Thread by @valent_projects

We examined 12,545 tweets using a pro #Sudan army hashtag. We found large-scale manipulation promoting escalation, whilst authentic Sudanese voices called for an end to the violence. The findings show how one of the most common deceptive practices on Twitter works #thread 🧵 1/13

Since fighting broke out between #Sudan army and the Rapid Support Force (RSF) on April 15, the hashtag جيش_واحد_شعب_واحد# (One Army, One Nation) has been used 12,545 times by pro-army accounts. All sides are running digital influence operations with much #disinformation 2/13

Firstly, we applied @benimmo’s Coefficient of Traffic Manipulation (CTM) to the dataset. CTM assigns a rating based on the number of retweets per user, percentage of retweets and the percentage of total traffic made up by the 50 most active users (see graph below). 3/13

@benimmo 2,267 accounts were responsible for the 12,545 tweets. Of those accounts, the top 50 were responsible for 49% of the total number of tweets. CTM for the dataset came in at a whopping 62.39. Organic hashtags rank at about 12.00 4/13

@benimmo The unusually high CTM figure warranted further examination of the dataset. We decided to identify and examine two control groups from amongst the users. We ranked the users by output and found they formed a distribution similar to a steep Lorenz curve 5/13

@benimmo We sampled the top 5% and bottom 50%. The top group was response for more than half the total number of tweets while the bottom group was responsible for only 10% 6/13

@benimmo Looking at the top 5% of accounts (113 accounts), we found classic indicators of bot/troll activity, including; generic account names, mass posting and account registration around the time of Sudan’s popular uprising against military rule in 2019 7/13 buff.ly/3CjM38V

@benimmo The bottom group who made up half the users, but only 10% of the total number of tweets, did not use the pro-army phrases at all. Examining the accounts in this group, we found most to likely be authentic users - see screen grab below 8/13

@benimmo Using an algorithm that examines key-word bias, we found the accounts in the top group commonly used specific pro-army phrases such as "Support the Sudanese army", "Neutrality is treason" and "We are all the army" 9/13

@benimmo The bottom group - making up the majority of posters - never used such jingoistic  phrases, tending instead towards more conciliatory and less extreme messaging focused on ending the violence and safeguarding civilians 10/13 buff.ly/426yaFm

@benimmo It is worth noting, the top group is made up of accounts designed to resemble ordinary citizens who happen to support the army. The overall intent is to gaslight Sudanese and observers into thinking there is a groundswell of support for the military to escalate 11/13

@benimmo Instead, in reality, those using the hashtag are supporting the military in the hopes of a quick end to the fighting and quick resolution to the fighting. 12/13

@benimmo What our analysis shows is that belligerent actors are astroturfing Sudanese, as well as foreign journalists, researchers and international actors into thinking there is an organic appetite for war when in fact the opposite is true. 13/13