Over the last two months, a number of political hashtags have entered Twitter’s top trending list in Egypt which have reflected the conflict between both fractions supporting and opposing the Egyptian President Abdel Fattah El-Sisi.
Both fractions launched a number of hashtags with the same meaning but are different in writing which lasted for days as each party tried to grab the attention of more Twitter user and gather as many followers as possible. During this period, we have observed 17 hashtags, the majority of which are in favor of the pro-Sisi fraction with 12 hashtags, with a percentage of 70% of total hashtags. On the other hand, total hashtags of the anti-Sisi fraction are 5.
Most of these hashtags are slightly different from each other as for how they are written, for example: on the supporting side “El-Sisi_Zaemy_Waftakher” (El-Sisi is my leader and I’m proud) and “El-Sisi_Raesy_Waftakher” (El-Sisi is my president and I’m proud), while the hashtags “Erhal_Ya_Sisi” and “Erhal_Yasisi” (Sisi, Leave) are launched by the opposing side.
This is not the last time for that to happen on Twitter, this conflict between El-Sisi supporters and opponents happened before when the hashtag “Erhal_Ya_Sisi” (Sisi, Leave) was launched on social media in 2015 after his statement that he was ready to leave his office only if this was the desire of “all Egyptian people”. So, his supporters launched the hashtag “Elshab_Behebak_Ya_Sisi” (People love you Sisi). There was also a hashtag of “Erhal” (Leave) which was launched in April 2016 after the president’s speech in which he affirmed that Tiran and Sanafer islands had always belonged to Saudi Arabia.
The last wave of opposing hashtags was launched as a result of many economic issues raised recently, such as the price hike of fuel and Metro tickets. This time, the polarization is clear between both fractions as each one tries to show its power and huge followers until El-Sisi expressed his “consent” of the opposing hashtag during his speech on the Sixth National Youth Forum at the end of last July.
InfoTimes team has investigated the last conflicting wave between the supporters and opponents on Twitter by analyzing the two conflicting hashtags “Erhal_Ya_Sisi” (Sisi, Leave) and “El-Sisi_Mesh_Hayerhal” (El-Sisi will not leave). Info Times has gathered a sample of 11,268 opposing tweets (re-tweet, reply to tweet and Mention) and 2,570 of supporting tweets during June. Then, this sample of tweets was analyzed as for quantity and quality in order to find out the nature of the accounts which launched these hashtags and led them to be in the top trending list.
#Sisi_Leave .. Starting Point
According to “Who Tweeted It First” tool available on internet, which displays the first tweet of any certain word, the first tweet mentioning the hashtag “Sisi_Leave” in the last wave of hashtags was launched on 11 June at 6:20 pm by an account related to a person who is active on Twitter. The first person to launch this tweet has confirmed, in a call with the investigation team, that he did not think to spread this hashtag, as he stated: “I used the hashtag just like many other people as I wanted to express my opinion, and I am not the first person to use this hashtag on Twitter”. He pointed out that this hashtag was there before in 2015 and 2016, and he refused to mention his name in investigation.When a sufficient number of automated accounts are available, they have the ability to spread large amounts of tweets, so that they are placed on the top trending topics. – Professor Mark Jones
Two days after the first tweet, an account called Po3leesh1, which has an unusual activity, tweeted the hashtag and added the sentence “Make it a trend”. By searching the list of tweets of this hashtag during the two-month investigation, we noticed few tweets by accounts of similar names; Po3leesh and Po3leesh11. Similarity of accounts’ names, even if it is strange, does not mean that it is an automated account. However, this leads us to search again for the source of this account. By analyzing Po3leesh1 account, we find that it has many features of automated accounts which are managed by “codes” to perform pre-set roles aiming to serve the purpose of its owner.
The account named Po3leesh1 has no profile photo nor a comprehensible name. It was made in the same month when the opposing hashtag was launched and was suspended on June 27 after the spread of the hashtag declined. Moreover, the large percentage of content published by that account during such period intensively used the hashtag opposing the president “Sisi_Leave”, in addition that most of its publications were re-tweets of the hashtag from other accounts.
According to Botometer tool, it is also likely to be an automated account, and such tool is developed by Indiana University Network Science Institute (IUNI). It helps to determine to what extent the account is likely to be automated by assessing it on the basis of a set of criteria. Then, the account gets a total average value from zero to five. The closer the score is to zero, the more it means that the account belongs to a human; and the closer it gets to the five, the more likely it is to be automated. Account Po3leesh1 gets 4 out of 5.
By analyzing the interactions on this hashtag in June, we found that about 23% of these interactions were done by only 20 out of the 489 accounts that participated in the interactions observed during last June. Among these 20 accounts, there are 9 accounts with a daily activity rate exceeding 100 tweets per day, and according to Oxford Internet Institute for Computational Propaganda, an account that publishes more than 100 tweets a day is suspicious enough to be classified as an automated account. Furthermore, we have identified 7 accounts out of such 20 to be missing, either deleted by their owners, or deleted by Twitter management for suspicious activity.
Along with the intensive tweeting and re-tweeting of hashtag “Sisi_Leave”, the majority of these accounts launch tweets about the Palestinian Issue and other tweets supporting the Muslim Brotherhood.
However, we found other accounts more influential in the discussion, which were not classified by Botometer as automated accounts, but its daily activity calls into question that it exceeds 200 tweets in each account and is not less than 500 tweets per day on average, the activity which can not belong to humans, such as th4ra, Mrabdo84033329, samah10203 and Pokarest1 which re-tweeted the hashtag 189 times and its tweets were retweeted by other accounts 92 times within the observed sample.
Professor Mark Jones, Assistant Professor at Institute of Arabic & Islamic Studies, University of Exeter, who is specialized in the study of discussions on the means of communication, says: “When a sufficient number of automated accounts are available, they have the ability to spread large amounts of tweets, so that they are placed on the top trending topics. Then, real people are attracted to discussing these topics, and so real heated discussions may begin from subjects originally launched by a set of programmed accounts”.
Accounts of media platforms have also reacted to the hashtag, such as Aljazeera Mubasher which retweeted 40 times, BBCArabic 16 times and Youm7 14 times, and its followers retweeted its publication 20 times in the observed sample.
The above visual design displays the network of participants in “Sisi_Leave” hashtag, each white point represents one account, and the red threads between them represent the interaction between each account and another. The more there is interaction between two accounts, the closer the points are to each other; and the more the point is in the middle of the network, the more it indicates that the point has a greater role in spreading the hashtag. The yellow circles represent accounts with suspicious activity that are more likely to be automated, while green circles are for media organizations.
What do you know about automated accounts?
An automated account, known as “bot”, is an account that has been programmed to perform a typical activity repeatedly on social platforms. People do not manage these accounts directly, but rather through codes that give orders to the account, and other similar accounts, to behave in a certain way, such as re-twittering tweets that include the hashtag “Sisi_Leave”. The activity of these accounts can be changed or disabled by changing the codes in control. You can also create automated accounts for creating new tweets that are pre-prepared or copied from other users, respond to other accounts, send private messages to them, or even follow them, and many other activities on various platforms, including Twitter.Creating bots is no longer a difficult programming task. – Saif Ahmed
“Creating bots is no longer a difficult programming task,” says Saif Ahmed, Director of WideBot, an Egyptian automated account company. “There are also online platforms that enable those who cannot write code to create automated accounts”.
However, not all automated accounts are of bad use. They may be used in researches such as Microsoft’s automated account TayandYou, in marketing such as Twitter’s own marketing account TwitterMktg, in serving people by alerting them to earthquakes such as earthquakesSF, or even to twitter every hour like the unofficial account of Big Ben clock big_ben_clock.
The automated accounts we have identified in our sample have been targeted to twitter and re-twitter the hashtag to amplify its popularity, which Jones believes are dangerous as they give no room for true discussions on these social platforms. He also added that malicious automated accounts destroy people’s trust in communication platforms and make them fear of trusting others on internet, as they also confuse content based on correct information with other controversial content and a lot of fake news”
#Sisi_will_not_leave .. Reaction
On June 24 at 1:51 am, four tweets with the hashtag “Sisi_Mash_Herhal” (Sisi will not leave) were launched at the same minute by four different accounts, separated by a few seconds.
The account which launched the first tweet maksully is highly likely to be an automated account by Botometer (score 4.8 out of 5) because it twittered only 17 tweets since it was created in January 2011, most of such tweets are in Russian and unrelated to each other.
This hashtag then was followed by other accounts with suspicious activity, including two accounts “If80fnrCM5dlFlM” and CQkXyT1n6JlmDmN, which significantly contributed to the spread of the hashtag by twittering it more than 100 times. At the same time, their average daily activity, whether in this hashtag or otherwise, was 100 tweets per day. So, Twitter management suspended the first account, while the second is still active with score 2.7 out of 5 on Botometer.
Like the opposing hashtag, 27% of interactions on this supporting hashtag were focused in only 20 out of the 554 accounts during the period we observed. At the top of those 20 accounts, there are 5 accounts which are related to each another under the name of “Gabhet Shaab Misr” (Egyptian People Bloc). Behind this name is only a blog defines itself as a “national bloc working for Egypt, not related to anyone and belongs to the homeland only”, and its page on Facebook provide no more information. The owner of bentel_nile account, who is the founder of the bloc as mentioned on her account, was at the top of twittering the hashtag by 34 tweets and 58 re-tweets; as well as other accounts under the name of the bloc such as faten_ahmedm, anaelmasri2121, Egypt_Ppl and ahmedelzyat123. These accounts could not be classified by Botometer whether they are automated or not as they got an average score, but they were among the top 20 interactive accounts on the hashtag.
Like the anti-Sisi hashtag, the above visualization represents the network of users reacting around the pro-Sisi hashtag. Nodes in yellow are suspicious ones, while nodes in green are media organizations.
The pro-president hashtag is distinguished from the anti-president one by the fact that more media outlets had a significant contribution to its spread. At the top is Youm7 newspaper by 74 re-tweets on the hashtag, as well as the account of On Live Channel OnLiveEgypt by 26 times of interaction, in addition to the account of Extra News Channel Extranewstv which interacted 21 times, and the account of TEN Channel TENTVNetwork which interacted 11 times.
According to a study prepared by the U.S. Pew Research Center on measuring public opinion and community issues published last April, two-thirds of tweets on Twitter are published by automated accounts. Moreover, a study made by Indiana and Northern California Universities estimated that from 9% to 15% of the accounts active on Twitter are automated, which means that the number of bots on the platform may reach 48 million accounts.
Using automated accounts on Twitter for the purposes of political propaganda has been used in Egypt only in the past few years, but it is widely used in other countries such as Venezuela, which is said by Freedom House Organization that “Its government has used social media platforms to spread its points of view and resist the political opposition in the country”. Justice and Development Party (AKP) which rules Turkey has been also using automated accounts since 2013 to increase its support on internet.
Twitter does not deny the use of automated accounts for the purposes of political propaganda, as 50 thousand accounts linked to Russian sources were identified which had published content influenced the general orientation of voters’ decision during the last US elections.
Twitter Changes Policy of Publishing
Twitter’s management has changed its policies last July in order to restrict the use of automated accounts for political propaganda. Twitter now blocks bots that make like on random tweets, replicate other real accounts, use mentions to address users who are not interested in this account, or follow other accounts in huge quantities at one time. These policies have led to a dramatic drop in the number of followers of all politicians and celebrities, indicating the extent to which automated accounts have penetrated into social platforms.
Such changes in policies make Saif, the Director of WideBot, optimistic about the future of automated accounts on various platforms, stressing that malicious accounts will be fought until they disappear, while the number of automated accounts with useful purposes will increase and demand for creating them will also increase.
If you’re interested in having a look at our dataset, click here.
Kanishk is an Indian data journalist who focuses on technology, privacy, internet culture, extremism and immigration. He studied at Columbia Journalism School and works for the American Atlantic Council research center.