The Twitter Underground Economy: A Blooming Business

Aug 03, 12 The Twitter Underground Economy: A Blooming Business

A study on Dealers, Abusers and fake Twitter Accounts

by Jason Ding, Research Scientist

Many people dream of becoming popular or famous, and Twitter provides an outlet to make this possible. Most Twitter users try the standard way to get popular and gain followers: constantly tweet funny quotes or comments, discuss breaking events, or disclose information that many people want (like Guy Adams did). However, some Twitter users look for unusual ways to make themselves appear more desirable and become popular faster. One of these ways is buying Twitter followers, which right or wrong, is a significantly growing trend.

At Barracuda Labs, we consistently find and study fake profiles on social media platforms (reference our study on Facebook Fake Profiles at http://barracudalabs.com/fbinfographic/) in order to better protect our 150,000 customers from being phished or harmed. For the past 75 days, we have been investigating the business of trading Twitter followers on eBay and other websites searched from Google. As it turns out, this underground economy on Twitter is blooming! The results show that this Twitter business is growing very fast to form a series of underground markets.

For quick snapshot, please refer to our most recent infographic, The Underground Economy of Buying Twitter Followers at http://barracudalabs.com/underground/.

The Study

As part of this study, beginning in May 2012, our team set up three Twitter accounts and purchased between 20,000 and 70,000 Twitter followers for each of them from eBay and another website searched from Google. After collecting these followers’ profiles via Twitter API, as well as additional information from eBay sellers and Google search results, we found many interesting highlights of this business, summarized as follows  into 3 categories.

Dealers (those users who create fake accounts and sell followings):

  • There are 20 eBay sellers and 58 websites (within top 100 returns of searching “buy twitter followers” in Google) where people can buy (fake) followers
  • Twitter username is used to purchase, no authentication is required
  • The average price of buying 1000 followers is $18
  • A Dealer can control as many as ~150,000 or more Twitter followers (contact @kashifrox)
  • A Dealer can earn as much as $800/day for 7 weeks of selling followings if they can control  20,000 fake accounts (estimated on several random fake accounts reaching 2000 followings in 7 weeks and assume each following involved a minimum $20 purchase)
  • In addition to selling followings from these fake accounts, there are numerous opportunities for expansion into other services: selling tweets/re-tweets to earn additional profits

 

twitter-account-selling-followers-150K-1

Figure 1: A dealer trying to sell Twitter followers for up to 150,000

Abusers (those users who bought followers (most of which are fake) in order to look more popular or to use the accounts for selling ads):

  • There were 11,283 Abusers identified (each has at least 470 fake followers; magic number 470 is explained later, hint: it is related to @GregoryDEvans)
  • The average Abuser has 48,885 followers
  • 53% of Abusers have 4,000-26,000 followers
  • 75% of abusers have set a URL in their profiles (compared to 31% for random Twitter users)

 

twitter-dist-followers-for-abusers-4k-26k-1-1024x414

Figure 2: The distribution of number of followers for 11,283 Abusers

Fake Accounts (created by dealers for selling followings or tweets business):

  • There were 72,212 unique fake accounts identified
  • 61% of these fake accounts are less than 3 months old (since April 16th, 2012)
  • Average age of these fake accounts is 19 weeks or about 5 months
  • 55% of fake accounts have ~2000 followings
  • The average number of following for a fake account is 1,799
  • The oldest fake account @krails is created on Jan 15th, 2007

 twitter-dist-creation-dates-for-fakers-1-1024x381

Figure 3: The distribution of creation time for all fake Twitter accounts

Considering there were more than 11,000 Abusers identified from only 3 purchases we made, we are amazed by the size of market for selling and buying Twitter followers.

Several quick conclusions about these statistics:

  • Dealers are controlling the following speed and total following number of these fake accounts to avoid being suspended by Twitter. Dealers built various business services from the controlled accounts, as every cent is still money, e.g., selling 2000 re-tweets for $5 at here.
  • Half of the abusers have 4,000-26,000 followers, which makes them the most likely to be “cheating” group; and 3/4 of Abusers have set a URL in their profiles, meaning they might buy followers for website promotional purposes.
  • Fake accounts normally follow a lot of people, but normally no bigger than 2001 followings, indicating Twitter may internally uses this number as a cutoff for abused account detection.
  • Still, these statistics of fake followers can be easily used for detection purposes. However, Dealers can apply obscure techniques to make them hard to detect, e.g., randomly following some famous and some average people, or posting tweets grabbed from the Twitter stream, etc. This is one reason that the prices of followers vary dramatically on eBay and other online websites, ranging from $2 to $55 per 1,000 followers. The higher the price is, the more real these followers look.
  • On the other side, Abusers can try to avoid being caught as well, by buying followers multiple times from different services. For example, since March 2nd 2012, the world’s “Top One” security expert Gregory D. Evans (@GregoryDEvans) seemed to be purchasing 4 times to gain at least 50,000 new followers from several resources, which we do not know.  But, by running an overlapping check between his followers and our purchased followers, we found 470 fake accounts shown in both list. Bingo! That is also the **magic** number used in our study to detect all other Abusers.

 

Figure 4 & 5: left) One partial of 470  accounts followed both @GregoryDEvans and us; right) The following trending stats of GregoryDEvans from twittercounter.com

 

Most interestingly, during our investigation, Republican nominee for US President Mitt Romney has been scrutinized recently for his abnormal increase in new followers (@mittromney), indicating that these followers had been purchased in the same way as the Dealers/Abuser scenario from our study. We do note that these followers could have been purchased by either himself, his associates or by his opponents. Particularly, on July 21st, 2012, his follower number went from 673,002 to 789,924, representing a gain of 116,922 or 17%.  As this story picked up momentum, we quickly pulled his newest followers since the big breakout, (resulting in 152,966 new Twitter accounts), and can disclose several interesting statistics of these followers.

Statistics of Romney’s newest 152,966 Twitter followers(between Jul 21st and Jul 26th 2012):

  • The number of Romney’s followers increased 17% (or 116,922) on a single day Jul 21, 2012, going from 673,002 to 789,924
  • 25% of these followers are less than 3 weeks old (created after July 17th 2012), 80% of them are less than 3 months old
  • 23% or about 1/4 of these followers have no tweet
  • 10% of these account has already been suspended by Twitter

Figure 6 & 7: left) The distribution of creation time of Romney’s recently followers; right) The following trending stats of MittRomney from twittercounter.com

 

Based on the above distinguishable features, we believe most of these recent followers of Romney are not from a general Twitter population but most likely from a paid Twitter follower service. Also notice that Romney’s newest followers did have few different statistical features than our purchased followers (e.g. much newer than ours), indicating his followers were purchased from a different data resource than ours.

It is important to note that authentication is not required when buying Twitter followers from eBay or other websites. It is possible for anyone to buy followers for other Twitter users. That being said, Romney’s newest followers could have been paid for by himself, his associates or by his opponents. So far, there is not a feasible way to confirm who is responsible.

Finally, creating fake Twitter accounts and buying/selling followers is against Twitter’s ToS, and gradually erodes the overall value of the social network. Twitter keeps on detecting fake accounts and followings, and suspending them in last few years. However, if they do not move faster and smarter, these fake accounts will continue to be created, blended into the massive Twitter population, bringing bigger and bigger impact.

This underground Twitter business is just blooming.