Share:
Twitter’s vast amount of fake accounts is relatively well-known. But how these fakes function often gets lost in the large numbers and technical assessments. Yet, whether it is rote AI bot, cyborg, or a troll-bot, how they function matters. What happens when a troll-bot is born? How does it act? What does it tweet? What accounts does it follow? Are there patterns?
Identifying functionality does a couple of things. First, it exposes how easily identifiable these accounts should be for Twitter’s fancy algorithms. But it also gives real users more tools to assess the fakes as they become more sophisticated and developed.
Troll-Bot’s First Tweet
We found this troll-bot just three hours after its first tweet. “Valeria,” allegedly located in Virginia, Minnesota (yes, that’s a real city in MN), managed to copy and paste a tweet’s text. Exposing the political intentions of this fake account, the troll-bot retweeted Trump supporter Andrew Klavan at 12:37 PM on Sept. 23, 2019.
At the three-hour mark, the account had only one tweet, was following nobody, and had somehow managed to grab four followers.
Why Walk When the Troll-Bot Can Run?
Then the troll-bot got busy. Approximately 3.5 hours later, the troll-bot retweeted another Trump supporter, Michelle Malkin, in much the same way it retweeted Klavan.
Nine minutes later, it retweeted the Trump-thumping “news” network, OANN.
Just one minute after the OANN retweet, the troll-bot plagiarized a tweet that had originated 12 minutes prior. The tweet bashed the Emmy Awards, Hollywood (generally), and took a religious swipe (“pagan statuette”).
At this point, we decided to poke the troll-bot. And Twitter Support.
Unsurprisingly, there has been no reply.
Magical Following Catches Up with Itself
As the troll-bot tweeted the latest four tweets, the numbers on the account showed it had also followed two accounts and “liked” seven tweets. Twitter’s algorithms hadn’t quite caught up with the account, however. It was following significantly more than two accounts.
The follower count, however, appeared accurate when adding the spam account Twitter hid from view.
The “likes” were certainly on-topic for a Right troll-bot. For example, it liked it’s own tweets and Right-leaning political tweets.
By the end of the day, the account’s following number had caught up to the number of accounts the troll-bot is following (77). And it had two, additional “likes.”
Day 2: Purposeful Slow Growth
On the 2nd day, the troll bot slowly increased its account numbers.
Following – 129
Followers – 8
Likes – 12
Tweets – 11
An intriguing development here is that the troll-bot’s boost in numbers included some American Left activity as well as some UK activity, including Likes and Following. (In the “following” below, the circled account is the last account the troll-bot had followed on the previous day.)
Following
Followers
Likes
Tweets
This Troll-Bot’s Sudden, Short Life-Span
On the 3rd day of this troll-bot’s existence, it again increased its numbers. Just before the person behind the account shut it down.
Following
Note that the troll-bot started off by following Right accounts (day 1), then it added some Left and UK accounts (day 2), and, on the 3rd day, it started following a number of Canadian accounts.
Followers
Likes
Tweets
The Troll-Bot’s Demise
Maybe the person behind the troll-bot finally saw our reply. Or maybe it was for some other (unknowable to us) reason. But at the end of the 3rd day of slowly building the account, the person behind the troll-bot shut it down.
Troll-Bot Functionality Tells Us A Lot
What the birth (and death) of this troll-bot tells us is extensive. First, as a part of a large network of fake accounts, this troll-bot showed a deliberate, slow-growth pattern. It seems reasonable to say that this pattern would help the fake accounts slide under Twitter’s algorithmic radar.
Secondly, the troll-bots following pattern is relatively distinct. It followed similar patterns in groups. One after the other. First Right-political accounts. Then Left- and UK-political accounts. Then Canadian-political accounts.
Third, this troll-bot, which only survived for three days, started off with exclusively spam and other fake accounts in its followers.
Lastly, while the account varied the topic of its retweets (and plagiarized tweets) and likes to some extent, it continued to like and retweet Right-friendly accounts throughout.
One Last Note
It’s important to keep in mind that we found this account, visually identified it according to basic troll-bot patterns, and we tracked it for three days. At the end of the three days, the person behind the account shut it down. In other words, Twitter’s fancy algorithms didn’t catch the troll-bot at sign-up. They didn’t catch the account as it built its numbers. And they didn’t catch the account before it shut down.
Twitter is a multi-billion dollar company with technical experts galore. A company that boasts about advancing its algorithms to catch accounts such as this. And yet, they completely missed this account. And, rest assured, there are millions more just like it.
Written by Unhackthevote
Read More about Fake Twitter Accounts:
Twitter’s Model Bots–Know Their Faces
Twitter Bots’ Quick-Change Profiles
Follow us:
