Surprise! Twitter is still teeming with fake accounts

fake twitterWhen it comes to social media popularity, the more the merrier – usually. However, that’s not the case when the “more” part is attained through fake followers and ghost accounts. This problem is especially rampant within Twitter, and although there are a few applications available out there that help eradicate or expose spammy profiles, they aren’t enough to keep the underground economy at bay.

According to new research from Barracuda Labs, Twitter still has the highest amount of social media shenanigans, brought on by people selling fake accounts to the highest bidder. To gauge just how bad the situation is, a team of researchers scoured Google, eBay, and Fiverr for merchants pawning off Twitter followers for different price ranges. These followers are then injected into the team’s controlled Twitter accounts for a more comprehensive statistical examination using the site’s API.

Results are segregated into three different groups: Sellers of fake followers, buyers of fake followers, and fake accounts used by sellers to conduct business. Results show that fake follower sellers are becoming extra crafty with their techniques, providing specialized followers according to location as well as offering special services that are too good to be true (just check out the packages presented by FastFollowerz), including bypassing fake follower detection by StatusPeople.

When it comes to fake follower buyers, only 121 out of 1,147 identified Twitter accounts were found to be fake. 16 users had more than 1 million followers, while 88 users were in the 100,000 follower range. The average buyer has over 52,000 followers and has had a Twitter account for almost two years.

Finally, through the team’s analysis, more than 99,000 fake accounts were identified, with an average of 60 users followed, 77 tweets posted, 32 followers, and account age of 30 weeks (roughly seven months old). 63 percent of the fake accounts discovered were created by duplicating real Twitter accounts along with the user’s display name, bio, and location – the only difference is the screen name has an extra character appended to it. Comparison between real accounts and their duplicates have also shown discrepancies only a machine-manipulated Twitter account can produce (five separate tweets of more than 60 characters each all published within a minute on a single account, for example).

As long as Twitter users show an interest in beefing up their online presence the quick and easy (and deceitful) way, the technology behind these fake follower sellers can only become more sophisticated over time – their algorithms already have the ability to fly under the radar of various services used to detect account fakeness, ones that are usually fool-proof (well, maybe not completely fool-proof, since these services openly discuss what makes a Twitter account “fake” and observant vendors can easily use that data to their advantage). Once they reach a point where they can pattern their fake accounts’ behavior seamlessly to the real accounts they shadow, there’s really one thing left for the entire Twitterverse to do: Stop relying on numbers altogether.

Editors' Recommendations