26 June 2014

ANALYZING THE ISIS “TWITTER STORM”

For the last eighteen months we’ve been closely monitoring the Syrian conflict. One of the ways we do this is through social media, using a range of tools to aid our work. For Twitter, we use Palantir’s Torch platform—a data analysis and visualization program—and decided to use it to analyze Friday’s ISIS #AllEyesOnISIS ‘”twitter storm.” The hashtag was first announced last week by user @Ansaar999 as the English counterpart to a larger campaign including the Arabic hashtag:

#حملة_المليار_مسلم_لنصرة_الدولة_الإسلامية

This translates roughly to “A billion Muslims in support of Islamic State of Iraq and Shaam,” that was ostensibly launched to demonstrate worldwide grassroots support for the organization.

#AllEyesOnISIS makes for an interesting case study because, as J.M. Berger noted in a recent article, press coverage can often overstate ISIS’s grassroots support online, while underplaying their prowess as strategic communicators that punch above their weight by effectively rolling out tightly-controlled campaigns to amplify the appearance of online grassroots support. We felt that tracking the development of the hashtag over time and digging into the numbers around tweeting patterns might reveal something interesting about their grassroots support on Twitter.

We began collecting data on #AllEyesOnISIS at 16:00 BST (British Summer Time) on June 19th and stopped at 10:00 on June 21st. The graph below shows the growth and decline of the #AllEyesOnISIS hashtag over this period.

Activity was initially slow but then began to spike after 11:00 and peaked at midday. This makes sense given that organizers of the campaign had called for it to officially start at 10:30. The graph below therefore shows all uses of #AllEyesOnISIS during a 24-hour period from 10:00 on Friday.

During this period there were around 31,500 tweets in total, peaking between 12:00-13:00 when #AllEyesOnISIS was tweeted 4,300 times. Over the same period the ISIS hashtag (#ISIS) received a comparable number of tweets, being used 34,100 times. However, the frequency of its use was much more evenly distributed throughout the day.

All mentions of #ISIS over 24-hour period coinciding with #AllEyesOnISIS “twitter storm” 
Going back to the “twitter storm” itself, we can also see that approximately two-thirds (~20K) of all tweets were sent in the first seven hours of the campaign’s start, demonstrating that most activity was concentrated in this period.

The most interesting aspect of all this is to assess the distribution of work carried across all users in the #AllEyesOnISIS communication space. Last week, J.M. Berger noted a user-to-tweet ratio of about 28% for a similar campaign using an Arabic hashtag, suggesting that the same users were just tweeting a lot more.

Our data appears to confirm Berger’s findings. The top 50 tweeters of #AllEyesOnISIS accounted for just under 20% of yesterday’s total tweet volume, accounting for 6,300 tweets, and averaging 126 per person. Some of the most prolific tweeters were: @nun0z, @ISISrtonly, and @truthsMaster.

This shows that a small number of enthusiastic and deeply invested activists shouldered the burden for one-fifth of the campaign’s overall output. Findings like this are not completely unexpected. Similar asymmetries in participation have been noted in a number of empirical studies of online social movement activity.

No comments: