Recent research has suggested that there are hundreds of thousands of fake Twitter accounts lying dormant on the site.
Furthermore, these fake accounts are all tied together, with the largest network to be uncovered so far being one of over 350,000 accounts. Further research suggests that there could be networks even larger than this.
The networks were accidentally uncovered by UK researchers who were experimenting with the site, looking at the behaviour of the site’s users and how they interact with the site. Juan Echeverria of UCL was analysing a sample equating to 1% of all users when the data showed some unusual results. When further probed, it showed that there were many linked accounts run by a single person or group. Furthermore, these accounts did not act like regular bot accounts on the site, yet were evidentially not being run by humans.
When researchers from UCL looked deeper into the accounts, many of them had been created to send spam, boost interest in a number of trending topics and even increase the number of followers an account has by creating fake followers.
A large number of accounts on the site are controlled by bots. These are remotely run with automated responses and activity. People have been known to pay to have bot accounts follow their account to boost follower numbers and dilute discussions around controversial topics.
Echeverria and his colleagues are now calling on the public to report these bot accounts to help them get a better idea of their prevalence. He pointed out that these accounts can be easy to spot as they are fairly new to the site, have strange user names, little content and few followers.
The network of 350,000 bot accounts they had earlier discovered stood out in their analysis because each of the accounts shared a number of subtle characteristics proving they were linked. This included tweets being posted from places where no-one lives, all messages being posted from Windows phones and the inclusion of Star Wars quotes.
Twitter has a strict policy in place that regulates automated accounts and prohibits users from writing code that will automatically follow and unfollow accounts or like tweets en masse, as this can degrade the Twitter experience for other users.
With these regulations in place, the discovery of such networks comes as a big surprise to both Twitter and the UCL researchers.