Twitter has been told to “look again” for evidence of Russian interference in UK politics as new research appears to show the social network under-reported the problem in evidence submitted to MPs.
Data unearthed by the Press Association shows more than 2,400 tweets about the UK, Brexit, the refugee crisis and last year’s general election came from at least 154 accounts linked to the Internet Research Agency (IRA), a Kremlin-linked organisation accused of sowing disinformation and discord on social media.
Here are the key questions:
What is the Internet Research Agency?
The Internet Research Agency (IRA) was a Russian organisation accused of spreading divisive messages across the internet, supporting the Russian government and attempting to weaken its rivals.
A US grand jury charged the IRA, among others, with conspiracy to defraud the United States in February. It named a senior Vladimir Putin ally, Yevgeny Prigozhin, as the organisation’s funder through his web of companies which regularly receive large government contracts.
Mr Prigozhin and Mr Putin have repeatedly denied any involvement.
Questioned this weekend by NBC News about allegations that Russian citizens had tried to interfere with US politics, Mr Putin said: “There are 146 million Russians. So what? I don’t care. I couldn’t care less. They do not represent the interests of the Russian state.”
What kind of tweets were sent?
The accounts regularly painted the refugee crisis as a Muslim invasion, highlighting terror attacks or crimes committed by immigrants across Europe and often sharing claims with little evidence to back them up.
All except one of the ten most-shared tweets were Islamophobic or anti-immigration in tone. The tenth was a Nigel Farage quote about Donald Trump.
Almost 500 tweets were sent on the day of the Brussels terror attack in March 2016, when the hashtags #IslamKills and #StopIslam featured heavily in the dataset, and #ReasonsToLeaveEU was popular among the accounts on the day of the Brexit vote.
Elsewhere, the infamous @ten_gop, which had more than 100,000 followers at its peak, said “It’s horrific! London looks like Baghdad!” on three occasions, garnering almost 1,000 retweets as a result.
The tweets were not all malicious or partisan in nature, however neither were they all original. The accounts run from St Petersburg would often retweet news organisations, politicians or prominent figures from across the political spectrum discussing the issues of the day.
What impact did the tweets have?
Only 278 of the 2,423 tweets about UK politics contained information about whether they were retweeted or favourited by other users, nearly half of which had no retweets or favourites at all.
The 142 tweets which did contain such information were shared more than 35,000 times and favourited almost 23,000 times.
Information about how many people see tweets or interact with them in other ways is only available to the accounts which post tweets or to Twitter themselves.
Twitter allows more access to data about its platform than other major social networks, but academics and researchers have called for collaboration between social networks, governments and universities to preserve data of historical significance.
How has Twitter reacted?
Twitter previously told MPs that 49 IRA-linked accounts had sent fewer than a thousand tweets about Brexit after months of pressure to reveal any posts linked to the United Kingdom.
In the wake of ongoing allegations about misuse of the platform, Twitter chief executive Jack Dorsey unveiled a new initiative aiming to improve the “health” of conversation on the network at the start of March.
“We don’t claim to have all the answers,” said a spokesman. “That’s why our approach to these critical issues at Twitter has been one of humility and collaboration from the off.”
The company is currently accepting proposals on how to achieve this, promising funding for the best ideas.
“Our primary aim is to ensure actors with malicious intent find it more and more difficult to abuse and undermine the integrity of our platform,” the spokesman added.
How has the Government reacted?
Matt Hancock, secretary of state for digital, culture, media and sport, will answer questions from the parliamentary inquiry on fake news on the Government’s response to online disinformation campaigns from foreign actors.
A Government spokesman said: “We have been clear on our concern about the rise of ‘fake news’ and misinformation, and its impact on the way that the public engages with politics and political information.
“To date, we have not seen evidence of successful interference in UK democratic processes by a foreign government.”
How was the data gathered?
The 154 accounts and 2,426 tweets were identified by combining datasets of millions of tweets from US and UK researchers. They were then filtered to include only tweets from known Russian accounts which discuss topics relevant to UK politics, including the EU referendum, the 2017 general election, UK political parties, politicians, immigration, the refugee crisis and the NHS.
The majority of tweets in this investigation came from a database of more than 200,000 tweets released by NBC News.
Researchers working with the Press Association scanned both databases for any mention of 117 keywords, hashtags and phrases relevant to UK politics, including mentions of politicians, political parties, countries, and social issues.
The Press Association then manually removed any tweets which did not specifically address UK politics or issues related to Brexit, UK sovereignty and immigration.
Anti-European and anti-immigration tweets were included in the final dataset because immigration regularly featured at the top of polls for reasons given by voters wanting to leave the European Union.
Ben Popken, a senior staff writer at NBC News who worked with US researchers to release the NBC dataset, said the full database constituted a spectrum of different activity, from political tweets to jokes around popular hashtags.
Neither the NBC database nor that gathered by UK researchers gives a full picture of all the tweets sent by the accounts in question, however.
“It’s a range,” Popken told the Press Association. “Our sources recovered over 200,000 tweets from 454 of the 2,752 named accounts. So the database could be the tip of the iceberg.”
All of the accounts and tweets identified in the data have been publicly deleted by Twitter.
To comply with Twitter’s terms of service, researchers who save copies of tweets are obliged to delete any tweets from accounts which Twitter suspends.
The researchers have asked to remain anonymous as a result.