love God God Bless President Trump #AmericaFirst #MAGA #2A #Military #Veterans Will Block Trump Haters Follow me”,”https://t.co/4srF9ALSJp”,”968″,”3871″,”2017-07-27″
“Retired IT geek from Veterans Admin, supports military and country! #MAGA #VETSFIRST, ? @POTUS, #1A, #2A #BanShariaLaw No Lists Follow me”,”https://t.co/4srF9ALSJp”,”202″,”947″,”2017-08-09″
“Registered Nurse, Entrepreneur, Wife, Christian, Pro-life, Trump 2020; pro-legal immigration, Vets & Trad. marriage Follow me”,”https://t.co/4srF9ALSJp”,”0″,”0″,”2017-07-25″,
“USA”,”Conservative wife, mother! Extremely proud of our veterans!! Lets take this once great country back!!!”,””,”2027″,”2694″,”2013-09-08″
“#USA????#Vets????#Police, and #Bama ?? ROLL TIDE! American and Southern by the Grace Of God, living in beautiful Idaho. Happily married to my HS… Follow me”,”https://t.co/4srF9ALSJp”,”0″,”0″,”2017-07-22
,”Equine Hoof Rehab/Soundness Mgmt Fairchild Cutting Horses & Cattle Co * God * Constitution * Conservative * 1A & 2A * Veterans * Freedom * John Wayne”,”https://t.co/4srF9AuhkP”,”904″,”52″,”2017-08-18″
“there are things known and things unknown and in between are the doors #twd #maga #beyou #veteranslivesmatter #conservative..”,”https://t.co/4srF9AuhkP”,”2208″,”1012″,”2017-08-09″
Guns, God and All-American Patriots (from Russia)
“Gun” showed up 22 times in this dataset, “#2A” 36, and “NRA” a hefty 39 times. “God” a whopping 106 times, “Patriot” 67 times, “Jesus” 16 times, “Christ” 70 times, (including instances of “Christian”) and #MAGA 67 times. Always as a negative, “Islam” showed up 21 times and “Sharia” 9 times. “Black” came up almost as often as “God” (and both significantly more than “MAGA”) at 100 instances. The word “lives”, used in hashtags “#BlackLivesMatter #VetsLivesMatter and #bluelivesmatter came up 50 times, and #BLM 11 times.
Trolls Love Poison, But Hate Policy
Compare these numbers to policy positions and they show that the trolls by far preferred to use emotions over facts. “Wall”, Trump’s most famous election pledge, only had 12 instances. “Border” had 2. “Immigration” did not come up once. “Immigrant” 3 times, “tax” 4 times, “abortion” just three times, “trade” once, and not in an economic context (the bio reads ‘”I’d love to trade weekdays for weekends. ⏰ ⏳ ⌛”). “Economy/ Economic” have five uses between them. “Defense” is never used.
The Americans – USSR to “USA”
These Russians really, really want you to think they are American. 152 instances of that word make it almost as big a winner as ‘Trump’. “United States” gets another 126 uses, and “Estados Unidos” a huge 227 uses. But if there is one overwhelming element of a Russian troll bio in this, the smallest dataset, it is the acronym “USA”. That gets 792 instances of use in the dataset, dwarfing everything else. That’s partly because “USA” is a location setting as well as a bio term, and if there is one thing the Russian troll “Blacktivist” and the Russian troll “Proud Deplorable” both want you to know about themselves, it’s that they are located in the USA.
Except they’re not. And now thanks to Twitter’s transparency, real American journalists will be able to analyze their propaganda. Patribotics will cover the larger datasets, including the media released by Twitter, in later articles.