Skip to main content

A new wave of disinformation emerges with anti-Trudeau hashtag

#29 of 84 articles from the Special Report: Democracy and Integrity Reporting Project
Prime Minister Justin Trudeau at an event with the European Council president and EU trade commissioner in Montreal on July 17, 2019. Photo by Josie Desmarais

The hashtag #TrudeauMustGo went viral last week, driven in part by inauthentic activity including artificial amplification and automation. After National Observer reported on the phenomenon, activity surrounding the hashtag surged for a second time, bringing with it a new wave of disinformation and social media manipulation.

That second wave of activity, beginning on July 18, also swept in a new cluster of hashtags, including #NotABot, which was used in an apparent attempt to deny or debunk allegations that #TrudeauMustGo was being amplified by inauthentic activity.

The hashtag #NotABot — and the associated narrative — could be described as a form of disinformation in itself, said Geoff Golberg, a social media manipulation researcher and founder of social media mapping firm Social Cartograph.

Purveyors of disinformation “tend to run a similar playbook when called out for engaging in co-ordinated inauthentic activity,” Golberg told National Observer.

The #TrudeauMustGo hashtag initially took off on July 16, in response to Prime Minister Justin Trudeau’s criticism of U.S. President Donald Trump’s ongoing racist attacks against four sitting congresswomen. By 8 p.m. on that date, the hashtag was being pushed out at a rate of approximately 2,100 tweets per hour.

While this may not have been the intended outcome, the #NotABot narrative ended up providing cover for inauthentic activity — and laying out a welcome mat for future disinformation campaigns.

Some of those aforementioned new wave of tweets called for a follow-up analysis, so we obliged.

More than 10,000 tweets tagged with #TrudeauMustGo were posted on July 16 and twice that many were posted the following day, according to data streamed via the Twitter API, indexed using Elasticsearch, and analyzed using the open source data visualization and analytics platform Kibana.

Activity on the hashtag levelled off at about 1,000 tweets per hour on July 17, before a second surge of activity started on July 18, climbing to about 2,300 tweets per hour at 10 p.m.

The activity level reached its highest point on July 19 around 11 a.m., when tweets using the hashtag were being pushed out at a rate of nearly 3,500 per hour.

More than 52,300 tweets using the hashtag were posted on that date. The numbers per day then declined, with more than 38,600 on July 20; more than 26,300 on July 21; and more than 17,000 on July 22.

Golberg explained just how purveyors of disinformation tend to operate when faced with being called out.

“One of the more common tactics involves misrepresenting what has been stated,” he explained. Then, when that misrepresentation is “debunked,” it is presented as proof that the initial reporting was wrong.

That's what happened with the #NotABot tweets, Golberg said.

“The #NotABot narrative seeks to discredit the data-driven reporting that identified #TrudeauMustGo as being weaponized. It attempts to create an alternate reality by twisting what was reported — setting up a scenario for the disinfo peddlers to ‘debunk.’”

Looking at the #NotABot tweets, most were based on the inaccurate premise that National Observer (and other media outlets that picked up on our analysis) cited "bots" as the reason #TrudeauMustGo became a trending hashtag. Many #NotABot tweets also implied that National Observer classified all accounts tweeting #TrudeauMustGo as "bots."

In fact, the word "bot" appeared only once in National Observer's report on the hashtag — and that one time was in the title. The term was not used once in the body of the article.

The hashtag cloud for #NotABot clearly shows that the hashtag is being used almost exclusively in reference to #TrudeauMustGo and associated hashtags, including #ButtsMustGo and #LiberalsMustGo.

A hashtag cloud for #NotABot shows that the hashtag is highly correlated with #TrudeauMustGo, indicating that the first hashtag is being used almost exclusively in reference to the second. (Photo: Screenshot).

Golberg also noted that he often observes inauthentic activity and other forms of social media manipulation surrounding hashtags like #NotABot.

Indeed, although the hashtag was meant to debunk reports of inauthentic activity, more evidence of inauthentic activity and digital manipulation emerged in our analysis of the hashtag.

Several of the most prolific tweeters using the hashtag #NotABot engaged in suspicious activity patterns and have since been suspended. One of the top contributors (by frequency of tweets) was @Jd0g77, an account created on July 20 and suspended by July 23. During its short life, the account tweeted 237 times, and a large proportion of those tweets were tagged with #NotABot. On July 23, a new account — @Jd0g17 — was created to replace the suspended account. It promptly started tweeting the hashtags #NotABot and #TrudeauMustGo.

Another top contributor to the hashtag #NotABot was @TCanuckchik, an account created on July 15. We identified this account in our initial analysis of #TrudeauMustGo, noting that it had tweeted 230 times within 24 hours of being created. According to the user’s profile, the account was created to replace an account (@ThatCanuckchik) that was restricted by Twitter.

Twitter also suspended two of the accounts identified by National Observer in our initial analysis of #TrudeauMustGo, including the top contributor to the hashtag, @CanadaProud10, which was tweeting an average of more than 140 times per day, and a newly created account, @aviator300E, which had tweeted more than 200 times a day since its creation on July 14.

Are these accounts “bots”? Probably not. However, they were engaged in a pattern of inauthentic activity that violated Twitter’s terms of service, which is likely why they were suspended or restricted.

The term "inauthentic activity" encompasses a range of tactics and behaviours, of which bots are only one. Twitter’s rules and policies describe some of these tactics, stipulating that it is prohibited “to use Twitter’s services in a manner intended to artificially amplify or suppress information or engage in behavior that manipulates or disrupts people’s experience on Twitter.”

Twitter also prohibits platform manipulation, which includes behaviours such as inauthentic engagement — defined as attempts to “make accounts or content appear more popular or active than they are” — and co-ordinated activity designed to “artificially influence conversations through the use of multiple accounts, fake accounts, automation and/or scripting.”

Strictly speaking, “bots” are fully automated accounts, meaning that they have been programed to autonomously carry out certain functions (replying, liking, retweeting, etc).

In contrast, “cyborgs” are accounts that are controlled by humans but also use automation or third-party apps to tweet or retweet more frequently.

Other examples of inauthentic activity include “followback parties” and “followback trains,” both of which are used to artificially inflate follower counts. Retweet rooms — groups created through Twitter’s direct message feature for the purpose of mass retweeting — may also be considered a form of co-ordinated inauthentic activity.

As National Observer reported recently, an estimated 25 to 30 per cent of accounts tweeting popular political hashtags like #abpoli, #ableg, #bcpoli and #cdnpoli display signs of inauthentic activity. Some of this activity includes bots, but most of it does not. However, all of it has the potential to disrupt the flow of information, distort online discourse and manipulate perceptions.

Bots and other forms of inauthentic activity are most effective when they blend in with the crowd by amplifying certain users, tweets and hashtags, thereby making them more visible to more people and increasing the likelihood that (authentic) users will engage with the tweets and boost them further.

While this may not have been the intended outcome, the #NotABot narrative ended up providing cover for inauthentic activity — and laying out a welcome mat for future disinformation campaigns.

Comments