How an online army of K-pop boy band fans exposes Twitter’s multiple account problem
The use of automated or multiple accounts promoting chosen messages are not exclusively limited to the domain of politics and international relations. The tactic is a useful marketing tool for everything from products to pop-culture to politics. The tactic on Twitter may even be more effective than paid advertising because automated content creates the facade of real users organically engaging on the message with the goal of trending, as opposed to the transactional label of “promoted content” stamped at the bottom of the tweet.
The case of the lead up to an awards show by fans of a Korean boy band provides a glimpse into the creation, tactics, and impact of manufactured messages on social media.
Bangtan Boys or BTS is a record-breaking K-pop (Korean pop) boy band, which formed in 2013. Their latest single Fake Love recently reached the top ten on the Billboard Hot 100 chart and became the 17th non-English song featured in chart in over 59 years.
BTS is a social media savvy band. Their YouTube videos are watched by millions around the world and average at around 300,000,000 views per video.
They are known for their highly engaged fan base, who call themselves A.R.M.Y, which is an acronym for “Adorable Representative MC for Youth”. The A.R.M.Y is very active on social media and take their role of promoting BTS very seriously. So much so, in fact, that they recently shut down a K-pop segment of a popular Toronto radio station for wanting to dedicate an hour of their broadcast to K-pop. The idea was attacked by BTS supporters on Twitter for promoting K-pop bands other than BTS. After the incident, the Canadian radio station decided to discontinue the K-pop hour.
Of all social networks, BTS’s fan base is concentrated on Twitter. Although there’s no reason to suspect that their followers are fake, the popularity of BTS hashtags ahead of the Billboard Music Awards (BBMAs), which took place on May 21, gives reason to doubt the authenticity of some of the BTS-supporting accounts.
Between April 24 and May 25, three BTS-supporting hashtags, #IVoteBTSBBMAS, #MTVLAkpopBTS and #MTVLAFanArmyBTS were used by 121,000,000 users who mentioned the three hashtags 272,000,000 times.
According to Twitter, the social network has 336 million monthly active users. It follows that one third (121,000,000) of those accounts were BTS fans for the one-month period between April 26 and May 25.
These numbers are suspicious because BTS hashtags were used by 121,000,000 accounts, but their two official English and Korean Twitter accounts were followed by respectively 15,000,000 and 11,000,000 users. This means that at least 80 percent of accounts who used BTS hashtags did not follow them on Twitter.
It appears that the use of the three hashtags peaked ahead of the BBMAs and plateaued the day after the awards ceremony.
A machine scan revealed that 75 percent of users retweeted tweets that mentioned the hashtags, as opposed to writing original tweets themselves.
Of all users, 45 percent had a low authority score (based on account activity and account’s influence) between one and four, indicating a high presence of automated or new accounts, which tend to have few followers and therefore a low influence score.
A more thorough machine scan revealed that 10.5 percent of all Twitter accounts using the hashtags, 12 million accounts, had an authority score of zero or one, which indicates a high presence of bots or new accounts.
Accounts with the low authority score generated 34,500,000 mentions, 90 percent of which were retweets.
Due to the extremely large number of tweets, @DFRLab was unable to analyze all of the accounts that used the hashtag, but a sample of 50,000 latest tweets revealed that the majority of accounts that used the hashtag were created in the past month, but did not appear to be automated.
The nature of BTS A.R.M.Y-ies makes it hard to distinguish between a genuine user and a bot. BTS fans remain anonymous, use BTS photos as their profile pictures and tweet exclusively about the BTS with most accounts solely retweeting tweets using BTS hashtags.
Despite that, most accounts analyzed did not appear to be automated, there were no observable patterns in their online activities — they tweeted at random times, but rarely during the night and the number of their tweets varied every day.
Despite that, they did not appear genuine and seemed to be dedicated to promoting BTS online. Further online research revealed that BTS fans often create multiple accounts on Twitter to support their favorite band.
@BTSVotingTeam was in particularly vocal encouraging BTS fans to create multiple Twitter accounts and their calls to action got more than 10,000 retweets.
@DFRLab also found some BTS fans bragging about their multiple accounts, with one user tweeting out the handles of their 17 accounts. Another replied, “I haven’t stopped making more accounts.”
Apart from Twitter, BTS fans also took it to BTS’s page on aminoapp.com to share tutorials on how to create multiple fake accounts using Fake Mail Generator.
Some of A.R.M.Y.’s most active members shared comprehensive tutorials on how to post multiple tweets supporting the BTS at the BBMAs without triggering Twitter’s spam detectors.
Raw data extracted from Sysomos appears to confirm that some BTS fans have multiple accounts as a lot of handles of accounts that used BTS hashtags appear to be semantically similar.
According to Twitter, having multiple accounts and posting “duplicative or substantially similar content, replies, or mentions” goes against the social network’s Rules and Policies, but it does not appear that this rule was enforced on the many BTS fans who promoted the band on their multiple accounts.
It is highly unlikely that BTS fans managed to collectively create more than 10,000,000 fake accounts, therefore, it is highly likely some of BTS supporters on Twitter are bots. Because BTS fans’ fake accounts have so many bot-like qualities, it is nearly impossible to distinguish between a fake account managed by a fan and a bot account run by an algorithm. This case study exposes the blurring distinction between a bot and a real person on Twitter.
Follow along for more in-depth analysis from our #DigitalSherlocks.