Marc Owen Jones tracking hate speech in the Gulf...
Some people have reported that Twitter hashtags relating to a siege on a Saudi town have been heavily contaminated by bots or PR accounts. The siege has been heavily politicised in the news, with different news sources reporting different things depending on their religious bias or sympathy towards Iran or Saudi. The Saudi Gazette, for example, has taken a pro government, tabloid-esque tone:
Trigger-happy terrorists ended the budding life of Saudi infant Jawad Al-Dagher during an indiscriminate firing at a development project site in Al-Masourah district of Al-Awamiyah in Qatif governorate in the Eastern Province. Jawad, who was 2 years old, was killed in front of his mother with a bullet to his head.
Meanwhile, the Al Bayt News Agency reported:
Al-Ahd news website quoted the local sources as saying that the Saudi army fired indiscriminately at the residents of al-Awamiyah and Qatif region and killed at least two civilians.
The official Saudi government line is that the two were killed by ‘terrorists’. Middle East Eye reported:
According to a statement from the Saudi interior ministry on Friday, workers on the project in Almosara “came under fire and their vehicles were targeted by explosives” from within the neighbourhood. A two-year-old Saudi boy – identified by the pro-government Al-Arabiya news site as Jawad al-Dagher – and a Pakistani resident were killed in the shooting.
The ministry did not specify when the attack took place but said criminals engaged in the drugs and arms trade had tried “to jeopardise the project and protect their terrorist activities that they launch from the abandoned houses in the neighbourhood”.
Activity on Twitters reflects a similar polarisation. An analsysis of the hashtag #حي_المسورة (Al Masoura neighbourood – in Al Awamiyya), reveals that there are a significant number of suspicious accounts, that may or may not be bots, but are certainly either 1) paid tweets 2) multiple accounts operated by a specific group of people, or 3) bots.
A sample of 12,000 Tweets was taken from between the 4th and 13th May using Twitter Archiver for Google Sheets. Duplicates were removed in Excel to reveal about 5832 individual accounts who had tweeted in the sample. A frequency analysis was done to quickly determine potentially suspicious accounts. The below graph shows spikes on certain dates where a lot of accounts were created (this usually suggests a single entity was creating a lot of accounts, ((or a major event prompted people to set up a Twitter account)) )
Closer inspection revealed it was unlikely/impossible that the spikes reflected actual individual humans setting up the accounts. Read on for more .
Below you can see a sample of Tweets from accounts using a Twitter App called ‘JAMDAM’. There were about 65 of these accounts in the sample. The fact it appears to be a bespoke app is suspicious, as is the fact that all these accounts were created on the same day. What’s interesting about this group is that whoever created them has gone to great lengths to make them appear unique, and they all have different locations, in both English and Arabic (See column on the far right). Each row represents a tweet from a unique account. The dates you see are the creation date of the account.
Whoever is behind the accounts clearly has an anti-Wahabi, anti-Saudi and anti-Israeli agenda. Most of the recent retweets by the accounts are criticising Saudi strikes on Yemen, or raising the awareness of the plight of Bahraini activists or cholera outbreaks in Yemen. The accounts appear to just retweets. For example, all JAMDAD accounts retweeted this tweet, which reads, ‘Wahahbis and Zionists are from the same womb’.
Moving on from JAMDAM. The next subset of suspicious accounts may be from a company that is paid to retweet. In addition to retweeting on the Al Masoura hashtag, these accounts tweet seemingly unrelated advertising content, including tweets about effective hair removal products. The first of the two tweets below says the ‘Security apparatus killed two wanted terrorists after the attackers opened fire on the demolition equipment in Al Masoura’. Below that is an advert to achieve smooth smooth legs.
Again, what makes these accounts similar, is not only that they retweet the same things, but they were all created on the same day. This particularl batch of pro Saudi hair removal bots was created on 12 May, 2017. In the sample there are about 52 unique accounts. Below you can see a snapshot. Again, each row represents a tweet from a unique account. The dates you see are the creation date of the account.
Another group of very similar advertising/political bots was identified. About 90 of these were created on 28th April 2017. The main tweet they contributed to the hahstag was supportive of the Saudi government’s official story, and reads: ‘A small group of terrorists who killed nobles of Al Awamiyya and the security forces will not find a safe haven after today’. See below.
(Their views on hair removal products are unclear.)
Below you can see another snippet of the data.
Another 100 or so of the accounts are perhaps a little creepier than the previous types. A bespoke app ‘Sdsdeb76’ was used. These accounts quote Suras from the Qu’ran, in addition to retweeting anti-Iranian tweets, mostly from this account >@Raml00g. Below for example, is a video of Hassan Nasrallah as a dog.
With regards to the Al Mansoura hashtag, a number of the retweets from the Sdsdeb76 accounts justified the action by the Saudi government. The below retweet says ‘For those who don’t know, Masoura is a very old neighbourhood where no one lives. It is not fit for habitation. Terrorists and wanted people hide there. Residents of houses were compensated with large sums’.
للي ما يعرف #حي_المسورة هو حي قديم جداً لا يسكنه احد ولا يصلح للسكن يختبى فيه الإرهابيين و المطلوبين تم تعويض اهالي المنازل بمبالغ عاليه— د رمــلــوج (@Raml00g) 12 May 2017
Some of the ‘Sdsdeb76’ accounts…
Given the agenda of these accounts, one of the tweets was particularly creepy. Again, it was a retweet of Ram100g, who cited 48:29 from the Quran, which translates as ‘ and those with him are forceful against the disbelievers, merciful among themselves’. It is not clear whether this is a dig against Shia, or just ‘terrorists’.
In summary, the hashtag on Al Masoura has been contaminated by a number of bots, with seemingly differing agendas. These tensions reflect the polarised news output, which reflect either an anti-Saudi/or anti-Iran bias (I will stick with political terms for now). The above blog just discusses unique accounts, and doesn’t necessarily discuss what percentage of the sample were from the suspicious accounts (I shall tabulate this and update soon).
The whole issue raises questions about the prevalence of certain people or institutions with ideological positions distorting the debate on Twitter by either creating or paying for the distribution of sectarian or partisan ideological content.