,
32 tweets,
16 min read
Read on Twitter
[Thread] I analysed around 13,000 retweets of Boris Johnson's tweet that read "The referendum result must be respected. We will leave the EU on 31st October. #LeaveOct31". He tweeted this twice, once on 15th Aug and again on 27th. The following is what I found: #StopTheCoup
Firstly, the first time he retweeted it it got Retweeted around 11.8k times: The second time, it received around 8.3k (see pic). So 13,000 retweets is not bad considering the total of around 20,000 tweets. At first, I was interested in automated accounts retweeting it (bots) 
That is to say, of 1025 accounts (8%) of the sample had biographies/descriptions that had the term MAGA in. That's not all though. I copied the text from the biographies and ran it through AntConc to count the frequencies of terms. Lo and behold, what did I find...
Excluding conjunctions, articles, pronouns etc (e.g the word 'the', the most common word among all the 12,177 bios was MAGA, shortly followed by Trump. You can see from the image MAGA was mentioned 1004 times (1025 when formatting issues accounted for). Trump was mentioned
853 times. This outranks #Brexit, which was mentioned about 508 times. So remember, this is an analysis of a British politican's retweets. One may reasonably assume that if you analysed those people who retweeted it, their bios would probably be more likely to include Brexit
than #MAGA or Trump. However, this is not the case. The implications are interesting. There are numerous possible explanations. If we assume that all the accounts retweeting Johnson's tweets are authentic (e.g. legitimate accounts), then the most cohesive community among them
are those that identify with supporting Donald Trump. By cohesive I mean a community that chooses to identify their Twitter account with the US President. Culturalists may suggest British people are less likely to wear their politics on their sleeve, an would be less likely to
have #Brexit plastered in their bios. However I don't think much of that. One might also think such numbers are logical as America has a larger population and global populism creates a commonality of interests among the Western right. However I am also not convinced by that
If we make another assumption, that many of the accounts are inorganic trolls (i.e. paid to promote Trump and his policies), then it probably makes more sense. A large proportion of the pro-Trump community also came into being in January 2017. As you can see from the graph, this
is quite significantly above the average. What's more, in a separate analysis of around 74000 Twitter unique accounts containing either the phrases #Maga #Qanon #LockHerUp and #KAG (i.e. Trump fans), we see the same spike at January 2017. This time we can see that 2,387 pro Trump
accounts were created in January 2017. That really is a lot considering the average per month is about 485. Although it is normal to see fluctuations in account creation date, this is way above even standard deviations. So what, you may say that January 2017 was Trump's
inauguration, and it makes sense that people would join then to show their support. Fair enough, but why didn't we see a similarly large spike when he won the election in December 2016. Also, why is there a similarly large spike in January 2018? Is it simply because it is the
anniversary of the inauguration. I doubt it, we don't see the same in January 2019. It could of course, be a coincidence, or indeed explained by some sort of call to Twitter in a right-wing populist memo in January 2017/18 that I missed. One hypothesis might be that,
if it is inorganic behaviour - a paid troll farm, that it was a yearly contract to be renewed in January. That might explain a boost in numbers at the same time each year. Certainly the 70,000 accounts or so show what, at first glance, an astonishing lack of diversity
in terms of biographical detail. They are also remarkably on message (won't go into that now until maybe some NLP analysis can be conducted). Anyway, we can conclude that of Johnson's retweets, a large number of those promoting the message are accounts that clearly align
themselves with Trump. In a related analysis, I also conducted the same analysis on the #LeaveOct31. This returned around 14,000 individual accounts and 20,000 tweets. Again, MAGA and Trump are the most popular descriptions within the biographies of accounts tweeting
on this hashtag. Of course many of those tweets will include those in the first analysis as Boris used the #LeaveOct31 hashtag. Nonetheless, it would seem the strongest community on that hashtag are the MAGA crew.
A somewhat formative analysis, but based on my ongoing analysis of Twitter propaganda, I would say organised entities are manipulating Twitter (including @BorisJohnson TL) in order to promote a hard-brexit. This #MAGA community's other tropes are familiar and consistent
@BorisJohnson - they are anti-Corbyn, anti-Islam, often self described 'patriots', or veterans, pro-Israel and anti-fascists. I don't doubt a huge number are real, but I am genuinely surprised by how many can consistently be so on message. Next tweet I will link to the data. #StopTheCoup
@BorisJohnson I should add before I link to the data that I downloaded @Nigel_Farage 's recent tweets and analysed the Retweets. On this occasion #Brexit was more frequent in the bios of the re tweeters, but MAGA and Trump came in 2nd and 3rd respectively
@BorisJohnson @Nigel_Farage Link to the Johnson Retweet Data: drive.google.com/open?id=1CrYPq… Link to the #LeaveOct1 data: drive.google.com/open?id=1MwOwB…
@BorisJohnson @Nigel_Farage For some extra information. The pro #MAGA accounts created in January 2017 - in addition to being suspiciously high in number, are also the most vociferous - i.e. they tweet the most. In this graph the red patches indicate groups of accounts that have more statuses. This graph
@BorisJohnson @Nigel_Farage therefore tells us that in addition to an anomalous number of MAGA accounts being created in January 2017, those accounts are more likely than most to be particularly active in promoting their MAGA message. That also makes the spike more suspicious, as if those accounts
@BorisJohnson @Nigel_Farage were created in a large volume to be particularly vocal....
@BorisJohnson @Nigel_Farage And as a bonus, for those interested in Islamophobia and what those 70,000 odd MAGA accounts think about Islam. 295 mention 'Islam' in their bios. Unsurprisingly, they all have remarkably similar views on their disdain for Islam...
@BorisJohnson @Nigel_Farage Ok that's enough for now. This is large scale analysis so I should re-emphasise a number of the accounts will be legitimate, but I'd wager a significant amount are absolutely not. We know enough now to assume that this type of behaviour is par for the course. It is sad
@BorisJohnson @Nigel_Farage so little is being done about it. Anyway, be vigilant, and angry, and above all, #StopTheCoup
@BorisJohnson @Nigel_Farage Just as an addendum. Manipulating a head of state's tweets is not unusual anymore. Back in 2017 I analysed how a bot army retweeted Donald Trump's tweet praising King Salman of Saudi Arabia. Even the President of the United States TL is fair game...
For those interested in more analysis of the weaponization of social media, this time from (probably) Saudi, you can read the following: The Gulf Info War| Propaganda, Fake News, and Fake Trends: The Weaponization of Twitter Bots in the Gulf Crisis ijoc.org/index.php/ijoc…
While compulsory verification would solve a lot of problems (and create others) I did an investigation that revealed how even verified accounts are hacked to spread propaganda. Even the accounts of dead people... Twitter did nothing alaraby.co.uk/english/indept…
Addendum: This is part two of the thread where I did more analysis. It makes essential reading
