We knew there was a problem with bots and trolls affecting Social Media around the Brexit vote.
We now know some were Kremlin farm staff, like David Jones, and we know academics have uncovered about 13,000 of them and they disappeared after posting tens of thousands of messages.
We have all seen them being repurposed too, flitting between Trump and Brexit often in a display of geographical identity confusion.
We also know what purpose they fulfilled and that there’s a whole underground market, and they’ve even been deployed again – such as in Qatar.
We even know Leave.EU used them, because they told us, though they’ve since decided they didn’t.
But can we prove it?
Basically, yes. Because of pencils.
Do you remember stuff like this?
At the time it was odd, utterly dismissable in fact, but the campaign was so well executed it ended up as headlines in every major newspaper and found itself factchecked on the BBC, who said: “some people on social media, seemingly mostly Brexit supporters, suspect that the vote will be anything but fair. They believe that the establishment is not above fixing the vote to thwart the democratic will of the electorate.”
“The run up to the referendum has seen the rise of the hashtag #usepens which urges people to reject the traditional pencils supplied at polling stations and instead use a pen to mark their cross on the voting paper. The thinking behind this is that it will then be impossible for some unknown hand to use an eraser to rub out your cross and make another mark in the other box,” they added.
Thankfully, the principle of every contact leaves a trace applies just as much to social media as it does in the forensic investigation of a crime.
The starting point was an analysis of the three hashtags used in this disinformation campaign, #TakeAPen, #UseAPen, and #UsePens.
Using the date range 01/06/16 to 30/06/16, the results of a trend analysis with Truthy are astonishing and show a tightly focused bot campaign which was targeted at a very specific point in time: the days up to the vote.
The campaign was then terminated immediately afterwards.
Even though the bot accounts were redeployed and often renamed, social
media keeps enough data to allow a hoax to remain visibile when it was
spread and give a good indication of scale.
Even with 65,000 bot tweets having been deleted, we can clearly identify a focused pattern in what remains on the internet, showing thousands of messages pushing the disinformation in a very short time period.
When you then start to run a network analysis, to see who precisely was talking about pens, things start to become a little more obvious. The conspiracy theory was most predominently discussed and proliferated at the UKIP end of the Leave spectrum.
Helpfully, we can also see which of the hashtags was the most pervasive, which real people used the tags a lot, and how this helped spread the message by pushing it into the discussions of mainstream reporters.
(Please note: a name appearing in trace evidence does not mean that person is necessarily a bot, just that they interacted with a bot-led activity which was exploiting Network Centrality.)
So some people were talking about it right, but how do we know bots were involved?
Well, by shifting our analysis away from mentions and retweets and looking at what other topics were being pushed in tandem – what’s called co-concurrence – we see this mysterious conspiracy theory was being pushed by people who were also pushing the tags #Ukraine, #Syria, #Aleppo, #ISIS, and #VoteLeave.
This allows us to see what the bot elements on social media were driving as core messaging in addition to the pen conspiracy during the same time period.
It gets even more interesting when the co-concurrent trends are set against each other over the same time period.
The analysis shows how the bots were specifically redeployed from regular Russian focus areas Ukraine, Syria, ISIS, and Aleppo to not only #UsePens but #VoteLeave too, with a huge push around referendum day, dwarfing all other bot traffic and after which the majority simply disappeared. (Spoiler: Trump waved them in afterwards).
So yes, bots were used to pushed the Leave message around the Brexit referendum.
Yes, the scale of the bot deployment was huge and did influence mainstream public discourse.
Yes, it can be connected to Russian disinformation too.
And yes, if they’d stuck to pencils we’d be none the wiser.
The key question has been put to Andy Wigmore and Arron Banks of Leave.EU, but at this time they haven’t replied.