Findings from our analysis of the ongoing tweet and retweet activity using the hashtag #saveGhouta help us understand programmatic behavior on Twitter in connection with the current war in Syria and the complexities of the information landscape.
This hashtag has been used in activist campaigns to bring the world's attention to Eastern Ghouta, an area on the outskirt of Syria's capital Damascus where armed rebel groups have been locked in a fight with the Syrian government forces besieging them. Both civilians and opposition fighters are entangled in this dense urban tissue, and civilian casualties from government airstrikes have been rapidly mounting. The tragedy of Ghouta is being promoted for fundraising purposes by groups based in the United States, such as the Syrian American Medical Society. The Society's Facebook verified page prominently features the hashtag on its cover photo and is currently holding fundraisers.
Upon returning to Washington from Syria in 2015, I closely followed the events going on back in Syria via the internet and social media. Connections I had made there regularly posted updates on what was happening in Syria on Facebook and Twitter.
At times, discerning facts from the information war could be a challenge. "This is a war," I thought, "and of course, warring parties will do their utmost to make sure that any piece of information they release will achieve the maximum impact for their cause and struggle." Syrian government channels labeled almost all opposition as "terrorists," while an Al Jazeera America executive told me in 2016 that he could not broadcast videos of opposition shelling of government-controlled part of Damascus.
It was in this context that last week Yasmina, who is originally from Damascus, now lives in Washington and working as an analyst of the Syrian conflict for leading organizations, John Gray, a social media researcher, and I looked into who was tweeting with the hashtag #saveGhouta, especially keeping an eye out for accounts that tweeted or retweeted using that particular hashtag a lot.
These events are taking place at a time when Americans have been reeling from revelations that Russia-aligned hackers had deployed networks of Twitter bots, i.e., automated fake accounts, to spread divisive content and junk news during and after the 2016 presidential election. It planted the seeds of doubts on the authenticity of any information published on social media.
In the wake of such phenomena against the American public following the 2016 presidential election, John and I were interested in the possibility that some of the accounts using this hashtag were bots as well. We were not the first to wonder this: the question of Twitter bots being used in the Syrian conflict had been raised by journalists as early as 2011. At the time, those interested in this possibility were looking at various government-aligned accounts that seemed aimed at discrediting anti-government activists. For example, The Guardian found that "pro-regime accounts have been set up to flood the pro-revolution narrative."
Accounts from which tweets using the #saveGhouta hashtag originated were mapped on select moments on February 23 and 24 of this year with the help of Mentionmapp, an analytical tool that John and his co-founder built.
John's research revealed that many accounts participating in the #saveGhouta campaign displayed a "bot-like" behavior.
He explained that a large number of the accounts tweeting or retweeting the hashtag #saveGhouta tweeted over fifty times per day, the definition of a programmatic activity, according to the Oxford Internet Institute. Some accounts even tweeted over one hundred times per day.
The finding that not all of the "propaganda" was being aimed at the people and coming from the government provided more insights into Twitter and the complexities of the information landscape. Clearly, if the Assad government could do it, why couldn't the opposition?
"If a public awareness campaign is getting fake accounts to amplify it via automation, should it bring into question the authenticity of that campaign?" I wondered.
In response to my musings, Yasmina did not deny the existence of fake accounts nor did she find them problematic. "The point is what," she challenged me, "[to show that] this is `all fake, a fake revolution? They are pro-revolution, especially this 'turtle lady' [TurtleWoman777, an account that tweeted 438 times a day on average]. Real people are being killed and real bombs have been targeting civilians and tweeting 50 times a day is not unusual."
She said there was "an agreed storm of hashtag campaign among the activists" around the current events in Eastern Ghouta.
All the same, in the light of events unfolding in the United States, I had to wonder whether the public ought to know when activists are deploying amplification bots in order to generate awareness for their campaign.
Should the Twitter's security algorithms somehow flag these bot accounts and inform the user?
If we expand our sample set, will we see the same kind of ratio of bots/non-bots?
A separate analysis could attempt to measure these accounts in terms of impact, engagement, or audience actions taken.
We will continue researching the Twitter hashtags connected with the Syrian conflict and report here.
Methodology and findings
Do tweets and retweets using the Twitter hashtag #saveGhouta display a 'bot-like' behavior, i.e. programmatic activity?
We looked at 2000 tweets that originated from 765 unique accounts collected at 10 specific moments in time Feb 23 and 24 and narrowed this data set to 88 accounts that had tweeted at least twice from this data set. Of these 88 accounts 42 exhibited bot-like behavior.
We found on a 7 day average:
42 accounts tweeted 50 + times per day = definition of programmatic activity
23 of these accounts tweet 100 + times per day
We subscribe to the Oxford Internet Institute baseline of 50 tweets per day to classify an account as a Cyborg, an account that exhibits bot-like behavior.
36 accounts joined Twitter since January 2017.
The map is a visual representation of how we see the conversation at that moment (mentionmapp.com)
The red icon (over the account's avatar thumbnail) indicates bot-like behavior.
Flavius Mihaies is an independent journalist and consultant at the World Bank. In May and August 2015 he traveled to Syria, where he visited Damascus, Homs and the Kurdish-controlled region in northwest Syria. Follow him on Twitter.
John Gray is CEO and co-founder of Mentionmapp Analytics Inc. and a freelance writer. He is social media researcher and reporter focused on the interactions between real people, bots, sockpuppets, and trolls. John's main consideration is how misinformation is impacting our socio-political discourse, and that cyberspace is becoming a less human place. He's co-authored "The Ecosystem of Fake: Bots, Information," and "Distorted Realities." John earned a Bachelor of Applied Science (Communications) and a B.A. (English) both from Simon Fraser University.