Between February 9-10 this year, Indian Prime Minister Narendra Modi laid the foundation stone of a health facility in Tirupur (Tamil Nadu) and then unveiled another 10 km section of the Chennai metro, and laid another foundation stone via video conference. Between February 9-10, Twitter was profusely abused by bot accounts which posted an inhuman volume of tweets to support both the ruling party and the opposition narrative, according to a research conducted by DFRLab. The Digital Forensics Lab (DFRLab) is the Atlantic Council‘s initiative to track disinformation. The pro-Modi traffic, the DFRLab’s report found, was far more heavily manipulated than any large-scale traffic flow that it has analyzed till now.

#TNwelcomesModi

1. High post volume, 50 most active accounts drove 60% of hashtag traffic: #TNwelcomesModi trended in India on February 9-10, and was posted over 7.7 lakh times in two days. DFRLab analyzed the first 49,000 tweets of this hashtag, which were tweeted over a period of almost 8 hours. The top three accounts alone posted #TNwelcomesModi close to 5,000 times, or roughly 10% of all traffic in the scan. The 50 most active accounts generated 30,446 tweets, or over 60% of all traffic. “This was an attempt at manipulation on an industrial scale, using a small number of hyper-tweeting bots to give the hashtag a massive boost,” concluded DFRLab.

2. Manipulation score double than average: The first 49,000 tweets for #TNwelcomesModi scored a CTM for 123.98, the highest score the lab has ever recorded, showing that it was very heavily manipulated by a very small group. “An “eyeball test” of the accounts that posted the hashtag most often confirmed that these accounts were bots,” says the lab. Previous scores for heavily manipulated traffic have short up to an average of 60.

3. Inhuman volume of posts every few seconds, and per day

  • The lab found that one account @SasiMaha6 posted one #TNwelcomesModi every 15 seconds. This account was created barely a week before Modi’s visit and posted 380 times a day and over 6,800 times by February 22.
  • @priyamanaval6, although created a week before the visit, posted roughly one tweet every 17 seconds continously for over seven hours. By February 22, the account was posting 381 posts a day, with over 6,800 posts in total. @ANGEL0310970276 posted once every 22 seconds.

The conclusion? – “These sustained rates are far too high for human posting.”

#GoBackModi

1. Posts volume lower than pro-BJP hashtag: On February 10, the hashtag #GoBackModi also trended raking up over 49,000 tweets in just three hours, almost the total number of posts peaked at just 4.47 lakh on February 9 & 10. #GoBackModi was heavily pushed by a small number of high-volume accounts that posted hundreds of times an hour. Unlike #TNwelcomesModi, these accounts were still not suspended.

2. More bots with fewer followers, nearly 900 accounts posted 50K tweets: The individual opposition bots were even more prolific than their pro-BJP rivals, but they had a larger number of lower-activity accounts. 50 most active accounts tweeting #GoBackModi generated 16,325 tweets during the three-hour scan , fewer than first hashtag but still far more any human user could have done. Nearly 50,000 tweets in the #TNwelcomesModi scan were posted by just 891 accounts, while the nearly 50,000 tweets in the #GoBackModi scan were posted by 7,394 accounts.

3. Individual accounts traffic

  • @PhillyTdp posted one tweet every 5.3 seconds for over three hours and post a total of 2,179 times as the hashtag took off . This account posted over 2400 times on February 10; 98 percent of all its tweets since February 9 were retweets.
  • Another hyperactive account @nritdpusa posted over 1800 times in three hours, or roughly one tweet every 6 seconds.
  • The third most active account @ap_cbn posted over 1100 tweets in just over three hours on the hashtag #GoBackModi on February 10, for an average rate of one post every 10 seconds. This account posted 1,807 times on February 10 and 96 percent of its posts since December 27 were retweets.