- Researchers have uncovered networks of Russian Twitter accounts which posted messages in the run-up to the UK’s EU referendum vote.
- Academics at Swansea University and Berkeley uncovered 150,000 accounts and 45,000 messages around Brexit.
- One of the most prolific Russia-linked accounts spread messages of Islamophobia, xenophobia, and negative messages around Sharia law.
- Analysis of the research showed humans were susceptible to the bots, and retweeted messages created by Russia-linked accounts.
- Prime minister Theresa May accused Russia of trying to “sow discord” through misinformation.
If you follow @sveta1972q1 on Twitter, you’ll see lots of seemingly harmless tweets about cloud computing dating from August 2016.
Had you followed the account in June 2016, though, you would have seen hundreds of tweets spreading messages of Islamophobia and xenophobia in the run-up to the Brexit vote
One such tweet — now apparently deleted — read: “Pathetic big freaking deal you’ll be singing another tune as sharia creeps in and takes hold #voteleave #brexit.”
Here’s another: “uk friends consider the quran on jihad and migration #voteleave #brexit.”
@sveta1972q1 was identified by researchers at Swansea University and Berkeley, California as being a possible Russian bot operating on Twitter during Brexit and the US presidential election.
Their wider research found 150,000 Russian-linked accounts and 45,000 messages sent on Twitter during Brexit.
The researchers collated data by collecting 28.6 million tweets tagged with the #Brexit hashtag between 24 May 2016 and 17 August 2016. The idea was to examine social media sentiment around major political events, so not all were bots. But they identified thousands of Russian bot accounts through the fact the accounts set their language description to Russian, not English. And their habit of posting hundreds of tweets in a day suggested they were bots.
Tho Pham, PhD candidate at Swansea University, identified @sveta1972q1 as “one of the most active accounts” in an email to Business Insider. The account remains active, but is labelled by Twitter as “temporarily restricted” due to unusual activity.
Here’s another sample tweet, provided by Pham: “these are not peaceful refugees they are violent invaders #calais.” The tweet was sent on 22 June 2016 — one day before the EU referendum vote. There had also been a story from The Daily Express that refugees in the Calais “Jungle” camp had attempted to jump onto a lorry and cross the border from France into the UK.
And one tweet where the account accused US-based British comedian John Oliver of propaganda read: “saw john oliver propaganda video can say he is been paid handsomely for all the crap he vomited in the video.”
Business Insider saw only a sample of @sveta1972q1’s tweets between June 20 and June 24 — the immediate run-up and aftermath to Brexit. The account sent 99 tweets in that time. This effectively amounts to a tweet on the hour, every hour over four days — not impossible, but highly improbable taking into account factors like sleep. Business Insider asked @sveta1972q1 whether they were a real person, but whoever is behind the account did not respond.
There are other clues that the account is not genuine. During those four days of activity, @sveta1972q1 variously claimed to be “a brown non-Brit,” “a Dutchman” and “of Scottish heritage.”
What is harder to determine is what impact the tweets had in terms of specific numbers. But the researchers generally found that human accounts were susceptible to retweeting messages created by bots. On average, an original bot tweet got five retweets from humans.
Here’s a graph showing just that:
According to the researchers, most of the Russian-linked accounts appeared before the Brexit vote.
“For the Referendum, the massive number of Russian-related tweets were only created few days before the voting day, reached its peak during the voting and result days then dropped immediately afterwards,” they wrote.
This diagram shows how Russian accounts spiked in activity around Brexit — the referendum date was June 23, 2016:
Pham told Business Insider that many of the accounts had now been suspended though some, like @sveta1972q1, appear to have slipped through the net and are now tweeting about other topics.
She added that she had actually expected more bots around the time of the vote.
“What I expected at the beginning was, let’s say, [there are] 100 users, that at least 40 should be bots,” she said. “But the number of bots [were more like] just 15 and 20, but their influence was significant.”
Pham said she had gathered data to replicate the research for the UK’s 2017 general election. Asked whether she would likely carry out a similar study, she said: “I would say yes.”
What is particularly alarming, she added, was that bots were upending the idea that the internet could make people better informed. “What we actually analyse and observe is that bots know how to pull the strings, they know how to manipulate people,” she said. “That was very interesting and shocking for us. And if a person knows exactly how to use bots they could — I couldn’t say easily — possibly use them to influence people.”
Sources close to Twitter pointed out the firm has done considerable work in filtering out “low-quality” accounts and tweets. Just because the tweets are on the platform doesn’t mean the bulk of users actually see them.
The company wouldn’t comment on individual accounts, but a spokesman said: “Twitter recognises that the integrity of the election process itself is integral to the health of a democracy. As such, we will continue to support formal investigations by government authorities into election interference as required.”