If you’re following any or all of the 27 Twitter hashtags (listed below) related to the European Union Elections on May 23 for a look at how social media is as being used as the polls prepare to open, you won’t see massive flow of tweets.
“But that doesn’t mean the activity is all legitimate,” said Andy Patel, Researcher, Artificial Intelligence Center of Excellence, F-Secure.
Andy is tracking one tweet per second related to the May 23rd elections compared to around 40 tweets per second for a popular Twitter term like ‘Trump”.
“If you’re running a disinformation campaign, you don’t want to stick out and look too obvious. So you’ll try to mirror organic activity levels in the targeted space. Even low volumes of activity generate hundreds of thousands of tweets over time. Combing through that much data is time consuming, and it’s easy to miss things,” Andy said.
“If you’re wondering if a piece of content is being unnaturally amplified, look for easy to spot signs such as a retweet count that is similar to, or greater than the number of likes, or for likes or retweets from accounts that don’t look like they should have engaged with the tweet in the first place — such as a Korean user retweeting content about Brexit.”
Coordination and manipulation in the Twitter Brexit debate
For two years, Andy has been monitoring tweets related to elections and political debates in order to understand how social media can be abused to spread misinformation.
Recently his research tracking 24 million tweets from over 1.65 million users related to the Brexit conversation made international news by showing that there was a “global effort amongst the far-right to amplify the ‘Leave’ side of the debate.”
That effort first became observable in Andy’s data with a flare up in some accounts focused on the Brexit debate suddenly turning their attention to the #franceprotests hashtag on December 11th of last year. By mapping interactions between the most active accounts, he found suspicious behavior on both sides of the Brexit debate but was far more prominent on the “leave” side.
The EU elections are not explicitly about Brexit. However, the results, especially in the United Kingdom, will likely have some impact on the never-ending debate over the vote to leave the EU. This may be why Andy is still seeing a strong pro-Brexit amplification campaign continuing over the past few week.
Social media is the “wild west” of electioneering
There are lots of reasons why journalists and other observers will turn to Twitter to try to understand the EU elections, beyond the vacuum of solid information that will exist until votes begin to be counted.
Social media disinformation seemed to play a starring role in the elections of 2016 in both the UK and the United States. And while election advertising on older media channels faces clear regulations, social media remains a “wild west.” Large-scale efforts (or efforts that appear large-scale) can be summoned with low costs and even less accountability.
As the social media tealeaves are being read, here are five things Andy suggested that we keep in mind.
1. Facebook is probably where the action will be.
“To be honest, the real campaigning will be happening on Facebook,” Andy said. Hence, most of the disinformation campaigns that have been identified recently are targeted at the world’s largest social network.
This makes sense for several reasons, including the basic fact that Facebook has about seven times more active users than Twitter globally.
Facebook is facing ever-increasing scrutiny following the 2016 elections and related and unrelated privacy issues that seem to make headlines every week. In response, new features that add transparency to advertisements are now in place. This should allow users to have a much clearer idea about who is behind the ads they are seeing. The site has also instituted new fact checking features aimed at stopping false stories from going viral.
2. The same things that worked before are still happening.
In 2016, activists and governments who embraced social media for so-called “memetic warfare”—which involves using strategic trolling to “expose and harass” political opponents—and “culture jamming”—which seeks to “disrupt dominant frequencies”—had the element of surprise on their side.
Now, many of the tactics that seemed to have work have been identified. However, some of the people who seem to be paying the most attention to these tactics are right-wing activists who seek to emulate them.
Erin Gallagher noted that Germany’s right-wing AfD party tracked the “#MemeWar” of 2016, possibly to port the same tactics for use in their own use. In the 2018 parliamentary elections, the party saw its best election results in its history.
3. Be careful what you call a bot.
“I never used the word ‘bot’,” Andy said, referencing his Brexit research.
He feels the tendency to call any sort of misuse of Twitter a “bot” is overblown. Automation that allows users to tweet incessantly on multiple accounts exists and is definitely in use. But the sense that there are “bots” or many large-scale networks of “bots” ignores that Twitter is pretty good at cracking down on that sort of misuse of the site.
“Most of the activity that I see is from activists who seem heavily invested in politics,” Andy said.
These users use multiple tactics to identify themselves to fellow travelers and use hashtags and “follow chains” to build up their accounts.
However, they often seem to cross the line into misuse of Twitter based on the way they are often suspended and banned. Andy noted conversation about how to avoid being banned is common with users often sharing tips. And when banned users return to the site, their community often helps them “level up” by reclaiming as many followers as possible.
These accounts tend to share similar characteristics, including a fondness for Donald Trump often expressed in the account masthead photo. They also have a tendency to tweet often with messages filled with usernames and/or hashtags intended to connect with other influencers:
True “bots” would likely quickly become obvious to Twitter.
The site seems to have capabilities on cracking down on users that go beyond what it often deploys. For instance, a Twitter employee recently told Motherboard that the sites machine learning that helps it instantly ban ISIS members isn’t being used to keep white supremacists. Why? The site fears this would lead to banning some Republican politicians from the United States.
4. It’s not just Russians.
The New York Times recently pointed out that it isn’t just Russia who spreads misinformation on social media. The story noted that analysis from Advance Democracy has found “a constellation of websites and social media accounts linked to Russia or far-right groups is spreading disinformation.”
But the urge to label all social media interference “Russian” ignores the larger story.
“The IRA was just jumping on what was already going on,” Andy said. “And they probably still operate in the same way. It is very difficult to separate nation-state disinformation activities from other groups on social media, such as the far-right. They just amplify each other.”
We don’t know what activity is coming from Russia because Twitter doesn’t even know.
Andy points to tools like TweetsAttacksPro that allow users to “run thousands of twitter accounts at the same time 24/7.” The tool includes features that “prevent Twitter from becoming suspicious about the account,” including the ability to hide or change your location.
5. Twitter is not reality.
Twitter’s ability to shape the conversation is remarkable considering less than 5% of the population uses the platform.
A 2015 study found that journalists are the “largest, most active verified group on Twitter.” The fact that the world’s most powerful human is also a Twitter power user who has been recently sending out dozens of tweets and retweets in a day also draws tons of attention to the platform.
But that doesn’t mean Twitter is real.
“Look at the recent local elections in the UK,” Andy said.
While the right wing dominated the Twitter debate, Liberal Democrats found surprising success at the polls.
“Twitter seems to be become increasingly become a bubble filled with far-right accounts,” Andy said.
So remember when you’re focusing at Twitter, you’re already through the looking glass.
For more insights, check out, Lisa-Maria Neudhert’s takeaways on “junk news” being shared related to the EU elections and Luca Hammer’s continuing Twitter thread tracking EU election-related social media activity.
EU Election-related hashtags: